Codebase intelligence for AI and humans: code health scores, auto-generated docs, git analytics, dead code detection, and architectural decisions via MCP.
Python aicode-complexitycode-healthcode-intelligencecode-qualitydead-codedeveloper-toolsdocumentationgit-analyticsmcpopen-sourcepythonrefactoringstatic-analysistechnical-debt
docs: correct code-health marker taxonomy and stale benchmark claims (#1338)
Raghav Chamadiya8c17ea0e99ce
diff --git a/.github/assets/health-loop.svg b/.github/assets/health-loop.svgindex 85ee7b80e..6ec3f3beb 100644--- a/.github/assets/health-loop.svg+++ b/.github/assets/health-loop.svg@@ -11,7 +11,7 @@<!-- top: the engine --><rect x="330" y="28" width="540" height="48" rx="24" fill="#F8FAFC" stroke="#E2E8F0"/>- <text x="600" y="52" text-anchor="middle" fill="#0F172A" font-size="16" font-weight="700">25 deterministic biomarkers</text>+ <text x="600" y="52" text-anchor="middle" fill="#0F172A" font-size="16" font-weight="700">49 deterministic detectors</text><text x="600" y="69" text-anchor="middle" fill="#94A3B8" font-size="12">tree-sitter + git history · zero LLM · under 30s</text><!-- fan-out connectors -->diff --git a/.github/assets/one-index-dark.svg b/.github/assets/one-index-dark.svgindex 1d6069300..97bc93aeb 100644--- a/.github/assets/one-index-dark.svg+++ b/.github/assets/one-index-dark.svg@@ -15,4 +15,4 @@.dec { font-size: 14px; font-weight: 600; fill: #EEEAF4; }.pill { font-size: 10.5px; font-weight: 600; }.mono { font-size: 11.5px; fill: #A79DB3; font-family: ui-monospace, SFMono-Regular, "SF Mono", Menlo, Consolas, monospace; }-</style><rect width="1600" height="812" fill="#17131D"/><text x="36" y="50" class="h1">One index. Everything your agent and your team keep asking for.</text><text x="36" y="80" class="sub">Index once, in seconds, no LLM required. Every answer below is already computed.</text><rect x="36" y="104" width="748" height="352" rx="14" fill="#211B29" stroke="rgba(213,197,232,.10)" stroke-width="1"/><text x="58" y="135" class="lbl">CODE HEALTH</text><text x="762" y="135" class="kick" text-anchor="end">25 markers · zero LLM · <30s</text><rect x="58" y="150" width="146" height="58" rx="9" fill="#110D17" stroke="rgba(213,197,232,.10)"/><text x="70" y="170" class="chiplbl">Defect risk</text><text x="70" y="197" class="chipval" fill="#F2A03D">7.5<tspan class="chipunit">/10</tspan></text><rect x="214" y="150" width="146" height="58" rx="9" fill="#110D17" stroke="rgba(213,197,232,.10)"/><text x="226" y="170" class="chiplbl">Maintainability</text><text x="226" y="197" class="chipval" fill="#34D399">8.6<tspan class="chipunit">/10</tspan></text><rect x="370" y="150" width="146" height="58" rx="9" fill="#110D17" stroke="rgba(213,197,232,.10)"/><text x="382" y="170" class="chiplbl">Perf risks</text><text x="382" y="197" class="chipval" fill="#A98FC4">268<tspan class="chipunit"></tspan></text><rect x="58" y="220" width="704" height="34" rx="8" fill="rgba(52,211,153,.10)" stroke="rgba(52,211,153,.26)"/><circle cx="76" cy="237" r="4.5" fill="#34D399"/><text x="90" y="242" class="trust">17 of the 20 lowest-health files were bug-fixed in the last 6 months, <tspan font-weight="700">3.63× the 23% baseline</tspan></text><rect x="58" y="266" width="452" height="170" rx="9" fill="#110D17"/><clipPath id="galaxy"><rect x="58" y="266" width="452" height="170" rx="9"/></clipPath><g clip-path="url(#galaxy)"><circle cx="150" cy="346" r="62" fill="rgba(213,197,232,.05)"/><circle cx="140.1" cy="365.8" r="5.4" fill="#8FCB6B" opacity="0.84"/><circle cx="157.6" cy="330.2" r="4.4" fill="#F2A03D" opacity="0.87"/><circle cx="139.0" cy="345.5" r="3.6" fill="#EE8A50" opacity="0.92"/><circle cx="152.5" cy="388.3" r="2.2" fill="#F2A03D" opacity="0.69"/><circle cx="169.2" cy="364.9" r="5.4" fill="#F2A03D" opacity="0.80"/><circle cx="190.4" cy="362.5" r="2.2" fill="#F2A03D" opacity="0.68"/><circle cx="200.6" cy="361.2" r="3.1" fill="#34D399" opacity="0.62"/><circle cx="108.3" cy="335.1" r="4.4" fill="#8FCB6B" opacity="0.69"/><circle cx="168.3" cy="385.4" r="5.4" fill="#34D399" opacity="0.68"/><circle cx="189.4" cy="373.7" r="4.4" fill="#34D399" opacity="0.65"/><circle cx="156.6" cy="307.6" r="3.6" fill="#8FCB6B" opacity="0.72"/><circle cx="150.2" cy="370.2" r="2.6" fill="#8FCB6B" opacity="0.82"/><circle cx="137.5" cy="384.2" r="3.1" fill="#8FCB6B" opacity="0.71"/><circle cx="136.7" cy="400.9" r="2.2" fill="#34D399" opacity="0.59"/><circle cx="100.7" cy="372.4" r="2.2" fill="#34D399" opacity="0.59"/><circle cx="102.1" cy="322.4" r="3.1" fill="#34D399" opacity="0.61"/><circle cx="126.3" cy="378.5" r="3.6" fill="#34D399" opacity="0.71"/><circle cx="142.8" cy="333.9" r="5.4" fill="#F2A03D" opacity="0.90"/><circle cx="118.3" cy="391.4" r="3.1" fill="#34D399" opacity="0.60"/><circle cx="160.3" cy="355.3" r="3.1" fill="#F2A03D" opacity="0.90"/><circle cx="169.8" cy="323.3" r="3.6" fill="#EE8A50" opacity="0.78"/><circle cx="135.6" cy="313.9" r="2.6" fill="#8FCB6B" opacity="0.74"/><circle cx="201.9" cy="349.9" r="2.6" fill="#34D399" opacity="0.62"/><circle cx="186.9" cy="346.9" r="3.1" fill="#8FCB6B" opacity="0.73"/><circle cx="193.9" cy="331.8" r="5.4" fill="#34D399" opacity="0.66"/><circle cx="190.7" cy="316.3" r="5.4" fill="#8FCB6B" opacity="0.63"/><circle cx="121.9" cy="368.5" r="2.2" fill="#8FCB6B" opacity="0.74"/><circle cx="129.8" cy="360.6" r="2.6" fill="#F2A03D" opacity="0.82"/><circle cx="172.2" cy="346.0" r="2.2" fill="#EE8A50" opacity="0.84"/><circle cx="118.4" cy="349.9" r="2.6" fill="#8FCB6B" opacity="0.77"/><circle cx="119.4" cy="382.2" r="2.2" fill="#8FCB6B" opacity="0.66"/><circle cx="178.5" cy="327.5" r="2.6" fill="#8FCB6B" opacity="0.75"/><circle cx="111.6" cy="312.9" r="2.6" fill="#8FCB6B" opacity="0.63"/><circle cx="170.4" cy="301.4" r="2.6" fill="#34D399" opacity="0.64"/><circle cx="101.3" cy="348.2" r="2.2" fill="#8FCB6B" opacity="0.65"/><circle cx="205.6" cy="341.8" r="3.1" fill="#34D399" opacity="0.60"/><circle cx="181.3" cy="369.4" r="3.1" fill="#34D399" opacity="0.72"/><circle cx="160.0" cy="297.6" r="3.6" fill="#8FCB6B" opacity="0.64"/><circle cx="108.4" cy="364.3" r="2.2" fill="#8FCB6B" opacity="0.67"/><circle cx="159.8" cy="363.8" r="2.6" fill="#EE8A50" opacity="0.85"/><circle cx="104.9" cy="356.0" r="4.4" fill="#8FCB6B" opacity="0.66"/><circle cx="99.9" cy="331.7" r="2.2" fill="#8FCB6B" opacity="0.62"/><circle cx="130.0" cy="330.2" r="3.1" fill="#F2A03D" opacity="0.82"/><circle cx="133.2" cy="295.3" r="3.1" fill="#8FCB6B" opacity="0.61"/><circle cx="128.6" cy="318.9" r="3.1" fill="#F2A03D" opacity="0.75"/><circle cx="176.2" cy="340.6" r="2.2" fill="#F2A03D" opacity="0.81"/><circle cx="93.4" cy="350.5" r="2.6" fill="#34D399" opacity="0.59"/><circle cx="120.0" cy="362.2" r="2.2" fill="#F2A03D" opacity="0.75"/><circle cx="120.9" cy="317.3" r="2.2" fill="#34D399" opacity="0.70"/><circle cx="173.8" cy="332.6" r="2.2" fill="#8FCB6B" opacity="0.80"/><circle cx="177.5" cy="308.0" r="3.1" fill="#34D399" opacity="0.66"/><circle cx="149.1" cy="353.3" r="2.2" fill="#E06A5A" opacity="0.95"/><circle cx="143.4" cy="396.7" r="3.1" fill="#34D399" opacity="0.63"/><circle cx="126.6" cy="310.9" r="3.6" fill="#F2A03D" opacity="0.69"/><circle cx="147.2" cy="324.9" r="3.1" fill="#EE8A50" opacity="0.85"/><circle cx="148.4" cy="317.2" r="2.6" fill="#8FCB6B" opacity="0.79"/><circle cx="140.6" cy="352.9" r="2.2" fill="#F2A03D" opacity="0.92"/><circle cx="142.7" cy="306.6" r="3.1" fill="#8FCB6B" opacity="0.71"/><circle cx="150.8" cy="303.0" r="2.2" fill="#F2A03D" opacity="0.69"/><circle cx="183.9" cy="337.2" r="3.6" fill="#8FCB6B" opacity="0.75"/><circle cx="198.2" cy="371.2" r="3.1" fill="#8FCB6B" opacity="0.60"/><circle cx="167.7" cy="399.6" r="3.6" fill="#34D399" opacity="0.59"/><circle cx="95.0" cy="337.6" r="2.6" fill="#34D399" opacity="0.60"/><circle cx="150.2" cy="293.8" r="2.2" fill="#8FCB6B" opacity="0.62"/><circle cx="109.0" cy="379.9" r="2.2" fill="#34D399" opacity="0.61"/><circle cx="123.1" cy="337.7" r="2.6" fill="#8FCB6B" opacity="0.80"/><circle cx="159.3" cy="322.1" r="2.2" fill="#EE8A50" opacity="0.81"/><circle cx="157.4" cy="374.7" r="2.2" fill="#F2A03D" opacity="0.78"/><circle cx="184.6" cy="301.7" r="2.6" fill="#8FCB6B" opacity="0.59"/><circle cx="144.3" cy="377.6" r="2.6" fill="#8FCB6B" opacity="0.77"/><circle cx="130.5" cy="393.8" r="2.6" fill="#34D399" opacity="0.63"/><circle cx="163.8" cy="341.2" r="2.6" fill="#F2A03D" opacity="0.89"/><circle cx="177.0" cy="375.8" r="2.2" fill="#F2A03D" opacity="0.71"/><circle cx="150.2" cy="360.5" r="2.2" fill="#EE8A50" opacity="0.89"/><circle cx="155.7" cy="315.6" r="3.1" fill="#F2A03D" opacity="0.78"/><circle cx="180.6" cy="390.0" r="4.4" fill="#34D399" opacity="0.61"/><circle cx="176.3" cy="357.3" r="2.6" fill="#F2A03D" opacity="0.79"/><circle cx="156.0" cy="396.9" r="5.4" fill="#34D399" opacity="0.63"/><circle cx="153.7" cy="346.9" r="2.6" fill="#E06A5A" opacity="0.97"/><circle cx="135.1" cy="325.0" r="2.2" fill="#EE8A50" opacity="0.81"/><circle cx="119.6" cy="329.8" r="2.6" fill="#8FCB6B" opacity="0.75"/><circle cx="116.4" cy="305.3" r="4.4" fill="#34D399" opacity="0.62"/><circle cx="169.0" cy="374.5" r="2.6" fill="#8FCB6B" opacity="0.75"/><circle cx="141.5" cy="292.6" r="2.2" fill="#8FCB6B" opacity="0.61"/><circle cx="125.9" cy="345.5" r="2.2" fill="#F2A03D" opacity="0.82"/><circle cx="163.4" cy="312.5" r="2.2" fill="#34D399" opacity="0.74"/><circle cx="112.5" cy="321.6" r="2.6" fill="#34D399" opacity="0.68"/><circle cx="199.4" cy="320.9" r="3.1" fill="#34D399" opacity="0.60"/><circle cx="168.4" cy="334.5" r="2.2" fill="#8FCB6B" opacity="0.84"/><circle cx="132.9" cy="339.1" r="3.1" fill="#F2A03D" opacity="0.87"/><circle cx="190.8" cy="356.0" r="2.6" fill="#8FCB6B" opacity="0.69"/><circle cx="149.5" cy="381.2" r="2.2" fill="#34D399" opacity="0.74"/><circle cx="276" cy="328" r="50" fill="rgba(213,197,232,.05)"/><circle cx="250.0" cy="342.6" r="5.4" fill="#F2A03D" opacity="0.73"/><circle cx="262.7" cy="320.7" r="3.1" fill="#EE8A50" opacity="0.86"/><circle cx="307.0" cy="315.7" r="2.6" fill="#F2A03D" opacity="0.70"/><circle cx="303.9" cy="357.4" r="5.4" fill="#34D399" opacity="0.64"/><circle cx="251.5" cy="297.6" r="2.6" fill="#34D399" opacity="0.65"/><circle cx="250.0" cy="361.6" r="3.1" fill="#8FCB6B" opacity="0.62"/><circle cx="258.5" cy="285.9" r="5.4" fill="#34D399" opacity="0.59"/><circle cx="315.1" cy="328.7" r="3.1" fill="#34D399" opacity="0.65"/><circle cx="261.3" cy="306.5" r="3.6" fill="#8FCB6B" opacity="0.77"/><circle cx="290.5" cy="348.7" r="3.6" fill="#8FCB6B" opacity="0.77"/><circle cx="296.6" cy="289.4" r="2.6" fill="#34D399" opacity="0.61"/><circle cx="266.4" cy="362.6" r="4.4" fill="#34D399" opacity="0.68"/><circle cx="280.2" cy="368.3" r="3.1" fill="#8FCB6B" opacity="0.64"/><circle cx="283.5" cy="356.2" r="4.4" fill="#8FCB6B" opacity="0.74"/><circle cx="303.8" cy="305.3" r="2.2" fill="#8FCB6B" opacity="0.68"/><circle cx="313.1" cy="348.0" r="5.4" fill="#8FCB6B" opacity="0.62"/><circle cx="253.7" cy="320.0" r="2.6" fill="#F2A03D" opacity="0.79"/><circle cx="242.1" cy="333.7" r="3.6" fill="#8FCB6B" opacity="0.69"/><circle cx="265.2" cy="297.4" r="3.1" fill="#F2A03D" opacity="0.71"/><circle cx="272.0" cy="322.7" r="2.2" fill="#E06A5A" opacity="0.94"/><circle cx="300.0" cy="317.0" r="2.2" fill="#F2A03D" opacity="0.76"/><circle cx="232.6" cy="332.2" r="2.2" fill="#34D399" opacity="0.61"/><circle cx="261.9" cy="336.3" r="2.2" fill="#EE8A50" opacity="0.85"/><circle cx="234.6" cy="313.3" r="3.1" fill="#34D399" opacity="0.60"/><circle cx="264.4" cy="345.2" r="3.6" fill="#EE8A50" opacity="0.81"/><circle cx="300.0" cy="335.7" r="2.2" fill="#8FCB6B" opacity="0.77"/><circle cx="289.9" cy="299.5" r="5.4" fill="#34D399" opacity="0.71"/><circle cx="279.7" cy="336.0" r="2.2" fill="#E06A5A" opacity="0.92"/><circle cx="288.8" cy="313.0" r="2.2" fill="#F2A03D" opacity="0.82"/><circle cx="307.5" cy="324.2" r="3.6" fill="#8FCB6B" opacity="0.71"/><circle cx="303.4" cy="343.7" r="2.6" fill="#8FCB6B" opacity="0.72"/><circle cx="279.2" cy="343.7" r="3.6" fill="#8FCB6B" opacity="0.86"/><circle cx="271.7" cy="305.8" r="5.4" fill="#F2A03D" opacity="0.80"/><circle cx="285.0" cy="322.5" r="2.2" fill="#F2A03D" opacity="0.90"/><circle cx="239.6" cy="301.5" r="3.6" fill="#34D399" opacity="0.59"/><circle cx="245.3" cy="354.8" r="3.1" fill="#8FCB6B" opacity="0.63"/><circle cx="258.4" cy="359.2" r="2.2" fill="#8FCB6B" opacity="0.68"/><circle cx="298.6" cy="324.6" r="3.6" fill="#F2A03D" opacity="0.79"/><circle cx="275.6" cy="329.8" r="3.1" fill="#EE8A50" opacity="0.98"/><circle cx="243.0" cy="320.7" r="3.6" fill="#F2A03D" opacity="0.70"/><circle cx="288.2" cy="365.9" r="2.6" fill="#34D399" opacity="0.64"/><circle cx="255.3" cy="327.2" r="2.6" fill="#8FCB6B" opacity="0.81"/><circle cx="236.0" cy="348.1" r="4.4" fill="#8FCB6B" opacity="0.60"/><circle cx="317.5" cy="313.7" r="2.2" fill="#8FCB6B" opacity="0.61"/><circle cx="310.7" cy="304.2" r="2.2" fill="#34D399" opacity="0.62"/><circle cx="292.4" cy="334.0" r="2.2" fill="#EE8A50" opacity="0.84"/><circle cx="281.5" cy="310.6" r="2.6" fill="#8FCB6B" opacity="0.84"/><circle cx="273.0" cy="357.4" r="2.6" fill="#34D399" opacity="0.73"/><circle cx="281.0" cy="289.5" r="2.6" fill="#8FCB6B" opacity="0.65"/><circle cx="276.1" cy="316.6" r="2.2" fill="#E06A5A" opacity="0.90"/><circle cx="246.7" cy="307.5" r="2.2" fill="#34D399" opacity="0.68"/><circle cx="312.0" cy="335.9" r="3.1" fill="#F2A03D" opacity="0.67"/><circle cx="232.8" cy="324.8" r="3.6" fill="#34D399" opacity="0.61"/><circle cx="250.9" cy="333.0" r="2.6" fill="#8FCB6B" opacity="0.77"/><circle cx="290.1" cy="319.6" r="2.2" fill="#8FCB6B" opacity="0.85"/><circle cx="257.1" cy="365.8" r="2.6" fill="#8FCB6B" opacity="0.62"/><circle cx="249.7" cy="314.7" r="2.2" fill="#34D399" opacity="0.73"/><circle cx="318.7" cy="340.1" r="2.6" fill="#34D399" opacity="0.60"/><circle cx="272.5" cy="287.9" r="2.2" fill="#8FCB6B" opacity="0.64"/><circle cx="264.0" cy="352.8" r="2.2" fill="#F2A03D" opacity="0.75"/><circle cx="296.0" cy="363.5" r="3.1" fill="#8FCB6B" opacity="0.63"/><circle cx="267.5" cy="333.9" r="2.6" fill="#EE8A50" opacity="0.91"/><circle cx="301.1" cy="295.8" r="3.6" fill="#34D399" opacity="0.63"/><circle cx="319.1" cy="320.1" r="2.6" fill="#8FCB6B" opacity="0.61"/><circle cx="254.5" cy="353.6" r="3.1" fill="#F2A03D" opacity="0.70"/><circle cx="272.2" cy="351.0" r="2.2" fill="#F2A03D" opacity="0.79"/><circle cx="262.2" cy="371.8" r="2.6" fill="#8FCB6B" opacity="0.59"/><circle cx="294.7" cy="307.9" r="2.2" fill="#8FCB6B" opacity="0.75"/><circle cx="288.9" cy="340.6" r="3.1" fill="#F2A03D" opacity="0.84"/><circle cx="376" cy="370" r="44" fill="rgba(213,197,232,.05)"/><circle cx="403.3" cy="359.1" r="3.1" fill="#8FCB6B" opacity="0.70"/><circle cx="411.6" cy="374.0" r="2.2" fill="#34D399" opacity="0.63"/><circle cx="371.8" cy="391.9" r="4.4" fill="#F2A03D" opacity="0.77"/><circle cx="388.9" cy="378.0" r="3.1" fill="#EE8A50" opacity="0.85"/><circle cx="379.6" cy="331.7" r="4.4" fill="#8FCB6B" opacity="0.61"/><circle cx="354.0" cy="402.6" r="5.4" fill="#34D399" opacity="0.60"/><circle cx="387.3" cy="391.1" r="3.6" fill="#8FCB6B" opacity="0.75"/><circle cx="375.2" cy="352.9" r="3.1" fill="#EE8A50" opacity="0.82"/><circle cx="371.5" cy="343.1" r="5.4" fill="#8FCB6B" opacity="0.72"/><circle cx="402.7" cy="388.7" r="2.6" fill="#34D399" opacity="0.67"/><circle cx="381.8" cy="360.5" r="5.4" fill="#EE8A50" opacity="0.89"/><circle cx="384.6" cy="384.3" r="2.2" fill="#F2A03D" opacity="0.83"/><circle cx="396.7" cy="338.4" r="3.6" fill="#34D399" opacity="0.61"/><circle cx="375.1" cy="383.2" r="2.6" fill="#F2A03D" opacity="0.86"/><circle cx="346.4" cy="370.7" r="2.2" fill="#34D399" opacity="0.70"/><circle cx="376.8" cy="400.5" r="2.2" fill="#8FCB6B" opacity="0.69"/><circle cx="363.9" cy="383.4" r="3.6" fill="#F2A03D" opacity="0.82"/><circle cx="395.8" cy="349.3" r="2.2" fill="#34D399" opacity="0.71"/><circle cx="345.6" cy="355.6" r="2.6" fill="#34D399" opacity="0.66"/><circle cx="393.6" cy="402.7" r="2.2" fill="#34D399" opacity="0.62"/><circle cx="366.5" cy="360.6" r="4.4" fill="#EE8A50" opacity="0.86"/><circle cx="397.1" cy="380.4" r="2.2" fill="#34D399" opacity="0.76"/><circle cx="351.2" cy="347.2" r="3.1" fill="#34D399" opacity="0.66"/><circle cx="386.2" cy="344.5" r="2.6" fill="#34D399" opacity="0.72"/><circle cx="353.0" cy="366.3" r="2.2" fill="#F2A03D" opacity="0.76"/><circle cx="359.7" cy="339.1" r="2.6" fill="#34D399" opacity="0.64"/><circle cx="404.3" cy="370.8" r="2.2" fill="#F2A03D" opacity="0.71"/><circle cx="405.3" cy="343.2" r="3.1" fill="#34D399" opacity="0.59"/><circle cx="373.0" cy="372.4" r="2.6" fill="#F2A03D" opacity="0.96"/><circle cx="406.5" cy="382.0" r="3.6" fill="#34D399" opacity="0.67"/><circle cx="394.2" cy="386.9" r="3.1" fill="#F2A03D" opacity="0.75"/><circle cx="337.9" cy="380.1" r="3.6" fill="#34D399" opacity="0.60"/><circle cx="412.6" cy="358.9" r="2.6" fill="#34D399" opacity="0.61"/><circle cx="387.1" cy="404.6" r="2.2" fill="#34D399" opacity="0.63"/><circle cx="348.6" cy="389.1" r="3.1" fill="#34D399" opacity="0.66"/><circle cx="395.5" cy="360.4" r="2.6" fill="#8FCB6B" opacity="0.78"/><circle cx="337.6" cy="369.9" r="2.2" fill="#8FCB6B" opacity="0.61"/><circle cx="388.2" cy="350.4" r="2.2" fill="#8FCB6B" opacity="0.76"/><circle cx="365.3" cy="407.7" r="3.1" fill="#34D399" opacity="0.60"/><circle cx="381.6" cy="407.3" r="2.2" fill="#8FCB6B" opacity="0.61"/><circle cx="363.6" cy="373.6" r="2.2" fill="#EE8A50" opacity="0.87"/><circle cx="414.6" cy="367.7" r="3.1" fill="#34D399" opacity="0.60"/><circle cx="354.6" cy="357.3" r="2.2" fill="#F2A03D" opacity="0.75"/><circle cx="387.7" cy="368.0" r="2.2" fill="#F2A03D" opacity="0.88"/><circle cx="394.3" cy="370.8" r="3.1" fill="#F2A03D" opacity="0.81"/><circle cx="406.0" cy="352.6" r="2.2" fill="#8FCB6B" opacity="0.65"/><circle cx="399.9" cy="396.8" r="3.1" fill="#34D399" opacity="0.63"/><circle cx="380.0" cy="376.5" r="3.1" fill="#EE8A50" opacity="0.92"/><circle cx="370.7" cy="378.8" r="2.2" fill="#E06A5A" opacity="0.90"/><circle cx="364.1" cy="350.8" r="3.1" fill="#8FCB6B" opacity="0.77"/><circle cx="342.3" cy="362.5" r="3.1" fill="#34D399" opacity="0.65"/><circle cx="387.1" cy="334.7" r="2.2" fill="#8FCB6B" opacity="0.62"/><circle cx="367.7" cy="331.3" r="3.1" fill="#34D399" opacity="0.59"/><circle cx="353.7" cy="373.3" r="2.2" fill="#8FCB6B" opacity="0.77"/><circle cx="347.0" cy="377.8" r="2.2" fill="#34D399" opacity="0.69"/><circle cx="361.0" cy="389.9" r="2.2" fill="#F2A03D" opacity="0.75"/><circle cx="383.9" cy="397.7" r="2.2" fill="#F2A03D" opacity="0.71"/><circle cx="364.2" cy="398.2" r="2.2" fill="#8FCB6B" opacity="0.69"/><circle cx="373.8" cy="405.5" r="2.2" fill="#34D399" opacity="0.64"/><circle cx="450" cy="318" r="32" fill="rgba(213,197,232,.05)"/><circle cx="472.5" cy="330.2" r="2.2" fill="#34D399" opacity="0.64"/><circle cx="463.4" cy="312.4" r="2.6" fill="#8FCB6B" opacity="0.80"/><circle cx="424.1" cy="329.8" r="2.2" fill="#34D399" opacity="0.60"/><circle cx="439.2" cy="323.2" r="2.2" fill="#F2A03D" opacity="0.83"/><circle cx="475.2" cy="316.4" r="5.4" fill="#34D399" opacity="0.65"/><circle cx="445.0" cy="313.0" r="4.4" fill="#F2A03D" opacity="0.90"/><circle cx="445.6" cy="297.4" r="2.6" fill="#34D399" opacity="0.70"/><circle cx="453.8" cy="332.5" r="3.6" fill="#8FCB6B" opacity="0.79"/><circle cx="447.2" cy="323.0" r="3.6" fill="#EE8A50" opacity="0.92"/><circle cx="462.1" cy="322.5" r="3.1" fill="#EE8A50" opacity="0.82"/><circle cx="431.8" cy="310.4" r="2.2" fill="#F2A03D" opacity="0.72"/><circle cx="458.0" cy="299.4" r="2.6" fill="#34D399" opacity="0.72"/><circle cx="455.2" cy="346.2" r="3.1" fill="#34D399" opacity="0.60"/><circle cx="455.5" cy="307.5" r="3.6" fill="#F2A03D" opacity="0.83"/><circle cx="441.9" cy="305.4" r="2.2" fill="#8FCB6B" opacity="0.79"/><circle cx="432.4" cy="334.4" r="2.2" fill="#34D399" opacity="0.66"/><circle cx="433.0" cy="319.8" r="3.1" fill="#8FCB6B" opacity="0.76"/><circle cx="422.7" cy="307.4" r="4.4" fill="#34D399" opacity="0.59"/><circle cx="425.0" cy="320.8" r="2.6" fill="#8FCB6B" opacity="0.65"/><circle cx="466.1" cy="299.1" r="3.6" fill="#34D399" opacity="0.65"/><circle cx="450.7" cy="292.8" r="2.6" fill="#8FCB6B" opacity="0.65"/><circle cx="436.7" cy="341.7" r="2.2" fill="#34D399" opacity="0.62"/><circle cx="436.0" cy="329.6" r="2.2" fill="#F2A03D" opacity="0.75"/><circle cx="453.1" cy="316.1" r="2.6" fill="#F2A03D" opacity="0.95"/><circle cx="431.6" cy="299.6" r="3.6" fill="#34D399" opacity="0.63"/><circle cx="442.1" cy="332.9" r="2.2" fill="#8FCB6B" opacity="0.76"/><circle cx="448.7" cy="341.6" r="2.6" fill="#8FCB6B" opacity="0.67"/><circle cx="465.9" cy="339.2" r="2.2" fill="#34D399" opacity="0.63"/><circle cx="462.2" cy="331.2" r="2.2" fill="#34D399" opacity="0.75"/><circle cx="474.2" cy="303.1" r="2.6" fill="#34D399" opacity="0.60"/><circle cx="450.5" cy="302.5" r="2.2" fill="#F2A03D" opacity="0.78"/><circle cx="443.4" cy="346.2" r="3.1" fill="#34D399" opacity="0.59"/><circle cx="438.7" cy="292.6" r="2.6" fill="#34D399" opacity="0.61"/><circle cx="467.6" cy="327.1" r="2.2" fill="#F2A03D" opacity="0.72"/><circle cx="461.7" cy="291.2" r="3.6" fill="#34D399" opacity="0.59"/><circle cx="459.5" cy="340.9" r="2.2" fill="#34D399" opacity="0.65"/><circle cx="420.8" cy="315.7" r="2.6" fill="#34D399" opacity="0.59"/><circle cx="469.2" cy="308.4" r="3.1" fill="#F2A03D" opacity="0.70"/><circle cx="208" cy="406" r="30" fill="rgba(213,197,232,.05)"/><circle cx="215.5" cy="430.4" r="3.6" fill="#34D399" opacity="0.62"/><circle cx="194.3" cy="402.3" r="2.2" fill="#EE8A50" opacity="0.79"/><circle cx="230.9" cy="399.5" r="4.4" fill="#34D399" opacity="0.64"/><circle cx="206.4" cy="403.2" r="5.4" fill="#E06A5A" opacity="0.95"/><circle cx="226.9" cy="421.4" r="3.6" fill="#8FCB6B" opacity="0.63"/><circle cx="220.2" cy="385.5" r="3.1" fill="#8FCB6B" opacity="0.64"/><circle cx="194.7" cy="393.1" r="4.4" fill="#34D399" opacity="0.72"/><circle cx="187.7" cy="410.6" r="4.4" fill="#8FCB6B" opacity="0.69"/><circle cx="201.9" cy="431.3" r="2.2" fill="#34D399" opacity="0.61"/><circle cx="188.1" cy="422.5" r="4.4" fill="#8FCB6B" opacity="0.61"/><circle cx="219.4" cy="416.8" r="2.2" fill="#F2A03D" opacity="0.76"/><circle cx="200.7" cy="420.4" r="2.2" fill="#8FCB6B" opacity="0.76"/><circle cx="212.8" cy="420.4" r="2.2" fill="#F2A03D" opacity="0.77"/><circle cx="201.7" cy="411.5" r="2.2" fill="#F2A03D" opacity="0.87"/><circle cx="218.4" cy="401.3" r="3.1" fill="#EE8A50" opacity="0.83"/><circle cx="206.6" cy="386.7" r="2.2" fill="#8FCB6B" opacity="0.71"/><circle cx="211.4" cy="394.7" r="2.6" fill="#EE8A50" opacity="0.82"/><circle cx="182.8" cy="402.1" r="2.6" fill="#34D399" opacity="0.62"/><circle cx="212.6" cy="412.1" r="3.1" fill="#EE8A50" opacity="0.89"/><circle cx="232.5" cy="416.4" r="2.2" fill="#34D399" opacity="0.60"/><circle cx="221.5" cy="393.6" r="3.1" fill="#8FCB6B" opacity="0.73"/><circle cx="199.4" cy="384.2" r="2.6" fill="#34D399" opacity="0.65"/><circle cx="208.0" cy="428.0" r="2.6" fill="#8FCB6B" opacity="0.67"/><circle cx="211.8" cy="379.2" r="4.4" fill="#34D399" opacity="0.59"/><circle cx="206.9" cy="420.1" r="2.2" fill="#8FCB6B" opacity="0.79"/><circle cx="229.8" cy="390.6" r="2.6" fill="#34D399" opacity="0.60"/><circle cx="195.7" cy="415.7" r="2.6" fill="#8FCB6B" opacity="0.77"/><circle cx="228.3" cy="408.5" r="3.1" fill="#F2A03D" opacity="0.69"/><circle cx="189.0" cy="387.5" r="2.2" fill="#34D399" opacity="0.60"/></g><text x="70" y="424" class="cap">2,802 files scored</text><text x="526" y="280" class="cap">OPEN FINDINGS</text><rect x="526" y="300" width="236" height="24" rx="6" fill="#110D17"/><rect x="526" y="300" width="3" height="24" rx="1.5" fill="#E06A5A"/><text x="539" y="316" class="find">CHANGE ENTROPY</text><text x="752" y="316" class="find" fill="#E06A5A" text-anchor="end" font-weight="700">−3.0</text><rect x="526" y="328" width="236" height="24" rx="6" fill="#110D17"/><rect x="526" y="328" width="3" height="24" rx="1.5" fill="#E06A5A"/><text x="539" y="344" class="find">NESTED COMPLEXITY</text><text x="752" y="344" class="find" fill="#E06A5A" text-anchor="end" font-weight="700">−2.4</text><rect x="526" y="356" width="236" height="24" rx="6" fill="#110D17"/><rect x="526" y="356" width="3" height="24" rx="1.5" fill="#F2A03D"/><text x="539" y="372" class="find">CHURN RISK</text><text x="752" y="372" class="find" fill="#F2A03D" text-anchor="end" font-weight="700">−1.8</text><rect x="526" y="384" width="236" height="24" rx="6" fill="#110D17"/><rect x="526" y="384" width="3" height="24" rx="1.5" fill="#F2A03D"/><text x="539" y="400" class="find">CO-CHANGE SCATTER</text><text x="752" y="400" class="find" fill="#F2A03D" text-anchor="end" font-weight="700">−1.5</text><rect x="526" y="412" width="236" height="24" rx="6" fill="#110D17"/><rect x="526" y="412" width="3" height="24" rx="1.5" fill="#F2A03D"/><text x="539" y="428" class="find">UNTESTED HOTSPOT</text><text x="752" y="428" class="find" fill="#F2A03D" text-anchor="end" font-weight="700">−1.2</text><rect x="802" y="104" width="380" height="352" rx="14" fill="#211B29" stroke="rgba(213,197,232,.10)" stroke-width="1"/><text x="824" y="135" class="lbl">DEPENDENCY GRAPH</text><text x="1160" y="135" class="kick" text-anchor="end">16 languages</text><rect x="820" y="150" width="344" height="240" rx="9" fill="#110D17"/><clipPath id="gclip"><rect x="820" y="150" width="344" height="240" rx="9"/></clipPath><g clip-path="url(#gclip)"><line x1="944.0" y1="201.8" x2="912" y2="224" stroke="rgba(213,197,232,.15)" stroke-width="1"/><line x1="907.3" y1="185.9" x2="912" y2="224" stroke="rgba(213,197,232,.15)" stroke-width="1"/><line x1="913.1" y1="193.3" x2="912" y2="224" stroke="rgba(213,197,232,.15)" stroke-width="1"/><line x1="948.1" y1="258.8" x2="912" y2="224" stroke="rgba(213,197,232,.15)" stroke-width="1"/><line x1="901.9" y1="182.8" x2="912" y2="224" stroke="rgba(213,197,232,.15)" stroke-width="1"/><line x1="895.4" y1="193.7" x2="912" y2="224" stroke="rgba(213,197,232,.15)" stroke-width="1"/><line x1="888.8" y1="197.7" x2="912" y2="224" stroke="rgba(213,197,232,.15)" stroke-width="1"/><line x1="929.0" y1="238.4" x2="912" y2="224" stroke="rgba(213,197,232,.15)" stroke-width="1"/><line x1="951.9" y1="214.2" x2="912" y2="224" stroke="rgba(213,197,232,.15)" stroke-width="1"/><line x1="1040.2" y1="228.2" x2="1052" y2="212" stroke="rgba(213,197,232,.15)" stroke-width="1"/><line x1="1048.4" y1="190.3" x2="1052" y2="212" stroke="rgba(213,197,232,.15)" stroke-width="1"/><line x1="1068.5" y1="249.3" x2="1052" y2="212" stroke="rgba(213,197,232,.15)" stroke-width="1"/><line x1="1013.8" y1="237.1" x2="1052" y2="212" stroke="rgba(213,197,232,.15)" stroke-width="1"/><line x1="1009.6" y1="237.8" x2="1052" y2="212" stroke="rgba(213,197,232,.15)" stroke-width="1"/><line x1="1065.3" y1="185.2" x2="1052" y2="212" stroke="rgba(213,197,232,.15)" stroke-width="1"/><line x1="1017.4" y1="194.6" x2="1052" y2="212" stroke="rgba(213,197,232,.15)" stroke-width="1"/><line x1="1107.4" y1="209.0" x2="1052" y2="212" stroke="rgba(213,197,232,.15)" stroke-width="1"/><line x1="1061.0" y1="174.5" x2="1052" y2="212" stroke="rgba(213,197,232,.15)" stroke-width="1"/><line x1="944.8" y1="323.3" x2="982" y2="322" stroke="rgba(213,197,232,.15)" stroke-width="1"/><line x1="955.1" y1="338.1" x2="982" y2="322" stroke="rgba(213,197,232,.15)" stroke-width="1"/><line x1="1001.7" y1="341.5" x2="982" y2="322" stroke="rgba(213,197,232,.15)" stroke-width="1"/><line x1="1022.9" y1="322.0" x2="982" y2="322" stroke="rgba(213,197,232,.15)" stroke-width="1"/><line x1="961.2" y1="356.5" x2="982" y2="322" stroke="rgba(213,197,232,.15)" stroke-width="1"/><line x1="942.1" y1="301.8" x2="982" y2="322" stroke="rgba(213,197,232,.15)" stroke-width="1"/><line x1="943.8" y1="346.7" x2="982" y2="322" stroke="rgba(213,197,232,.15)" stroke-width="1"/><line x1="983.8" y1="296.5" x2="982" y2="322" stroke="rgba(213,197,232,.15)" stroke-width="1"/><line x1="949.1" y1="320.7" x2="982" y2="322" stroke="rgba(213,197,232,.15)" stroke-width="1"/><line x1="1131.2" y1="323.1" x2="1098" y2="310" stroke="rgba(213,197,232,.15)" stroke-width="1"/><line x1="1085.5" y1="327.7" x2="1098" y2="310" stroke="rgba(213,197,232,.15)" stroke-width="1"/><line x1="1106.7" y1="286.0" x2="1098" y2="310" stroke="rgba(213,197,232,.15)" stroke-width="1"/><line x1="1152.2" y1="302.0" x2="1098" y2="310" stroke="rgba(213,197,232,.15)" stroke-width="1"/><line x1="1068.9" y1="335.5" x2="1098" y2="310" stroke="rgba(213,197,232,.15)" stroke-width="1"/><line x1="1148.6" y1="288.2" x2="1098" y2="310" stroke="rgba(213,197,232,.15)" stroke-width="1"/><line x1="1064.8" y1="323.3" x2="1098" y2="310" stroke="rgba(213,197,232,.15)" stroke-width="1"/><line x1="1046.2" y1="319.4" x2="1098" y2="310" stroke="rgba(213,197,232,.15)" stroke-width="1"/><line x1="1137.2" y1="290.0" x2="1098" y2="310" stroke="rgba(213,197,232,.15)" stroke-width="1"/><line x1="912" y1="224" x2="1052" y2="212" stroke="rgba(213,197,232,.30)" stroke-width="1.6"/><line x1="912" y1="224" x2="982" y2="322" stroke="rgba(213,197,232,.30)" stroke-width="1.6"/><line x1="912" y1="224" x2="1098" y2="310" stroke="rgba(213,197,232,.30)" stroke-width="1.6"/><line x1="1052" y1="212" x2="982" y2="322" stroke="rgba(213,197,232,.30)" stroke-width="1.6"/><line x1="982" y1="322" x2="1098" y2="310" stroke="rgba(213,197,232,.30)" stroke-width="1.6"/><circle cx="912.0" cy="224.0" r="9.0" fill="#F59520" stroke="#211B29" stroke-width="1.5"/><circle cx="944.0" cy="201.8" r="4.0" fill="#C97A1A"/><circle cx="907.3" cy="185.9" r="3.3" fill="#C97A1A"/><circle cx="913.1" cy="193.3" r="2.8" fill="#C97A1A"/><circle cx="948.1" cy="258.8" r="3.8" fill="#C97A1A"/><circle cx="901.9" cy="182.8" r="3.1" fill="#C97A1A"/><circle cx="895.4" cy="193.7" r="3.4" fill="#C97A1A"/><circle cx="888.8" cy="197.7" r="3.3" fill="#C97A1A"/><circle cx="929.0" cy="238.4" r="3.8" fill="#C97A1A"/><circle cx="951.9" cy="214.2" r="3.6" fill="#C97A1A"/><circle cx="1052.0" cy="212.0" r="9.0" fill="#A98FC4" stroke="#211B29" stroke-width="1.5"/><circle cx="1040.2" cy="228.2" r="3.4" fill="#7D659C"/><circle cx="1048.4" cy="190.3" r="3.5" fill="#7D659C"/><circle cx="1068.5" cy="249.3" r="3.5" fill="#7D659C"/><circle cx="1013.8" cy="237.1" r="3.7" fill="#7D659C"/><circle cx="1009.6" cy="237.8" r="3.8" fill="#7D659C"/><circle cx="1065.3" cy="185.2" r="4.1" fill="#7D659C"/><circle cx="1017.4" cy="194.6" r="4.2" fill="#7D659C"/><circle cx="1107.4" cy="209.0" r="3.7" fill="#7D659C"/><circle cx="1061.0" cy="174.5" r="3.3" fill="#7D659C"/><circle cx="982.0" cy="322.0" r="9.0" fill="#E06A5A" stroke="#211B29" stroke-width="1.5"/><circle cx="944.8" cy="323.3" r="4.2" fill="#B04C3E"/><circle cx="955.1" cy="338.1" r="4.0" fill="#B04C3E"/><circle cx="1001.7" cy="341.5" r="3.2" fill="#B04C3E"/><circle cx="1022.9" cy="322.0" r="3.0" fill="#B04C3E"/><circle cx="961.2" cy="356.5" r="4.0" fill="#B04C3E"/><circle cx="942.1" cy="301.8" r="3.5" fill="#B04C3E"/><circle cx="943.8" cy="346.7" r="3.7" fill="#B04C3E"/><circle cx="983.8" cy="296.5" r="2.9" fill="#B04C3E"/><circle cx="949.1" cy="320.7" r="2.6" fill="#B04C3E"/><circle cx="1098.0" cy="310.0" r="9.0" fill="#A9BB6F" stroke="#211B29" stroke-width="1.5"/><circle cx="1131.2" cy="323.1" r="3.2" fill="#7E8F4E"/><circle cx="1085.5" cy="327.7" r="4.3" fill="#7E8F4E"/><circle cx="1106.7" cy="286.0" r="2.9" fill="#7E8F4E"/><circle cx="1152.2" cy="302.0" r="3.9" fill="#7E8F4E"/><circle cx="1068.9" cy="335.5" r="2.8" fill="#7E8F4E"/><circle cx="1148.6" cy="288.2" r="4.2" fill="#7E8F4E"/><circle cx="1064.8" cy="323.3" r="3.6" fill="#7E8F4E"/><circle cx="1046.2" cy="319.4" r="4.0" fill="#7E8F4E"/><circle cx="1137.2" cy="290.0" r="2.9" fill="#7E8F4E"/></g><text x="832" y="380" class="cap">28,126 nodes · 11 communities</text><text x="824" y="414" class="out">Who calls this? What breaks if</text><text x="824" y="434" class="out">I change it?</text><rect x="1200" y="104" width="364" height="352" rx="14" fill="#211B29" stroke="rgba(213,197,232,.10)" stroke-width="1"/><text x="1222" y="135" class="lbl">GIT HISTORY</text><text x="1542" y="135" class="kick" text-anchor="end">1,050 commits</text><rect x="1218" y="150" width="328" height="240" rx="9" fill="#110D17"/><rect x="1230.0" y="312.2" width="16.0" height="13.8" rx="2.5" fill="#34D399" opacity="0.82"/><rect x="1249.0" y="300.7" width="16.0" height="25.3" rx="2.5" fill="#34D399" opacity="0.82"/><rect x="1268.0" y="282.3" width="16.0" height="43.7" rx="2.5" fill="#34D399" opacity="0.82"/><rect x="1287.0" y="261.6" width="16.0" height="64.4" rx="2.5" fill="#34D399" opacity="0.82"/><rect x="1306.0" y="231.7" width="16.0" height="94.3" rx="2.5" fill="#34D399" opacity="0.82"/><rect x="1325.0" y="206.4" width="16.0" height="119.6" rx="2.5" fill="#34D399" opacity="0.82"/><rect x="1344.0" y="181.1" width="16.0" height="144.9" rx="2.5" fill="#F2A03D" opacity="0.82"/><rect x="1363.0" y="192.6" width="16.0" height="133.4" rx="2.5" fill="#F2A03D" opacity="0.82"/><rect x="1382.0" y="217.9" width="16.0" height="108.1" rx="2.5" fill="#F2A03D" opacity="0.82"/><rect x="1401.0" y="243.2" width="16.0" height="82.8" rx="2.5" fill="#F2A03D" opacity="0.82"/><rect x="1420.0" y="263.9" width="16.0" height="62.1" rx="2.5" fill="#F2A03D" opacity="0.82"/><rect x="1439.0" y="282.3" width="16.0" height="43.7" rx="2.5" fill="#E06A5A" opacity="0.82"/><rect x="1458.0" y="296.1" width="16.0" height="29.9" rx="2.5" fill="#E06A5A" opacity="0.82"/><rect x="1477.0" y="305.3" width="16.0" height="20.7" rx="2.5" fill="#E06A5A" opacity="0.82"/><rect x="1496.0" y="312.2" width="16.0" height="13.8" rx="2.5" fill="#E06A5A" opacity="0.82"/><rect x="1515.0" y="316.8" width="16.0" height="9.2" rx="2.5" fill="#E06A5A" opacity="0.82"/><line x1="1230" y1="327" x2="1534" y2="327" stroke="rgba(213,197,232,.20)"/><text x="1230" y="348" class="cap">low risk</text><text x="1534" y="348" class="cap" text-anchor="end">high risk</text><text x="1230" y="176" class="cap">CHANGE-RISK DISTRIBUTION</text><text x="1230" y="380" class="cap">hotspots · ownership · co-change · bus factor</text><text x="1222" y="414" class="out">Which of these 40 files is</text><text x="1222" y="434" class="out">actually dangerous?</text><rect x="36" y="474" width="496" height="302" rx="14" fill="#211B29" stroke="rgba(213,197,232,.10)" stroke-width="1"/><text x="58" y="505" class="lbl">GENERATED DOCS</text><text x="510" y="505" class="kick" text-anchor="end">1,274 pages</text><rect x="54" y="520" width="460" height="176" rx="9" fill="#110D17"/><rect x="70" y="536" width="265" height="11" rx="3" fill="#A98FC4" opacity=".85"/><rect x="70" y="562" width="394" height="6" rx="3" fill="rgba(213,197,232,.24)"/><rect x="70" y="578" width="368" height="6" rx="3" fill="rgba(213,197,232,.24)"/><rect x="70" y="594" width="304" height="6" rx="3" fill="rgba(213,197,232,.24)"/><rect x="70" y="607" width="188" height="9" rx="3" fill="#F59520" opacity=".55"/><rect x="70" y="630" width="377" height="6" rx="3" fill="rgba(213,197,232,.24)"/><rect x="70" y="646" width="338" height="6" rx="3" fill="rgba(213,197,232,.24)"/><rect x="70" y="662" width="235" height="6" rx="3" fill="rgba(213,197,232,.24)"/><rect x="416" y="530" width="86" height="20" rx="10" fill="rgba(52,211,153,.14)"/><text x="459" y="544" class="pill" fill="#34D399" text-anchor="middle">fresh · v6</text><text x="58" y="730" class="out">Why does auth work this way?</text><text x="58" y="751" class="outsub">Answered from the wiki, not from 12 file reads.</text><rect x="550" y="474" width="496" height="302" rx="14" fill="#211B29" stroke="rgba(213,197,232,.10)" stroke-width="1"/><text x="572" y="505" class="lbl">ARCHITECTURAL DECISIONS</text><text x="1024" y="505" class="kick" text-anchor="end">mined from 8 sources</text><rect x="568" y="520" width="460" height="176" rx="9" fill="#110D17"/><text x="584" y="548" class="dec">Rebuild the KG skeleton only when the</text><text x="584" y="568" class="dec">graph fingerprint changes</text><rect x="584" y="582" width="392" height="1" fill="rgba(213,197,232,.15)"/><rect x="584" y="596" width="101" height="21" rx="10.5" fill="#211B29" stroke="rgba(213,197,232,.18)"/><text x="635" y="610" class="pill" fill="#34D399" text-anchor="middle">Verified quote</text><rect x="693.4" y="596" width="108" height="21" rx="10.5" fill="#211B29" stroke="rgba(213,197,232,.18)"/><text x="747" y="610" class="pill" fill="#A98FC4" text-anchor="middle">supersedes #418</text><rect x="809.4" y="596" width="115" height="21" rx="10.5" fill="#211B29" stroke="rgba(213,197,232,.18)"/><text x="867" y="610" class="pill" fill="#F59520" text-anchor="middle">governs 24 files</text><text x="584" y="660" class="cap">Delivered back at session start, and the moment</text><text x="584" y="676" class="cap">your agent edits a file the decision governs.</text><text x="572" y="730" class="out">Don’t re-litigate what we settled.</text><text x="572" y="751" class="outsub">Captured nowhere else.</text><rect x="1064" y="474" width="500" height="302" rx="14" fill="#211B29" stroke="rgba(213,197,232,.10)" stroke-width="1"/><text x="1086" y="505" class="lbl">SERVED TO YOUR AGENT</text><text x="1542" y="505" class="kick" text-anchor="end">10 MCP tools</text><rect x="1084" y="528" width="95" height="24" rx="6" fill="#110D17" stroke="rgba(213,197,232,.10)"/><text x="1131" y="544" class="mono" text-anchor="middle">get_overview</text><rect x="1187" y="528" width="81" height="24" rx="6" fill="#110D17" stroke="rgba(213,197,232,.10)"/><text x="1227" y="544" class="mono" text-anchor="middle">get_answer</text><rect x="1276" y="528" width="88" height="24" rx="6" fill="#110D17" stroke="rgba(213,197,232,.10)"/><text x="1320" y="544" class="mono" text-anchor="middle">get_context</text><rect x="1372" y="528" width="81" height="24" rx="6" fill="#110D17" stroke="rgba(213,197,232,.10)"/><text x="1412" y="544" class="mono" text-anchor="middle">get_symbol</text><rect x="1084" y="558" width="116" height="24" rx="6" fill="#110D17" stroke="rgba(213,197,232,.10)"/><text x="1142" y="574" class="mono" text-anchor="middle">search_codebase</text><rect x="1208" y="558" width="67" height="24" rx="6" fill="#110D17" stroke="rgba(213,197,232,.10)"/><text x="1241" y="574" class="mono" text-anchor="middle">get_risk</text><rect x="1283" y="558" width="116" height="24" rx="6" fill="#110D17" stroke="rgba(213,197,232,.10)"/><text x="1340" y="574" class="mono" text-anchor="middle">get_change_risk</text><rect x="1406" y="558" width="60" height="24" rx="6" fill="#110D17" stroke="rgba(213,197,232,.10)"/><text x="1436" y="574" class="mono" text-anchor="middle">get_why</text><rect x="1084" y="588" width="102" height="24" rx="6" fill="#110D17" stroke="rgba(213,197,232,.10)"/><text x="1135" y="604" class="mono" text-anchor="middle">get_dead_code</text><rect x="1194" y="588" width="81" height="24" rx="6" fill="#110D17" stroke="rgba(213,197,232,.10)"/><text x="1234" y="604" class="mono" text-anchor="middle">get_health</text><text x="1084" y="636" class="cap">Every response carries index age, indexed commit, and a staleness warning.</text><rect x="1084" y="650" width="460" height="46" rx="9" fill="rgba(245,149,32,.12)" stroke="rgba(245,149,32,.32)"/><text x="1098" y="670" class="chipval" fill="#F59520" font-size="19">−96%<tspan class="chipunit" fill="#A79DB3"> context tokens</tspan></text><text x="1098" y="688" class="cap">−70% tool calls · −89% file reads · fewer round-trips per answer</text><text x="1086" y="730" class="out">Claude Code, Codex, Cursor, VS Code.</text><text x="1086" y="751" class="outsub">Plus hooks, so it arrives unasked.</text></svg>\ No newline at end of file+</style><rect width="1600" height="812" fill="#17131D"/><text x="36" y="50" class="h1">One index. Everything your agent and your team keep asking for.</text><text x="36" y="80" class="sub">Index once, in seconds, no LLM required. Every answer below is already computed.</text><rect x="36" y="104" width="748" height="352" rx="14" fill="#211B29" stroke="rgba(213,197,232,.10)" stroke-width="1"/><text x="58" y="135" class="lbl">CODE HEALTH</text><text x="762" y="135" class="kick" text-anchor="end">49 detectors · zero LLM · <30s</text><rect x="58" y="150" width="146" height="58" rx="9" fill="#110D17" stroke="rgba(213,197,232,.10)"/><text x="70" y="170" class="chiplbl">Defect risk</text><text x="70" y="197" class="chipval" fill="#F2A03D">7.5<tspan class="chipunit">/10</tspan></text><rect x="214" y="150" width="146" height="58" rx="9" fill="#110D17" stroke="rgba(213,197,232,.10)"/><text x="226" y="170" class="chiplbl">Maintainability</text><text x="226" y="197" class="chipval" fill="#34D399">8.6<tspan class="chipunit">/10</tspan></text><rect x="370" y="150" width="146" height="58" rx="9" fill="#110D17" stroke="rgba(213,197,232,.10)"/><text x="382" y="170" class="chiplbl">Perf risks</text><text x="382" y="197" class="chipval" fill="#A98FC4">268<tspan class="chipunit"></tspan></text><rect x="58" y="220" width="704" height="34" rx="8" fill="rgba(52,211,153,.10)" stroke="rgba(52,211,153,.26)"/><circle cx="76" cy="237" r="4.5" fill="#34D399"/><text x="90" y="242" class="trust">17 of the 20 lowest-health files were bug-fixed in the last 6 months, <tspan font-weight="700">3.63× the 23% baseline</tspan></text><rect x="58" y="266" width="452" height="170" rx="9" fill="#110D17"/><clipPath id="galaxy"><rect x="58" y="266" width="452" height="170" rx="9"/></clipPath><g clip-path="url(#galaxy)"><circle cx="150" cy="346" r="62" fill="rgba(213,197,232,.05)"/><circle cx="140.1" cy="365.8" r="5.4" fill="#8FCB6B" opacity="0.84"/><circle cx="157.6" cy="330.2" r="4.4" fill="#F2A03D" opacity="0.87"/><circle cx="139.0" cy="345.5" r="3.6" fill="#EE8A50" opacity="0.92"/><circle cx="152.5" cy="388.3" r="2.2" fill="#F2A03D" opacity="0.69"/><circle cx="169.2" cy="364.9" r="5.4" fill="#F2A03D" opacity="0.80"/><circle cx="190.4" cy="362.5" r="2.2" fill="#F2A03D" opacity="0.68"/><circle cx="200.6" cy="361.2" r="3.1" fill="#34D399" opacity="0.62"/><circle cx="108.3" cy="335.1" r="4.4" fill="#8FCB6B" opacity="0.69"/><circle cx="168.3" cy="385.4" r="5.4" fill="#34D399" opacity="0.68"/><circle cx="189.4" cy="373.7" r="4.4" fill="#34D399" opacity="0.65"/><circle cx="156.6" cy="307.6" r="3.6" fill="#8FCB6B" opacity="0.72"/><circle cx="150.2" cy="370.2" r="2.6" fill="#8FCB6B" opacity="0.82"/><circle cx="137.5" cy="384.2" r="3.1" fill="#8FCB6B" opacity="0.71"/><circle cx="136.7" cy="400.9" r="2.2" fill="#34D399" opacity="0.59"/><circle cx="100.7" cy="372.4" r="2.2" fill="#34D399" opacity="0.59"/><circle cx="102.1" cy="322.4" r="3.1" fill="#34D399" opacity="0.61"/><circle cx="126.3" cy="378.5" r="3.6" fill="#34D399" opacity="0.71"/><circle cx="142.8" cy="333.9" r="5.4" fill="#F2A03D" opacity="0.90"/><circle cx="118.3" cy="391.4" r="3.1" fill="#34D399" opacity="0.60"/><circle cx="160.3" cy="355.3" r="3.1" fill="#F2A03D" opacity="0.90"/><circle cx="169.8" cy="323.3" r="3.6" fill="#EE8A50" opacity="0.78"/><circle cx="135.6" cy="313.9" r="2.6" fill="#8FCB6B" opacity="0.74"/><circle cx="201.9" cy="349.9" r="2.6" fill="#34D399" opacity="0.62"/><circle cx="186.9" cy="346.9" r="3.1" fill="#8FCB6B" opacity="0.73"/><circle cx="193.9" cy="331.8" r="5.4" fill="#34D399" opacity="0.66"/><circle cx="190.7" cy="316.3" r="5.4" fill="#8FCB6B" opacity="0.63"/><circle cx="121.9" cy="368.5" r="2.2" fill="#8FCB6B" opacity="0.74"/><circle cx="129.8" cy="360.6" r="2.6" fill="#F2A03D" opacity="0.82"/><circle cx="172.2" cy="346.0" r="2.2" fill="#EE8A50" opacity="0.84"/><circle cx="118.4" cy="349.9" r="2.6" fill="#8FCB6B" opacity="0.77"/><circle cx="119.4" cy="382.2" r="2.2" fill="#8FCB6B" opacity="0.66"/><circle cx="178.5" cy="327.5" r="2.6" fill="#8FCB6B" opacity="0.75"/><circle cx="111.6" cy="312.9" r="2.6" fill="#8FCB6B" opacity="0.63"/><circle cx="170.4" cy="301.4" r="2.6" fill="#34D399" opacity="0.64"/><circle cx="101.3" cy="348.2" r="2.2" fill="#8FCB6B" opacity="0.65"/><circle cx="205.6" cy="341.8" r="3.1" fill="#34D399" opacity="0.60"/><circle cx="181.3" cy="369.4" r="3.1" fill="#34D399" opacity="0.72"/><circle cx="160.0" cy="297.6" r="3.6" fill="#8FCB6B" opacity="0.64"/><circle cx="108.4" cy="364.3" r="2.2" fill="#8FCB6B" opacity="0.67"/><circle cx="159.8" cy="363.8" r="2.6" fill="#EE8A50" opacity="0.85"/><circle cx="104.9" cy="356.0" r="4.4" fill="#8FCB6B" opacity="0.66"/><circle cx="99.9" cy="331.7" r="2.2" fill="#8FCB6B" opacity="0.62"/><circle cx="130.0" cy="330.2" r="3.1" fill="#F2A03D" opacity="0.82"/><circle cx="133.2" cy="295.3" r="3.1" fill="#8FCB6B" opacity="0.61"/><circle cx="128.6" cy="318.9" r="3.1" fill="#F2A03D" opacity="0.75"/><circle cx="176.2" cy="340.6" r="2.2" fill="#F2A03D" opacity="0.81"/><circle cx="93.4" cy="350.5" r="2.6" fill="#34D399" opacity="0.59"/><circle cx="120.0" cy="362.2" r="2.2" fill="#F2A03D" opacity="0.75"/><circle cx="120.9" cy="317.3" r="2.2" fill="#34D399" opacity="0.70"/><circle cx="173.8" cy="332.6" r="2.2" fill="#8FCB6B" opacity="0.80"/><circle cx="177.5" cy="308.0" r="3.1" fill="#34D399" opacity="0.66"/><circle cx="149.1" cy="353.3" r="2.2" fill="#E06A5A" opacity="0.95"/><circle cx="143.4" cy="396.7" r="3.1" fill="#34D399" opacity="0.63"/><circle cx="126.6" cy="310.9" r="3.6" fill="#F2A03D" opacity="0.69"/><circle cx="147.2" cy="324.9" r="3.1" fill="#EE8A50" opacity="0.85"/><circle cx="148.4" cy="317.2" r="2.6" fill="#8FCB6B" opacity="0.79"/><circle cx="140.6" cy="352.9" r="2.2" fill="#F2A03D" opacity="0.92"/><circle cx="142.7" cy="306.6" r="3.1" fill="#8FCB6B" opacity="0.71"/><circle cx="150.8" cy="303.0" r="2.2" fill="#F2A03D" opacity="0.69"/><circle cx="183.9" cy="337.2" r="3.6" fill="#8FCB6B" opacity="0.75"/><circle cx="198.2" cy="371.2" r="3.1" fill="#8FCB6B" opacity="0.60"/><circle cx="167.7" cy="399.6" r="3.6" fill="#34D399" opacity="0.59"/><circle cx="95.0" cy="337.6" r="2.6" fill="#34D399" opacity="0.60"/><circle cx="150.2" cy="293.8" r="2.2" fill="#8FCB6B" opacity="0.62"/><circle cx="109.0" cy="379.9" r="2.2" fill="#34D399" opacity="0.61"/><circle cx="123.1" cy="337.7" r="2.6" fill="#8FCB6B" opacity="0.80"/><circle cx="159.3" cy="322.1" r="2.2" fill="#EE8A50" opacity="0.81"/><circle cx="157.4" cy="374.7" r="2.2" fill="#F2A03D" opacity="0.78"/><circle cx="184.6" cy="301.7" r="2.6" fill="#8FCB6B" opacity="0.59"/><circle cx="144.3" cy="377.6" r="2.6" fill="#8FCB6B" opacity="0.77"/><circle cx="130.5" cy="393.8" r="2.6" fill="#34D399" opacity="0.63"/><circle cx="163.8" cy="341.2" r="2.6" fill="#F2A03D" opacity="0.89"/><circle cx="177.0" cy="375.8" r="2.2" fill="#F2A03D" opacity="0.71"/><circle cx="150.2" cy="360.5" r="2.2" fill="#EE8A50" opacity="0.89"/><circle cx="155.7" cy="315.6" r="3.1" fill="#F2A03D" opacity="0.78"/><circle cx="180.6" cy="390.0" r="4.4" fill="#34D399" opacity="0.61"/><circle cx="176.3" cy="357.3" r="2.6" fill="#F2A03D" opacity="0.79"/><circle cx="156.0" cy="396.9" r="5.4" fill="#34D399" opacity="0.63"/><circle cx="153.7" cy="346.9" r="2.6" fill="#E06A5A" opacity="0.97"/><circle cx="135.1" cy="325.0" r="2.2" fill="#EE8A50" opacity="0.81"/><circle cx="119.6" cy="329.8" r="2.6" fill="#8FCB6B" opacity="0.75"/><circle cx="116.4" cy="305.3" r="4.4" fill="#34D399" opacity="0.62"/><circle cx="169.0" cy="374.5" r="2.6" fill="#8FCB6B" opacity="0.75"/><circle cx="141.5" cy="292.6" r="2.2" fill="#8FCB6B" opacity="0.61"/><circle cx="125.9" cy="345.5" r="2.2" fill="#F2A03D" opacity="0.82"/><circle cx="163.4" cy="312.5" r="2.2" fill="#34D399" opacity="0.74"/><circle cx="112.5" cy="321.6" r="2.6" fill="#34D399" opacity="0.68"/><circle cx="199.4" cy="320.9" r="3.1" fill="#34D399" opacity="0.60"/><circle cx="168.4" cy="334.5" r="2.2" fill="#8FCB6B" opacity="0.84"/><circle cx="132.9" cy="339.1" r="3.1" fill="#F2A03D" opacity="0.87"/><circle cx="190.8" cy="356.0" r="2.6" fill="#8FCB6B" opacity="0.69"/><circle cx="149.5" cy="381.2" r="2.2" fill="#34D399" opacity="0.74"/><circle cx="276" cy="328" r="50" fill="rgba(213,197,232,.05)"/><circle cx="250.0" cy="342.6" r="5.4" fill="#F2A03D" opacity="0.73"/><circle cx="262.7" cy="320.7" r="3.1" fill="#EE8A50" opacity="0.86"/><circle cx="307.0" cy="315.7" r="2.6" fill="#F2A03D" opacity="0.70"/><circle cx="303.9" cy="357.4" r="5.4" fill="#34D399" opacity="0.64"/><circle cx="251.5" cy="297.6" r="2.6" fill="#34D399" opacity="0.65"/><circle cx="250.0" cy="361.6" r="3.1" fill="#8FCB6B" opacity="0.62"/><circle cx="258.5" cy="285.9" r="5.4" fill="#34D399" opacity="0.59"/><circle cx="315.1" cy="328.7" r="3.1" fill="#34D399" opacity="0.65"/><circle cx="261.3" cy="306.5" r="3.6" fill="#8FCB6B" opacity="0.77"/><circle cx="290.5" cy="348.7" r="3.6" fill="#8FCB6B" opacity="0.77"/><circle cx="296.6" cy="289.4" r="2.6" fill="#34D399" opacity="0.61"/><circle cx="266.4" cy="362.6" r="4.4" fill="#34D399" opacity="0.68"/><circle cx="280.2" cy="368.3" r="3.1" fill="#8FCB6B" opacity="0.64"/><circle cx="283.5" cy="356.2" r="4.4" fill="#8FCB6B" opacity="0.74"/><circle cx="303.8" cy="305.3" r="2.2" fill="#8FCB6B" opacity="0.68"/><circle cx="313.1" cy="348.0" r="5.4" fill="#8FCB6B" opacity="0.62"/><circle cx="253.7" cy="320.0" r="2.6" fill="#F2A03D" opacity="0.79"/><circle cx="242.1" cy="333.7" r="3.6" fill="#8FCB6B" opacity="0.69"/><circle cx="265.2" cy="297.4" r="3.1" fill="#F2A03D" opacity="0.71"/><circle cx="272.0" cy="322.7" r="2.2" fill="#E06A5A" opacity="0.94"/><circle cx="300.0" cy="317.0" r="2.2" fill="#F2A03D" opacity="0.76"/><circle cx="232.6" cy="332.2" r="2.2" fill="#34D399" opacity="0.61"/><circle cx="261.9" cy="336.3" r="2.2" fill="#EE8A50" opacity="0.85"/><circle cx="234.6" cy="313.3" r="3.1" fill="#34D399" opacity="0.60"/><circle cx="264.4" cy="345.2" r="3.6" fill="#EE8A50" opacity="0.81"/><circle cx="300.0" cy="335.7" r="2.2" fill="#8FCB6B" opacity="0.77"/><circle cx="289.9" cy="299.5" r="5.4" fill="#34D399" opacity="0.71"/><circle cx="279.7" cy="336.0" r="2.2" fill="#E06A5A" opacity="0.92"/><circle cx="288.8" cy="313.0" r="2.2" fill="#F2A03D" opacity="0.82"/><circle cx="307.5" cy="324.2" r="3.6" fill="#8FCB6B" opacity="0.71"/><circle cx="303.4" cy="343.7" r="2.6" fill="#8FCB6B" opacity="0.72"/><circle cx="279.2" cy="343.7" r="3.6" fill="#8FCB6B" opacity="0.86"/><circle cx="271.7" cy="305.8" r="5.4" fill="#F2A03D" opacity="0.80"/><circle cx="285.0" cy="322.5" r="2.2" fill="#F2A03D" opacity="0.90"/><circle cx="239.6" cy="301.5" r="3.6" fill="#34D399" opacity="0.59"/><circle cx="245.3" cy="354.8" r="3.1" fill="#8FCB6B" opacity="0.63"/><circle cx="258.4" cy="359.2" r="2.2" fill="#8FCB6B" opacity="0.68"/><circle cx="298.6" cy="324.6" r="3.6" fill="#F2A03D" opacity="0.79"/><circle cx="275.6" cy="329.8" r="3.1" fill="#EE8A50" opacity="0.98"/><circle cx="243.0" cy="320.7" r="3.6" fill="#F2A03D" opacity="0.70"/><circle cx="288.2" cy="365.9" r="2.6" fill="#34D399" opacity="0.64"/><circle cx="255.3" cy="327.2" r="2.6" fill="#8FCB6B" opacity="0.81"/><circle cx="236.0" cy="348.1" r="4.4" fill="#8FCB6B" opacity="0.60"/><circle cx="317.5" cy="313.7" r="2.2" fill="#8FCB6B" opacity="0.61"/><circle cx="310.7" cy="304.2" r="2.2" fill="#34D399" opacity="0.62"/><circle cx="292.4" cy="334.0" r="2.2" fill="#EE8A50" opacity="0.84"/><circle cx="281.5" cy="310.6" r="2.6" fill="#8FCB6B" opacity="0.84"/><circle cx="273.0" cy="357.4" r="2.6" fill="#34D399" opacity="0.73"/><circle cx="281.0" cy="289.5" r="2.6" fill="#8FCB6B" opacity="0.65"/><circle cx="276.1" cy="316.6" r="2.2" fill="#E06A5A" opacity="0.90"/><circle cx="246.7" cy="307.5" r="2.2" fill="#34D399" opacity="0.68"/><circle cx="312.0" cy="335.9" r="3.1" fill="#F2A03D" opacity="0.67"/><circle cx="232.8" cy="324.8" r="3.6" fill="#34D399" opacity="0.61"/><circle cx="250.9" cy="333.0" r="2.6" fill="#8FCB6B" opacity="0.77"/><circle cx="290.1" cy="319.6" r="2.2" fill="#8FCB6B" opacity="0.85"/><circle cx="257.1" cy="365.8" r="2.6" fill="#8FCB6B" opacity="0.62"/><circle cx="249.7" cy="314.7" r="2.2" fill="#34D399" opacity="0.73"/><circle cx="318.7" cy="340.1" r="2.6" fill="#34D399" opacity="0.60"/><circle cx="272.5" cy="287.9" r="2.2" fill="#8FCB6B" opacity="0.64"/><circle cx="264.0" cy="352.8" r="2.2" fill="#F2A03D" opacity="0.75"/><circle cx="296.0" cy="363.5" r="3.1" fill="#8FCB6B" opacity="0.63"/><circle cx="267.5" cy="333.9" r="2.6" fill="#EE8A50" opacity="0.91"/><circle cx="301.1" cy="295.8" r="3.6" fill="#34D399" opacity="0.63"/><circle cx="319.1" cy="320.1" r="2.6" fill="#8FCB6B" opacity="0.61"/><circle cx="254.5" cy="353.6" r="3.1" fill="#F2A03D" opacity="0.70"/><circle cx="272.2" cy="351.0" r="2.2" fill="#F2A03D" opacity="0.79"/><circle cx="262.2" cy="371.8" r="2.6" fill="#8FCB6B" opacity="0.59"/><circle cx="294.7" cy="307.9" r="2.2" fill="#8FCB6B" opacity="0.75"/><circle cx="288.9" cy="340.6" r="3.1" fill="#F2A03D" opacity="0.84"/><circle cx="376" cy="370" r="44" fill="rgba(213,197,232,.05)"/><circle cx="403.3" cy="359.1" r="3.1" fill="#8FCB6B" opacity="0.70"/><circle cx="411.6" cy="374.0" r="2.2" fill="#34D399" opacity="0.63"/><circle cx="371.8" cy="391.9" r="4.4" fill="#F2A03D" opacity="0.77"/><circle cx="388.9" cy="378.0" r="3.1" fill="#EE8A50" opacity="0.85"/><circle cx="379.6" cy="331.7" r="4.4" fill="#8FCB6B" opacity="0.61"/><circle cx="354.0" cy="402.6" r="5.4" fill="#34D399" opacity="0.60"/><circle cx="387.3" cy="391.1" r="3.6" fill="#8FCB6B" opacity="0.75"/><circle cx="375.2" cy="352.9" r="3.1" fill="#EE8A50" opacity="0.82"/><circle cx="371.5" cy="343.1" r="5.4" fill="#8FCB6B" opacity="0.72"/><circle cx="402.7" cy="388.7" r="2.6" fill="#34D399" opacity="0.67"/><circle cx="381.8" cy="360.5" r="5.4" fill="#EE8A50" opacity="0.89"/><circle cx="384.6" cy="384.3" r="2.2" fill="#F2A03D" opacity="0.83"/><circle cx="396.7" cy="338.4" r="3.6" fill="#34D399" opacity="0.61"/><circle cx="375.1" cy="383.2" r="2.6" fill="#F2A03D" opacity="0.86"/><circle cx="346.4" cy="370.7" r="2.2" fill="#34D399" opacity="0.70"/><circle cx="376.8" cy="400.5" r="2.2" fill="#8FCB6B" opacity="0.69"/><circle cx="363.9" cy="383.4" r="3.6" fill="#F2A03D" opacity="0.82"/><circle cx="395.8" cy="349.3" r="2.2" fill="#34D399" opacity="0.71"/><circle cx="345.6" cy="355.6" r="2.6" fill="#34D399" opacity="0.66"/><circle cx="393.6" cy="402.7" r="2.2" fill="#34D399" opacity="0.62"/><circle cx="366.5" cy="360.6" r="4.4" fill="#EE8A50" opacity="0.86"/><circle cx="397.1" cy="380.4" r="2.2" fill="#34D399" opacity="0.76"/><circle cx="351.2" cy="347.2" r="3.1" fill="#34D399" opacity="0.66"/><circle cx="386.2" cy="344.5" r="2.6" fill="#34D399" opacity="0.72"/><circle cx="353.0" cy="366.3" r="2.2" fill="#F2A03D" opacity="0.76"/><circle cx="359.7" cy="339.1" r="2.6" fill="#34D399" opacity="0.64"/><circle cx="404.3" cy="370.8" r="2.2" fill="#F2A03D" opacity="0.71"/><circle cx="405.3" cy="343.2" r="3.1" fill="#34D399" opacity="0.59"/><circle cx="373.0" cy="372.4" r="2.6" fill="#F2A03D" opacity="0.96"/><circle cx="406.5" cy="382.0" r="3.6" fill="#34D399" opacity="0.67"/><circle cx="394.2" cy="386.9" r="3.1" fill="#F2A03D" opacity="0.75"/><circle cx="337.9" cy="380.1" r="3.6" fill="#34D399" opacity="0.60"/><circle cx="412.6" cy="358.9" r="2.6" fill="#34D399" opacity="0.61"/><circle cx="387.1" cy="404.6" r="2.2" fill="#34D399" opacity="0.63"/><circle cx="348.6" cy="389.1" r="3.1" fill="#34D399" opacity="0.66"/><circle cx="395.5" cy="360.4" r="2.6" fill="#8FCB6B" opacity="0.78"/><circle cx="337.6" cy="369.9" r="2.2" fill="#8FCB6B" opacity="0.61"/><circle cx="388.2" cy="350.4" r="2.2" fill="#8FCB6B" opacity="0.76"/><circle cx="365.3" cy="407.7" r="3.1" fill="#34D399" opacity="0.60"/><circle cx="381.6" cy="407.3" r="2.2" fill="#8FCB6B" opacity="0.61"/><circle cx="363.6" cy="373.6" r="2.2" fill="#EE8A50" opacity="0.87"/><circle cx="414.6" cy="367.7" r="3.1" fill="#34D399" opacity="0.60"/><circle cx="354.6" cy="357.3" r="2.2" fill="#F2A03D" opacity="0.75"/><circle cx="387.7" cy="368.0" r="2.2" fill="#F2A03D" opacity="0.88"/><circle cx="394.3" cy="370.8" r="3.1" fill="#F2A03D" opacity="0.81"/><circle cx="406.0" cy="352.6" r="2.2" fill="#8FCB6B" opacity="0.65"/><circle cx="399.9" cy="396.8" r="3.1" fill="#34D399" opacity="0.63"/><circle cx="380.0" cy="376.5" r="3.1" fill="#EE8A50" opacity="0.92"/><circle cx="370.7" cy="378.8" r="2.2" fill="#E06A5A" opacity="0.90"/><circle cx="364.1" cy="350.8" r="3.1" fill="#8FCB6B" opacity="0.77"/><circle cx="342.3" cy="362.5" r="3.1" fill="#34D399" opacity="0.65"/><circle cx="387.1" cy="334.7" r="2.2" fill="#8FCB6B" opacity="0.62"/><circle cx="367.7" cy="331.3" r="3.1" fill="#34D399" opacity="0.59"/><circle cx="353.7" cy="373.3" r="2.2" fill="#8FCB6B" opacity="0.77"/><circle cx="347.0" cy="377.8" r="2.2" fill="#34D399" opacity="0.69"/><circle cx="361.0" cy="389.9" r="2.2" fill="#F2A03D" opacity="0.75"/><circle cx="383.9" cy="397.7" r="2.2" fill="#F2A03D" opacity="0.71"/><circle cx="364.2" cy="398.2" r="2.2" fill="#8FCB6B" opacity="0.69"/><circle cx="373.8" cy="405.5" r="2.2" fill="#34D399" opacity="0.64"/><circle cx="450" cy="318" r="32" fill="rgba(213,197,232,.05)"/><circle cx="472.5" cy="330.2" r="2.2" fill="#34D399" opacity="0.64"/><circle cx="463.4" cy="312.4" r="2.6" fill="#8FCB6B" opacity="0.80"/><circle cx="424.1" cy="329.8" r="2.2" fill="#34D399" opacity="0.60"/><circle cx="439.2" cy="323.2" r="2.2" fill="#F2A03D" opacity="0.83"/><circle cx="475.2" cy="316.4" r="5.4" fill="#34D399" opacity="0.65"/><circle cx="445.0" cy="313.0" r="4.4" fill="#F2A03D" opacity="0.90"/><circle cx="445.6" cy="297.4" r="2.6" fill="#34D399" opacity="0.70"/><circle cx="453.8" cy="332.5" r="3.6" fill="#8FCB6B" opacity="0.79"/><circle cx="447.2" cy="323.0" r="3.6" fill="#EE8A50" opacity="0.92"/><circle cx="462.1" cy="322.5" r="3.1" fill="#EE8A50" opacity="0.82"/><circle cx="431.8" cy="310.4" r="2.2" fill="#F2A03D" opacity="0.72"/><circle cx="458.0" cy="299.4" r="2.6" fill="#34D399" opacity="0.72"/><circle cx="455.2" cy="346.2" r="3.1" fill="#34D399" opacity="0.60"/><circle cx="455.5" cy="307.5" r="3.6" fill="#F2A03D" opacity="0.83"/><circle cx="441.9" cy="305.4" r="2.2" fill="#8FCB6B" opacity="0.79"/><circle cx="432.4" cy="334.4" r="2.2" fill="#34D399" opacity="0.66"/><circle cx="433.0" cy="319.8" r="3.1" fill="#8FCB6B" opacity="0.76"/><circle cx="422.7" cy="307.4" r="4.4" fill="#34D399" opacity="0.59"/><circle cx="425.0" cy="320.8" r="2.6" fill="#8FCB6B" opacity="0.65"/><circle cx="466.1" cy="299.1" r="3.6" fill="#34D399" opacity="0.65"/><circle cx="450.7" cy="292.8" r="2.6" fill="#8FCB6B" opacity="0.65"/><circle cx="436.7" cy="341.7" r="2.2" fill="#34D399" opacity="0.62"/><circle cx="436.0" cy="329.6" r="2.2" fill="#F2A03D" opacity="0.75"/><circle cx="453.1" cy="316.1" r="2.6" fill="#F2A03D" opacity="0.95"/><circle cx="431.6" cy="299.6" r="3.6" fill="#34D399" opacity="0.63"/><circle cx="442.1" cy="332.9" r="2.2" fill="#8FCB6B" opacity="0.76"/><circle cx="448.7" cy="341.6" r="2.6" fill="#8FCB6B" opacity="0.67"/><circle cx="465.9" cy="339.2" r="2.2" fill="#34D399" opacity="0.63"/><circle cx="462.2" cy="331.2" r="2.2" fill="#34D399" opacity="0.75"/><circle cx="474.2" cy="303.1" r="2.6" fill="#34D399" opacity="0.60"/><circle cx="450.5" cy="302.5" r="2.2" fill="#F2A03D" opacity="0.78"/><circle cx="443.4" cy="346.2" r="3.1" fill="#34D399" opacity="0.59"/><circle cx="438.7" cy="292.6" r="2.6" fill="#34D399" opacity="0.61"/><circle cx="467.6" cy="327.1" r="2.2" fill="#F2A03D" opacity="0.72"/><circle cx="461.7" cy="291.2" r="3.6" fill="#34D399" opacity="0.59"/><circle cx="459.5" cy="340.9" r="2.2" fill="#34D399" opacity="0.65"/><circle cx="420.8" cy="315.7" r="2.6" fill="#34D399" opacity="0.59"/><circle cx="469.2" cy="308.4" r="3.1" fill="#F2A03D" opacity="0.70"/><circle cx="208" cy="406" r="30" fill="rgba(213,197,232,.05)"/><circle cx="215.5" cy="430.4" r="3.6" fill="#34D399" opacity="0.62"/><circle cx="194.3" cy="402.3" r="2.2" fill="#EE8A50" opacity="0.79"/><circle cx="230.9" cy="399.5" r="4.4" fill="#34D399" opacity="0.64"/><circle cx="206.4" cy="403.2" r="5.4" fill="#E06A5A" opacity="0.95"/><circle cx="226.9" cy="421.4" r="3.6" fill="#8FCB6B" opacity="0.63"/><circle cx="220.2" cy="385.5" r="3.1" fill="#8FCB6B" opacity="0.64"/><circle cx="194.7" cy="393.1" r="4.4" fill="#34D399" opacity="0.72"/><circle cx="187.7" cy="410.6" r="4.4" fill="#8FCB6B" opacity="0.69"/><circle cx="201.9" cy="431.3" r="2.2" fill="#34D399" opacity="0.61"/><circle cx="188.1" cy="422.5" r="4.4" fill="#8FCB6B" opacity="0.61"/><circle cx="219.4" cy="416.8" r="2.2" fill="#F2A03D" opacity="0.76"/><circle cx="200.7" cy="420.4" r="2.2" fill="#8FCB6B" opacity="0.76"/><circle cx="212.8" cy="420.4" r="2.2" fill="#F2A03D" opacity="0.77"/><circle cx="201.7" cy="411.5" r="2.2" fill="#F2A03D" opacity="0.87"/><circle cx="218.4" cy="401.3" r="3.1" fill="#EE8A50" opacity="0.83"/><circle cx="206.6" cy="386.7" r="2.2" fill="#8FCB6B" opacity="0.71"/><circle cx="211.4" cy="394.7" r="2.6" fill="#EE8A50" opacity="0.82"/><circle cx="182.8" cy="402.1" r="2.6" fill="#34D399" opacity="0.62"/><circle cx="212.6" cy="412.1" r="3.1" fill="#EE8A50" opacity="0.89"/><circle cx="232.5" cy="416.4" r="2.2" fill="#34D399" opacity="0.60"/><circle cx="221.5" cy="393.6" r="3.1" fill="#8FCB6B" opacity="0.73"/><circle cx="199.4" cy="384.2" r="2.6" fill="#34D399" opacity="0.65"/><circle cx="208.0" cy="428.0" r="2.6" fill="#8FCB6B" opacity="0.67"/><circle cx="211.8" cy="379.2" r="4.4" fill="#34D399" opacity="0.59"/><circle cx="206.9" cy="420.1" r="2.2" fill="#8FCB6B" opacity="0.79"/><circle cx="229.8" cy="390.6" r="2.6" fill="#34D399" opacity="0.60"/><circle cx="195.7" cy="415.7" r="2.6" fill="#8FCB6B" opacity="0.77"/><circle cx="228.3" cy="408.5" r="3.1" fill="#F2A03D" opacity="0.69"/><circle cx="189.0" cy="387.5" r="2.2" fill="#34D399" opacity="0.60"/></g><text x="70" y="424" class="cap">2,802 files scored</text><text x="526" y="280" class="cap">OPEN FINDINGS</text><rect x="526" y="300" width="236" height="24" rx="6" fill="#110D17"/><rect x="526" y="300" width="3" height="24" rx="1.5" fill="#E06A5A"/><text x="539" y="316" class="find">CHANGE ENTROPY</text><text x="752" y="316" class="find" fill="#E06A5A" text-anchor="end" font-weight="700">−3.0</text><rect x="526" y="328" width="236" height="24" rx="6" fill="#110D17"/><rect x="526" y="328" width="3" height="24" rx="1.5" fill="#E06A5A"/><text x="539" y="344" class="find">NESTED COMPLEXITY</text><text x="752" y="344" class="find" fill="#E06A5A" text-anchor="end" font-weight="700">−2.4</text><rect x="526" y="356" width="236" height="24" rx="6" fill="#110D17"/><rect x="526" y="356" width="3" height="24" rx="1.5" fill="#F2A03D"/><text x="539" y="372" class="find">CHURN RISK</text><text x="752" y="372" class="find" fill="#F2A03D" text-anchor="end" font-weight="700">−1.8</text><rect x="526" y="384" width="236" height="24" rx="6" fill="#110D17"/><rect x="526" y="384" width="3" height="24" rx="1.5" fill="#F2A03D"/><text x="539" y="400" class="find">CO-CHANGE SCATTER</text><text x="752" y="400" class="find" fill="#F2A03D" text-anchor="end" font-weight="700">−1.5</text><rect x="526" y="412" width="236" height="24" rx="6" fill="#110D17"/><rect x="526" y="412" width="3" height="24" rx="1.5" fill="#F2A03D"/><text x="539" y="428" class="find">UNTESTED HOTSPOT</text><text x="752" y="428" class="find" fill="#F2A03D" text-anchor="end" font-weight="700">−1.2</text><rect x="802" y="104" width="380" height="352" rx="14" fill="#211B29" stroke="rgba(213,197,232,.10)" stroke-width="1"/><text x="824" y="135" class="lbl">DEPENDENCY GRAPH</text><text x="1160" y="135" class="kick" text-anchor="end">16 languages</text><rect x="820" y="150" width="344" height="240" rx="9" fill="#110D17"/><clipPath id="gclip"><rect x="820" y="150" width="344" height="240" rx="9"/></clipPath><g clip-path="url(#gclip)"><line x1="944.0" y1="201.8" x2="912" y2="224" stroke="rgba(213,197,232,.15)" stroke-width="1"/><line x1="907.3" y1="185.9" x2="912" y2="224" stroke="rgba(213,197,232,.15)" stroke-width="1"/><line x1="913.1" y1="193.3" x2="912" y2="224" stroke="rgba(213,197,232,.15)" stroke-width="1"/><line x1="948.1" y1="258.8" x2="912" y2="224" stroke="rgba(213,197,232,.15)" stroke-width="1"/><line x1="901.9" y1="182.8" x2="912" y2="224" stroke="rgba(213,197,232,.15)" stroke-width="1"/><line x1="895.4" y1="193.7" x2="912" y2="224" stroke="rgba(213,197,232,.15)" stroke-width="1"/><line x1="888.8" y1="197.7" x2="912" y2="224" stroke="rgba(213,197,232,.15)" stroke-width="1"/><line x1="929.0" y1="238.4" x2="912" y2="224" stroke="rgba(213,197,232,.15)" stroke-width="1"/><line x1="951.9" y1="214.2" x2="912" y2="224" stroke="rgba(213,197,232,.15)" stroke-width="1"/><line x1="1040.2" y1="228.2" x2="1052" y2="212" stroke="rgba(213,197,232,.15)" stroke-width="1"/><line x1="1048.4" y1="190.3" x2="1052" y2="212" stroke="rgba(213,197,232,.15)" stroke-width="1"/><line x1="1068.5" y1="249.3" x2="1052" y2="212" stroke="rgba(213,197,232,.15)" stroke-width="1"/><line x1="1013.8" y1="237.1" x2="1052" y2="212" stroke="rgba(213,197,232,.15)" stroke-width="1"/><line x1="1009.6" y1="237.8" x2="1052" y2="212" stroke="rgba(213,197,232,.15)" stroke-width="1"/><line x1="1065.3" y1="185.2" x2="1052" y2="212" stroke="rgba(213,197,232,.15)" stroke-width="1"/><line x1="1017.4" y1="194.6" x2="1052" y2="212" stroke="rgba(213,197,232,.15)" stroke-width="1"/><line x1="1107.4" y1="209.0" x2="1052" y2="212" stroke="rgba(213,197,232,.15)" stroke-width="1"/><line x1="1061.0" y1="174.5" x2="1052" y2="212" stroke="rgba(213,197,232,.15)" stroke-width="1"/><line x1="944.8" y1="323.3" x2="982" y2="322" stroke="rgba(213,197,232,.15)" stroke-width="1"/><line x1="955.1" y1="338.1" x2="982" y2="322" stroke="rgba(213,197,232,.15)" stroke-width="1"/><line x1="1001.7" y1="341.5" x2="982" y2="322" stroke="rgba(213,197,232,.15)" stroke-width="1"/><line x1="1022.9" y1="322.0" x2="982" y2="322" stroke="rgba(213,197,232,.15)" stroke-width="1"/><line x1="961.2" y1="356.5" x2="982" y2="322" stroke="rgba(213,197,232,.15)" stroke-width="1"/><line x1="942.1" y1="301.8" x2="982" y2="322" stroke="rgba(213,197,232,.15)" stroke-width="1"/><line x1="943.8" y1="346.7" x2="982" y2="322" stroke="rgba(213,197,232,.15)" stroke-width="1"/><line x1="983.8" y1="296.5" x2="982" y2="322" stroke="rgba(213,197,232,.15)" stroke-width="1"/><line x1="949.1" y1="320.7" x2="982" y2="322" stroke="rgba(213,197,232,.15)" stroke-width="1"/><line x1="1131.2" y1="323.1" x2="1098" y2="310" stroke="rgba(213,197,232,.15)" stroke-width="1"/><line x1="1085.5" y1="327.7" x2="1098" y2="310" stroke="rgba(213,197,232,.15)" stroke-width="1"/><line x1="1106.7" y1="286.0" x2="1098" y2="310" stroke="rgba(213,197,232,.15)" stroke-width="1"/><line x1="1152.2" y1="302.0" x2="1098" y2="310" stroke="rgba(213,197,232,.15)" stroke-width="1"/><line x1="1068.9" y1="335.5" x2="1098" y2="310" stroke="rgba(213,197,232,.15)" stroke-width="1"/><line x1="1148.6" y1="288.2" x2="1098" y2="310" stroke="rgba(213,197,232,.15)" stroke-width="1"/><line x1="1064.8" y1="323.3" x2="1098" y2="310" stroke="rgba(213,197,232,.15)" stroke-width="1"/><line x1="1046.2" y1="319.4" x2="1098" y2="310" stroke="rgba(213,197,232,.15)" stroke-width="1"/><line x1="1137.2" y1="290.0" x2="1098" y2="310" stroke="rgba(213,197,232,.15)" stroke-width="1"/><line x1="912" y1="224" x2="1052" y2="212" stroke="rgba(213,197,232,.30)" stroke-width="1.6"/><line x1="912" y1="224" x2="982" y2="322" stroke="rgba(213,197,232,.30)" stroke-width="1.6"/><line x1="912" y1="224" x2="1098" y2="310" stroke="rgba(213,197,232,.30)" stroke-width="1.6"/><line x1="1052" y1="212" x2="982" y2="322" stroke="rgba(213,197,232,.30)" stroke-width="1.6"/><line x1="982" y1="322" x2="1098" y2="310" stroke="rgba(213,197,232,.30)" stroke-width="1.6"/><circle cx="912.0" cy="224.0" r="9.0" fill="#F59520" stroke="#211B29" stroke-width="1.5"/><circle cx="944.0" cy="201.8" r="4.0" fill="#C97A1A"/><circle cx="907.3" cy="185.9" r="3.3" fill="#C97A1A"/><circle cx="913.1" cy="193.3" r="2.8" fill="#C97A1A"/><circle cx="948.1" cy="258.8" r="3.8" fill="#C97A1A"/><circle cx="901.9" cy="182.8" r="3.1" fill="#C97A1A"/><circle cx="895.4" cy="193.7" r="3.4" fill="#C97A1A"/><circle cx="888.8" cy="197.7" r="3.3" fill="#C97A1A"/><circle cx="929.0" cy="238.4" r="3.8" fill="#C97A1A"/><circle cx="951.9" cy="214.2" r="3.6" fill="#C97A1A"/><circle cx="1052.0" cy="212.0" r="9.0" fill="#A98FC4" stroke="#211B29" stroke-width="1.5"/><circle cx="1040.2" cy="228.2" r="3.4" fill="#7D659C"/><circle cx="1048.4" cy="190.3" r="3.5" fill="#7D659C"/><circle cx="1068.5" cy="249.3" r="3.5" fill="#7D659C"/><circle cx="1013.8" cy="237.1" r="3.7" fill="#7D659C"/><circle cx="1009.6" cy="237.8" r="3.8" fill="#7D659C"/><circle cx="1065.3" cy="185.2" r="4.1" fill="#7D659C"/><circle cx="1017.4" cy="194.6" r="4.2" fill="#7D659C"/><circle cx="1107.4" cy="209.0" r="3.7" fill="#7D659C"/><circle cx="1061.0" cy="174.5" r="3.3" fill="#7D659C"/><circle cx="982.0" cy="322.0" r="9.0" fill="#E06A5A" stroke="#211B29" stroke-width="1.5"/><circle cx="944.8" cy="323.3" r="4.2" fill="#B04C3E"/><circle cx="955.1" cy="338.1" r="4.0" fill="#B04C3E"/><circle cx="1001.7" cy="341.5" r="3.2" fill="#B04C3E"/><circle cx="1022.9" cy="322.0" r="3.0" fill="#B04C3E"/><circle cx="961.2" cy="356.5" r="4.0" fill="#B04C3E"/><circle cx="942.1" cy="301.8" r="3.5" fill="#B04C3E"/><circle cx="943.8" cy="346.7" r="3.7" fill="#B04C3E"/><circle cx="983.8" cy="296.5" r="2.9" fill="#B04C3E"/><circle cx="949.1" cy="320.7" r="2.6" fill="#B04C3E"/><circle cx="1098.0" cy="310.0" r="9.0" fill="#A9BB6F" stroke="#211B29" stroke-width="1.5"/><circle cx="1131.2" cy="323.1" r="3.2" fill="#7E8F4E"/><circle cx="1085.5" cy="327.7" r="4.3" fill="#7E8F4E"/><circle cx="1106.7" cy="286.0" r="2.9" fill="#7E8F4E"/><circle cx="1152.2" cy="302.0" r="3.9" fill="#7E8F4E"/><circle cx="1068.9" cy="335.5" r="2.8" fill="#7E8F4E"/><circle cx="1148.6" cy="288.2" r="4.2" fill="#7E8F4E"/><circle cx="1064.8" cy="323.3" r="3.6" fill="#7E8F4E"/><circle cx="1046.2" cy="319.4" r="4.0" fill="#7E8F4E"/><circle cx="1137.2" cy="290.0" r="2.9" fill="#7E8F4E"/></g><text x="832" y="380" class="cap">28,126 nodes · 11 communities</text><text x="824" y="414" class="out">Who calls this? What breaks if</text><text x="824" y="434" class="out">I change it?</text><rect x="1200" y="104" width="364" height="352" rx="14" fill="#211B29" stroke="rgba(213,197,232,.10)" stroke-width="1"/><text x="1222" y="135" class="lbl">GIT HISTORY</text><text x="1542" y="135" class="kick" text-anchor="end">1,050 commits</text><rect x="1218" y="150" width="328" height="240" rx="9" fill="#110D17"/><rect x="1230.0" y="312.2" width="16.0" height="13.8" rx="2.5" fill="#34D399" opacity="0.82"/><rect x="1249.0" y="300.7" width="16.0" height="25.3" rx="2.5" fill="#34D399" opacity="0.82"/><rect x="1268.0" y="282.3" width="16.0" height="43.7" rx="2.5" fill="#34D399" opacity="0.82"/><rect x="1287.0" y="261.6" width="16.0" height="64.4" rx="2.5" fill="#34D399" opacity="0.82"/><rect x="1306.0" y="231.7" width="16.0" height="94.3" rx="2.5" fill="#34D399" opacity="0.82"/><rect x="1325.0" y="206.4" width="16.0" height="119.6" rx="2.5" fill="#34D399" opacity="0.82"/><rect x="1344.0" y="181.1" width="16.0" height="144.9" rx="2.5" fill="#F2A03D" opacity="0.82"/><rect x="1363.0" y="192.6" width="16.0" height="133.4" rx="2.5" fill="#F2A03D" opacity="0.82"/><rect x="1382.0" y="217.9" width="16.0" height="108.1" rx="2.5" fill="#F2A03D" opacity="0.82"/><rect x="1401.0" y="243.2" width="16.0" height="82.8" rx="2.5" fill="#F2A03D" opacity="0.82"/><rect x="1420.0" y="263.9" width="16.0" height="62.1" rx="2.5" fill="#F2A03D" opacity="0.82"/><rect x="1439.0" y="282.3" width="16.0" height="43.7" rx="2.5" fill="#E06A5A" opacity="0.82"/><rect x="1458.0" y="296.1" width="16.0" height="29.9" rx="2.5" fill="#E06A5A" opacity="0.82"/><rect x="1477.0" y="305.3" width="16.0" height="20.7" rx="2.5" fill="#E06A5A" opacity="0.82"/><rect x="1496.0" y="312.2" width="16.0" height="13.8" rx="2.5" fill="#E06A5A" opacity="0.82"/><rect x="1515.0" y="316.8" width="16.0" height="9.2" rx="2.5" fill="#E06A5A" opacity="0.82"/><line x1="1230" y1="327" x2="1534" y2="327" stroke="rgba(213,197,232,.20)"/><text x="1230" y="348" class="cap">low risk</text><text x="1534" y="348" class="cap" text-anchor="end">high risk</text><text x="1230" y="176" class="cap">CHANGE-RISK DISTRIBUTION</text><text x="1230" y="380" class="cap">hotspots · ownership · co-change · bus factor</text><text x="1222" y="414" class="out">Which of these 40 files is</text><text x="1222" y="434" class="out">actually dangerous?</text><rect x="36" y="474" width="496" height="302" rx="14" fill="#211B29" stroke="rgba(213,197,232,.10)" stroke-width="1"/><text x="58" y="505" class="lbl">GENERATED DOCS</text><text x="510" y="505" class="kick" text-anchor="end">1,274 pages</text><rect x="54" y="520" width="460" height="176" rx="9" fill="#110D17"/><rect x="70" y="536" width="265" height="11" rx="3" fill="#A98FC4" opacity=".85"/><rect x="70" y="562" width="394" height="6" rx="3" fill="rgba(213,197,232,.24)"/><rect x="70" y="578" width="368" height="6" rx="3" fill="rgba(213,197,232,.24)"/><rect x="70" y="594" width="304" height="6" rx="3" fill="rgba(213,197,232,.24)"/><rect x="70" y="607" width="188" height="9" rx="3" fill="#F59520" opacity=".55"/><rect x="70" y="630" width="377" height="6" rx="3" fill="rgba(213,197,232,.24)"/><rect x="70" y="646" width="338" height="6" rx="3" fill="rgba(213,197,232,.24)"/><rect x="70" y="662" width="235" height="6" rx="3" fill="rgba(213,197,232,.24)"/><rect x="416" y="530" width="86" height="20" rx="10" fill="rgba(52,211,153,.14)"/><text x="459" y="544" class="pill" fill="#34D399" text-anchor="middle">fresh · v6</text><text x="58" y="730" class="out">Why does auth work this way?</text><text x="58" y="751" class="outsub">Answered from the wiki, not from 12 file reads.</text><rect x="550" y="474" width="496" height="302" rx="14" fill="#211B29" stroke="rgba(213,197,232,.10)" stroke-width="1"/><text x="572" y="505" class="lbl">ARCHITECTURAL DECISIONS</text><text x="1024" y="505" class="kick" text-anchor="end">mined from 8 sources</text><rect x="568" y="520" width="460" height="176" rx="9" fill="#110D17"/><text x="584" y="548" class="dec">Rebuild the KG skeleton only when the</text><text x="584" y="568" class="dec">graph fingerprint changes</text><rect x="584" y="582" width="392" height="1" fill="rgba(213,197,232,.15)"/><rect x="584" y="596" width="101" height="21" rx="10.5" fill="#211B29" stroke="rgba(213,197,232,.18)"/><text x="635" y="610" class="pill" fill="#34D399" text-anchor="middle">Verified quote</text><rect x="693.4" y="596" width="108" height="21" rx="10.5" fill="#211B29" stroke="rgba(213,197,232,.18)"/><text x="747" y="610" class="pill" fill="#A98FC4" text-anchor="middle">supersedes #418</text><rect x="809.4" y="596" width="115" height="21" rx="10.5" fill="#211B29" stroke="rgba(213,197,232,.18)"/><text x="867" y="610" class="pill" fill="#F59520" text-anchor="middle">governs 24 files</text><text x="584" y="660" class="cap">Delivered back at session start, and the moment</text><text x="584" y="676" class="cap">your agent edits a file the decision governs.</text><text x="572" y="730" class="out">Don’t re-litigate what we settled.</text><text x="572" y="751" class="outsub">Captured nowhere else.</text><rect x="1064" y="474" width="500" height="302" rx="14" fill="#211B29" stroke="rgba(213,197,232,.10)" stroke-width="1"/><text x="1086" y="505" class="lbl">SERVED TO YOUR AGENT</text><text x="1542" y="505" class="kick" text-anchor="end">10 MCP tools</text><rect x="1084" y="528" width="95" height="24" rx="6" fill="#110D17" stroke="rgba(213,197,232,.10)"/><text x="1131" y="544" class="mono" text-anchor="middle">get_overview</text><rect x="1187" y="528" width="81" height="24" rx="6" fill="#110D17" stroke="rgba(213,197,232,.10)"/><text x="1227" y="544" class="mono" text-anchor="middle">get_answer</text><rect x="1276" y="528" width="88" height="24" rx="6" fill="#110D17" stroke="rgba(213,197,232,.10)"/><text x="1320" y="544" class="mono" text-anchor="middle">get_context</text><rect x="1372" y="528" width="81" height="24" rx="6" fill="#110D17" stroke="rgba(213,197,232,.10)"/><text x="1412" y="544" class="mono" text-anchor="middle">get_symbol</text><rect x="1084" y="558" width="116" height="24" rx="6" fill="#110D17" stroke="rgba(213,197,232,.10)"/><text x="1142" y="574" class="mono" text-anchor="middle">search_codebase</text><rect x="1208" y="558" width="67" height="24" rx="6" fill="#110D17" stroke="rgba(213,197,232,.10)"/><text x="1241" y="574" class="mono" text-anchor="middle">get_risk</text><rect x="1283" y="558" width="116" height="24" rx="6" fill="#110D17" stroke="rgba(213,197,232,.10)"/><text x="1340" y="574" class="mono" text-anchor="middle">get_change_risk</text><rect x="1406" y="558" width="60" height="24" rx="6" fill="#110D17" stroke="rgba(213,197,232,.10)"/><text x="1436" y="574" class="mono" text-anchor="middle">get_why</text><rect x="1084" y="588" width="102" height="24" rx="6" fill="#110D17" stroke="rgba(213,197,232,.10)"/><text x="1135" y="604" class="mono" text-anchor="middle">get_dead_code</text><rect x="1194" y="588" width="81" height="24" rx="6" fill="#110D17" stroke="rgba(213,197,232,.10)"/><text x="1234" y="604" class="mono" text-anchor="middle">get_health</text><text x="1084" y="636" class="cap">Every response carries index age, indexed commit, and a staleness warning.</text><rect x="1084" y="650" width="460" height="46" rx="9" fill="rgba(245,149,32,.12)" stroke="rgba(245,149,32,.32)"/><text x="1098" y="670" class="chipval" fill="#F59520" font-size="19">−96%<tspan class="chipunit" fill="#A79DB3"> context tokens</tspan></text><text x="1098" y="688" class="cap">−70% tool calls · −89% file reads · fewer round-trips per answer</text><text x="1086" y="730" class="out">Claude Code, Codex, Cursor, VS Code.</text><text x="1086" y="751" class="outsub">Plus hooks, so it arrives unasked.</text></svg>\ No newline at end of filediff --git a/.github/assets/one-index.svg b/.github/assets/one-index.svgindex ac3c4204c..cfdef4fda 100644--- a/.github/assets/one-index.svg+++ b/.github/assets/one-index.svg@@ -15,4 +15,4 @@.dec { font-size: 14px; font-weight: 600; fill: #241B2C; }.pill { font-size: 10.5px; font-weight: 600; }.mono { font-size: 11.5px; fill: #5E5360; font-family: ui-monospace, SFMono-Regular, "SF Mono", Menlo, Consolas, monospace; }-</style><rect width="1600" height="812" fill="#FBF6F1"/><text x="36" y="50" class="h1">One index. Everything your agent and your team keep asking for.</text><text x="36" y="80" class="sub">Index once, in seconds, no LLM required. Every answer below is already computed.</text><rect x="36" y="104" width="748" height="352" rx="14" fill="#FFFFFF" stroke="rgba(88,67,108,.12)" stroke-width="1"/><text x="58" y="135" class="lbl">CODE HEALTH</text><text x="762" y="135" class="kick" text-anchor="end">25 markers · zero LLM · <30s</text><rect x="58" y="150" width="146" height="58" rx="9" fill="#F4EAE1" stroke="rgba(88,67,108,.12)"/><text x="70" y="170" class="chiplbl">Defect risk</text><text x="70" y="197" class="chipval" fill="#9A6614">7.5<tspan class="chipunit">/10</tspan></text><rect x="214" y="150" width="146" height="58" rx="9" fill="#F4EAE1" stroke="rgba(88,67,108,.12)"/><text x="226" y="170" class="chiplbl">Maintainability</text><text x="226" y="197" class="chipval" fill="#1D8155">8.6<tspan class="chipunit">/10</tspan></text><rect x="370" y="150" width="146" height="58" rx="9" fill="#F4EAE1" stroke="rgba(88,67,108,.12)"/><text x="382" y="170" class="chiplbl">Perf risks</text><text x="382" y="197" class="chipval" fill="#58436C">268<tspan class="chipunit"></tspan></text><rect x="58" y="220" width="704" height="34" rx="8" fill="rgba(29,129,85,.09)" stroke="rgba(29,129,85,.22)"/><circle cx="76" cy="237" r="4.5" fill="#1D8155"/><text x="90" y="242" class="trust">17 of the 20 lowest-health files were bug-fixed in the last 6 months, <tspan font-weight="700">3.63× the 23% baseline</tspan></text><rect x="58" y="266" width="452" height="170" rx="9" fill="#F4EAE1"/><clipPath id="galaxy"><rect x="58" y="266" width="452" height="170" rx="9"/></clipPath><g clip-path="url(#galaxy)"><circle cx="150" cy="346" r="62" fill="rgba(88,67,108,.05)"/><circle cx="140.1" cy="365.8" r="5.4" fill="#6B9A3D" opacity="0.84"/><circle cx="157.6" cy="330.2" r="4.4" fill="#9A6614" opacity="0.87"/><circle cx="139.0" cy="345.5" r="3.6" fill="#C4762A" opacity="0.92"/><circle cx="152.5" cy="388.3" r="2.2" fill="#9A6614" opacity="0.69"/><circle cx="169.2" cy="364.9" r="5.4" fill="#9A6614" opacity="0.80"/><circle cx="190.4" cy="362.5" r="2.2" fill="#9A6614" opacity="0.68"/><circle cx="200.6" cy="361.2" r="3.1" fill="#1D8155" opacity="0.62"/><circle cx="108.3" cy="335.1" r="4.4" fill="#6B9A3D" opacity="0.69"/><circle cx="168.3" cy="385.4" r="5.4" fill="#1D8155" opacity="0.68"/><circle cx="189.4" cy="373.7" r="4.4" fill="#1D8155" opacity="0.65"/><circle cx="156.6" cy="307.6" r="3.6" fill="#6B9A3D" opacity="0.72"/><circle cx="150.2" cy="370.2" r="2.6" fill="#6B9A3D" opacity="0.82"/><circle cx="137.5" cy="384.2" r="3.1" fill="#6B9A3D" opacity="0.71"/><circle cx="136.7" cy="400.9" r="2.2" fill="#1D8155" opacity="0.59"/><circle cx="100.7" cy="372.4" r="2.2" fill="#1D8155" opacity="0.59"/><circle cx="102.1" cy="322.4" r="3.1" fill="#1D8155" opacity="0.61"/><circle cx="126.3" cy="378.5" r="3.6" fill="#1D8155" opacity="0.71"/><circle cx="142.8" cy="333.9" r="5.4" fill="#9A6614" opacity="0.90"/><circle cx="118.3" cy="391.4" r="3.1" fill="#1D8155" opacity="0.60"/><circle cx="160.3" cy="355.3" r="3.1" fill="#9A6614" opacity="0.90"/><circle cx="169.8" cy="323.3" r="3.6" fill="#C4762A" opacity="0.78"/><circle cx="135.6" cy="313.9" r="2.6" fill="#6B9A3D" opacity="0.74"/><circle cx="201.9" cy="349.9" r="2.6" fill="#1D8155" opacity="0.62"/><circle cx="186.9" cy="346.9" r="3.1" fill="#6B9A3D" opacity="0.73"/><circle cx="193.9" cy="331.8" r="5.4" fill="#1D8155" opacity="0.66"/><circle cx="190.7" cy="316.3" r="5.4" fill="#6B9A3D" opacity="0.63"/><circle cx="121.9" cy="368.5" r="2.2" fill="#6B9A3D" opacity="0.74"/><circle cx="129.8" cy="360.6" r="2.6" fill="#9A6614" opacity="0.82"/><circle cx="172.2" cy="346.0" r="2.2" fill="#C4762A" opacity="0.84"/><circle cx="118.4" cy="349.9" r="2.6" fill="#6B9A3D" opacity="0.77"/><circle cx="119.4" cy="382.2" r="2.2" fill="#6B9A3D" opacity="0.66"/><circle cx="178.5" cy="327.5" r="2.6" fill="#6B9A3D" opacity="0.75"/><circle cx="111.6" cy="312.9" r="2.6" fill="#6B9A3D" opacity="0.63"/><circle cx="170.4" cy="301.4" r="2.6" fill="#1D8155" opacity="0.64"/><circle cx="101.3" cy="348.2" r="2.2" fill="#6B9A3D" opacity="0.65"/><circle cx="205.6" cy="341.8" r="3.1" fill="#1D8155" opacity="0.60"/><circle cx="181.3" cy="369.4" r="3.1" fill="#1D8155" opacity="0.72"/><circle cx="160.0" cy="297.6" r="3.6" fill="#6B9A3D" opacity="0.64"/><circle cx="108.4" cy="364.3" r="2.2" fill="#6B9A3D" opacity="0.67"/><circle cx="159.8" cy="363.8" r="2.6" fill="#C4762A" opacity="0.85"/><circle cx="104.9" cy="356.0" r="4.4" fill="#6B9A3D" opacity="0.66"/><circle cx="99.9" cy="331.7" r="2.2" fill="#6B9A3D" opacity="0.62"/><circle cx="130.0" cy="330.2" r="3.1" fill="#9A6614" opacity="0.82"/><circle cx="133.2" cy="295.3" r="3.1" fill="#6B9A3D" opacity="0.61"/><circle cx="128.6" cy="318.9" r="3.1" fill="#9A6614" opacity="0.75"/><circle cx="176.2" cy="340.6" r="2.2" fill="#9A6614" opacity="0.81"/><circle cx="93.4" cy="350.5" r="2.6" fill="#1D8155" opacity="0.59"/><circle cx="120.0" cy="362.2" r="2.2" fill="#9A6614" opacity="0.75"/><circle cx="120.9" cy="317.3" r="2.2" fill="#1D8155" opacity="0.70"/><circle cx="173.8" cy="332.6" r="2.2" fill="#6B9A3D" opacity="0.80"/><circle cx="177.5" cy="308.0" r="3.1" fill="#1D8155" opacity="0.66"/><circle cx="149.1" cy="353.3" r="2.2" fill="#B23A2E" opacity="0.95"/><circle cx="143.4" cy="396.7" r="3.1" fill="#1D8155" opacity="0.63"/><circle cx="126.6" cy="310.9" r="3.6" fill="#9A6614" opacity="0.69"/><circle cx="147.2" cy="324.9" r="3.1" fill="#C4762A" opacity="0.85"/><circle cx="148.4" cy="317.2" r="2.6" fill="#6B9A3D" opacity="0.79"/><circle cx="140.6" cy="352.9" r="2.2" fill="#9A6614" opacity="0.92"/><circle cx="142.7" cy="306.6" r="3.1" fill="#6B9A3D" opacity="0.71"/><circle cx="150.8" cy="303.0" r="2.2" fill="#9A6614" opacity="0.69"/><circle cx="183.9" cy="337.2" r="3.6" fill="#6B9A3D" opacity="0.75"/><circle cx="198.2" cy="371.2" r="3.1" fill="#6B9A3D" opacity="0.60"/><circle cx="167.7" cy="399.6" r="3.6" fill="#1D8155" opacity="0.59"/><circle cx="95.0" cy="337.6" r="2.6" fill="#1D8155" opacity="0.60"/><circle cx="150.2" cy="293.8" r="2.2" fill="#6B9A3D" opacity="0.62"/><circle cx="109.0" cy="379.9" r="2.2" fill="#1D8155" opacity="0.61"/><circle cx="123.1" cy="337.7" r="2.6" fill="#6B9A3D" opacity="0.80"/><circle cx="159.3" cy="322.1" r="2.2" fill="#C4762A" opacity="0.81"/><circle cx="157.4" cy="374.7" r="2.2" fill="#9A6614" opacity="0.78"/><circle cx="184.6" cy="301.7" r="2.6" fill="#6B9A3D" opacity="0.59"/><circle cx="144.3" cy="377.6" r="2.6" fill="#6B9A3D" opacity="0.77"/><circle cx="130.5" cy="393.8" r="2.6" fill="#1D8155" opacity="0.63"/><circle cx="163.8" cy="341.2" r="2.6" fill="#9A6614" opacity="0.89"/><circle cx="177.0" cy="375.8" r="2.2" fill="#9A6614" opacity="0.71"/><circle cx="150.2" cy="360.5" r="2.2" fill="#C4762A" opacity="0.89"/><circle cx="155.7" cy="315.6" r="3.1" fill="#9A6614" opacity="0.78"/><circle cx="180.6" cy="390.0" r="4.4" fill="#1D8155" opacity="0.61"/><circle cx="176.3" cy="357.3" r="2.6" fill="#9A6614" opacity="0.79"/><circle cx="156.0" cy="396.9" r="5.4" fill="#1D8155" opacity="0.63"/><circle cx="153.7" cy="346.9" r="2.6" fill="#B23A2E" opacity="0.97"/><circle cx="135.1" cy="325.0" r="2.2" fill="#C4762A" opacity="0.81"/><circle cx="119.6" cy="329.8" r="2.6" fill="#6B9A3D" opacity="0.75"/><circle cx="116.4" cy="305.3" r="4.4" fill="#1D8155" opacity="0.62"/><circle cx="169.0" cy="374.5" r="2.6" fill="#6B9A3D" opacity="0.75"/><circle cx="141.5" cy="292.6" r="2.2" fill="#6B9A3D" opacity="0.61"/><circle cx="125.9" cy="345.5" r="2.2" fill="#9A6614" opacity="0.82"/><circle cx="163.4" cy="312.5" r="2.2" fill="#1D8155" opacity="0.74"/><circle cx="112.5" cy="321.6" r="2.6" fill="#1D8155" opacity="0.68"/><circle cx="199.4" cy="320.9" r="3.1" fill="#1D8155" opacity="0.60"/><circle cx="168.4" cy="334.5" r="2.2" fill="#6B9A3D" opacity="0.84"/><circle cx="132.9" cy="339.1" r="3.1" fill="#9A6614" opacity="0.87"/><circle cx="190.8" cy="356.0" r="2.6" fill="#6B9A3D" opacity="0.69"/><circle cx="149.5" cy="381.2" r="2.2" fill="#1D8155" opacity="0.74"/><circle cx="276" cy="328" r="50" fill="rgba(88,67,108,.05)"/><circle cx="250.0" cy="342.6" r="5.4" fill="#9A6614" opacity="0.73"/><circle cx="262.7" cy="320.7" r="3.1" fill="#C4762A" opacity="0.86"/><circle cx="307.0" cy="315.7" r="2.6" fill="#9A6614" opacity="0.70"/><circle cx="303.9" cy="357.4" r="5.4" fill="#1D8155" opacity="0.64"/><circle cx="251.5" cy="297.6" r="2.6" fill="#1D8155" opacity="0.65"/><circle cx="250.0" cy="361.6" r="3.1" fill="#6B9A3D" opacity="0.62"/><circle cx="258.5" cy="285.9" r="5.4" fill="#1D8155" opacity="0.59"/><circle cx="315.1" cy="328.7" r="3.1" fill="#1D8155" opacity="0.65"/><circle cx="261.3" cy="306.5" r="3.6" fill="#6B9A3D" opacity="0.77"/><circle cx="290.5" cy="348.7" r="3.6" fill="#6B9A3D" opacity="0.77"/><circle cx="296.6" cy="289.4" r="2.6" fill="#1D8155" opacity="0.61"/><circle cx="266.4" cy="362.6" r="4.4" fill="#1D8155" opacity="0.68"/><circle cx="280.2" cy="368.3" r="3.1" fill="#6B9A3D" opacity="0.64"/><circle cx="283.5" cy="356.2" r="4.4" fill="#6B9A3D" opacity="0.74"/><circle cx="303.8" cy="305.3" r="2.2" fill="#6B9A3D" opacity="0.68"/><circle cx="313.1" cy="348.0" r="5.4" fill="#6B9A3D" opacity="0.62"/><circle cx="253.7" cy="320.0" r="2.6" fill="#9A6614" opacity="0.79"/><circle cx="242.1" cy="333.7" r="3.6" fill="#6B9A3D" opacity="0.69"/><circle cx="265.2" cy="297.4" r="3.1" fill="#9A6614" opacity="0.71"/><circle cx="272.0" cy="322.7" r="2.2" fill="#B23A2E" opacity="0.94"/><circle cx="300.0" cy="317.0" r="2.2" fill="#9A6614" opacity="0.76"/><circle cx="232.6" cy="332.2" r="2.2" fill="#1D8155" opacity="0.61"/><circle cx="261.9" cy="336.3" r="2.2" fill="#C4762A" opacity="0.85"/><circle cx="234.6" cy="313.3" r="3.1" fill="#1D8155" opacity="0.60"/><circle cx="264.4" cy="345.2" r="3.6" fill="#C4762A" opacity="0.81"/><circle cx="300.0" cy="335.7" r="2.2" fill="#6B9A3D" opacity="0.77"/><circle cx="289.9" cy="299.5" r="5.4" fill="#1D8155" opacity="0.71"/><circle cx="279.7" cy="336.0" r="2.2" fill="#B23A2E" opacity="0.92"/><circle cx="288.8" cy="313.0" r="2.2" fill="#9A6614" opacity="0.82"/><circle cx="307.5" cy="324.2" r="3.6" fill="#6B9A3D" opacity="0.71"/><circle cx="303.4" cy="343.7" r="2.6" fill="#6B9A3D" opacity="0.72"/><circle cx="279.2" cy="343.7" r="3.6" fill="#6B9A3D" opacity="0.86"/><circle cx="271.7" cy="305.8" r="5.4" fill="#9A6614" opacity="0.80"/><circle cx="285.0" cy="322.5" r="2.2" fill="#9A6614" opacity="0.90"/><circle cx="239.6" cy="301.5" r="3.6" fill="#1D8155" opacity="0.59"/><circle cx="245.3" cy="354.8" r="3.1" fill="#6B9A3D" opacity="0.63"/><circle cx="258.4" cy="359.2" r="2.2" fill="#6B9A3D" opacity="0.68"/><circle cx="298.6" cy="324.6" r="3.6" fill="#9A6614" opacity="0.79"/><circle cx="275.6" cy="329.8" r="3.1" fill="#C4762A" opacity="0.98"/><circle cx="243.0" cy="320.7" r="3.6" fill="#9A6614" opacity="0.70"/><circle cx="288.2" cy="365.9" r="2.6" fill="#1D8155" opacity="0.64"/><circle cx="255.3" cy="327.2" r="2.6" fill="#6B9A3D" opacity="0.81"/><circle cx="236.0" cy="348.1" r="4.4" fill="#6B9A3D" opacity="0.60"/><circle cx="317.5" cy="313.7" r="2.2" fill="#6B9A3D" opacity="0.61"/><circle cx="310.7" cy="304.2" r="2.2" fill="#1D8155" opacity="0.62"/><circle cx="292.4" cy="334.0" r="2.2" fill="#C4762A" opacity="0.84"/><circle cx="281.5" cy="310.6" r="2.6" fill="#6B9A3D" opacity="0.84"/><circle cx="273.0" cy="357.4" r="2.6" fill="#1D8155" opacity="0.73"/><circle cx="281.0" cy="289.5" r="2.6" fill="#6B9A3D" opacity="0.65"/><circle cx="276.1" cy="316.6" r="2.2" fill="#B23A2E" opacity="0.90"/><circle cx="246.7" cy="307.5" r="2.2" fill="#1D8155" opacity="0.68"/><circle cx="312.0" cy="335.9" r="3.1" fill="#9A6614" opacity="0.67"/><circle cx="232.8" cy="324.8" r="3.6" fill="#1D8155" opacity="0.61"/><circle cx="250.9" cy="333.0" r="2.6" fill="#6B9A3D" opacity="0.77"/><circle cx="290.1" cy="319.6" r="2.2" fill="#6B9A3D" opacity="0.85"/><circle cx="257.1" cy="365.8" r="2.6" fill="#6B9A3D" opacity="0.62"/><circle cx="249.7" cy="314.7" r="2.2" fill="#1D8155" opacity="0.73"/><circle cx="318.7" cy="340.1" r="2.6" fill="#1D8155" opacity="0.60"/><circle cx="272.5" cy="287.9" r="2.2" fill="#6B9A3D" opacity="0.64"/><circle cx="264.0" cy="352.8" r="2.2" fill="#9A6614" opacity="0.75"/><circle cx="296.0" cy="363.5" r="3.1" fill="#6B9A3D" opacity="0.63"/><circle cx="267.5" cy="333.9" r="2.6" fill="#C4762A" opacity="0.91"/><circle cx="301.1" cy="295.8" r="3.6" fill="#1D8155" opacity="0.63"/><circle cx="319.1" cy="320.1" r="2.6" fill="#6B9A3D" opacity="0.61"/><circle cx="254.5" cy="353.6" r="3.1" fill="#9A6614" opacity="0.70"/><circle cx="272.2" cy="351.0" r="2.2" fill="#9A6614" opacity="0.79"/><circle cx="262.2" cy="371.8" r="2.6" fill="#6B9A3D" opacity="0.59"/><circle cx="294.7" cy="307.9" r="2.2" fill="#6B9A3D" opacity="0.75"/><circle cx="288.9" cy="340.6" r="3.1" fill="#9A6614" opacity="0.84"/><circle cx="376" cy="370" r="44" fill="rgba(88,67,108,.05)"/><circle cx="403.3" cy="359.1" r="3.1" fill="#6B9A3D" opacity="0.70"/><circle cx="411.6" cy="374.0" r="2.2" fill="#1D8155" opacity="0.63"/><circle cx="371.8" cy="391.9" r="4.4" fill="#9A6614" opacity="0.77"/><circle cx="388.9" cy="378.0" r="3.1" fill="#C4762A" opacity="0.85"/><circle cx="379.6" cy="331.7" r="4.4" fill="#6B9A3D" opacity="0.61"/><circle cx="354.0" cy="402.6" r="5.4" fill="#1D8155" opacity="0.60"/><circle cx="387.3" cy="391.1" r="3.6" fill="#6B9A3D" opacity="0.75"/><circle cx="375.2" cy="352.9" r="3.1" fill="#C4762A" opacity="0.82"/><circle cx="371.5" cy="343.1" r="5.4" fill="#6B9A3D" opacity="0.72"/><circle cx="402.7" cy="388.7" r="2.6" fill="#1D8155" opacity="0.67"/><circle cx="381.8" cy="360.5" r="5.4" fill="#C4762A" opacity="0.89"/><circle cx="384.6" cy="384.3" r="2.2" fill="#9A6614" opacity="0.83"/><circle cx="396.7" cy="338.4" r="3.6" fill="#1D8155" opacity="0.61"/><circle cx="375.1" cy="383.2" r="2.6" fill="#9A6614" opacity="0.86"/><circle cx="346.4" cy="370.7" r="2.2" fill="#1D8155" opacity="0.70"/><circle cx="376.8" cy="400.5" r="2.2" fill="#6B9A3D" opacity="0.69"/><circle cx="363.9" cy="383.4" r="3.6" fill="#9A6614" opacity="0.82"/><circle cx="395.8" cy="349.3" r="2.2" fill="#1D8155" opacity="0.71"/><circle cx="345.6" cy="355.6" r="2.6" fill="#1D8155" opacity="0.66"/><circle cx="393.6" cy="402.7" r="2.2" fill="#1D8155" opacity="0.62"/><circle cx="366.5" cy="360.6" r="4.4" fill="#C4762A" opacity="0.86"/><circle cx="397.1" cy="380.4" r="2.2" fill="#1D8155" opacity="0.76"/><circle cx="351.2" cy="347.2" r="3.1" fill="#1D8155" opacity="0.66"/><circle cx="386.2" cy="344.5" r="2.6" fill="#1D8155" opacity="0.72"/><circle cx="353.0" cy="366.3" r="2.2" fill="#9A6614" opacity="0.76"/><circle cx="359.7" cy="339.1" r="2.6" fill="#1D8155" opacity="0.64"/><circle cx="404.3" cy="370.8" r="2.2" fill="#9A6614" opacity="0.71"/><circle cx="405.3" cy="343.2" r="3.1" fill="#1D8155" opacity="0.59"/><circle cx="373.0" cy="372.4" r="2.6" fill="#9A6614" opacity="0.96"/><circle cx="406.5" cy="382.0" r="3.6" fill="#1D8155" opacity="0.67"/><circle cx="394.2" cy="386.9" r="3.1" fill="#9A6614" opacity="0.75"/><circle cx="337.9" cy="380.1" r="3.6" fill="#1D8155" opacity="0.60"/><circle cx="412.6" cy="358.9" r="2.6" fill="#1D8155" opacity="0.61"/><circle cx="387.1" cy="404.6" r="2.2" fill="#1D8155" opacity="0.63"/><circle cx="348.6" cy="389.1" r="3.1" fill="#1D8155" opacity="0.66"/><circle cx="395.5" cy="360.4" r="2.6" fill="#6B9A3D" opacity="0.78"/><circle cx="337.6" cy="369.9" r="2.2" fill="#6B9A3D" opacity="0.61"/><circle cx="388.2" cy="350.4" r="2.2" fill="#6B9A3D" opacity="0.76"/><circle cx="365.3" cy="407.7" r="3.1" fill="#1D8155" opacity="0.60"/><circle cx="381.6" cy="407.3" r="2.2" fill="#6B9A3D" opacity="0.61"/><circle cx="363.6" cy="373.6" r="2.2" fill="#C4762A" opacity="0.87"/><circle cx="414.6" cy="367.7" r="3.1" fill="#1D8155" opacity="0.60"/><circle cx="354.6" cy="357.3" r="2.2" fill="#9A6614" opacity="0.75"/><circle cx="387.7" cy="368.0" r="2.2" fill="#9A6614" opacity="0.88"/><circle cx="394.3" cy="370.8" r="3.1" fill="#9A6614" opacity="0.81"/><circle cx="406.0" cy="352.6" r="2.2" fill="#6B9A3D" opacity="0.65"/><circle cx="399.9" cy="396.8" r="3.1" fill="#1D8155" opacity="0.63"/><circle cx="380.0" cy="376.5" r="3.1" fill="#C4762A" opacity="0.92"/><circle cx="370.7" cy="378.8" r="2.2" fill="#B23A2E" opacity="0.90"/><circle cx="364.1" cy="350.8" r="3.1" fill="#6B9A3D" opacity="0.77"/><circle cx="342.3" cy="362.5" r="3.1" fill="#1D8155" opacity="0.65"/><circle cx="387.1" cy="334.7" r="2.2" fill="#6B9A3D" opacity="0.62"/><circle cx="367.7" cy="331.3" r="3.1" fill="#1D8155" opacity="0.59"/><circle cx="353.7" cy="373.3" r="2.2" fill="#6B9A3D" opacity="0.77"/><circle cx="347.0" cy="377.8" r="2.2" fill="#1D8155" opacity="0.69"/><circle cx="361.0" cy="389.9" r="2.2" fill="#9A6614" opacity="0.75"/><circle cx="383.9" cy="397.7" r="2.2" fill="#9A6614" opacity="0.71"/><circle cx="364.2" cy="398.2" r="2.2" fill="#6B9A3D" opacity="0.69"/><circle cx="373.8" cy="405.5" r="2.2" fill="#1D8155" opacity="0.64"/><circle cx="450" cy="318" r="32" fill="rgba(88,67,108,.05)"/><circle cx="472.5" cy="330.2" r="2.2" fill="#1D8155" opacity="0.64"/><circle cx="463.4" cy="312.4" r="2.6" fill="#6B9A3D" opacity="0.80"/><circle cx="424.1" cy="329.8" r="2.2" fill="#1D8155" opacity="0.60"/><circle cx="439.2" cy="323.2" r="2.2" fill="#9A6614" opacity="0.83"/><circle cx="475.2" cy="316.4" r="5.4" fill="#1D8155" opacity="0.65"/><circle cx="445.0" cy="313.0" r="4.4" fill="#9A6614" opacity="0.90"/><circle cx="445.6" cy="297.4" r="2.6" fill="#1D8155" opacity="0.70"/><circle cx="453.8" cy="332.5" r="3.6" fill="#6B9A3D" opacity="0.79"/><circle cx="447.2" cy="323.0" r="3.6" fill="#C4762A" opacity="0.92"/><circle cx="462.1" cy="322.5" r="3.1" fill="#C4762A" opacity="0.82"/><circle cx="431.8" cy="310.4" r="2.2" fill="#9A6614" opacity="0.72"/><circle cx="458.0" cy="299.4" r="2.6" fill="#1D8155" opacity="0.72"/><circle cx="455.2" cy="346.2" r="3.1" fill="#1D8155" opacity="0.60"/><circle cx="455.5" cy="307.5" r="3.6" fill="#9A6614" opacity="0.83"/><circle cx="441.9" cy="305.4" r="2.2" fill="#6B9A3D" opacity="0.79"/><circle cx="432.4" cy="334.4" r="2.2" fill="#1D8155" opacity="0.66"/><circle cx="433.0" cy="319.8" r="3.1" fill="#6B9A3D" opacity="0.76"/><circle cx="422.7" cy="307.4" r="4.4" fill="#1D8155" opacity="0.59"/><circle cx="425.0" cy="320.8" r="2.6" fill="#6B9A3D" opacity="0.65"/><circle cx="466.1" cy="299.1" r="3.6" fill="#1D8155" opacity="0.65"/><circle cx="450.7" cy="292.8" r="2.6" fill="#6B9A3D" opacity="0.65"/><circle cx="436.7" cy="341.7" r="2.2" fill="#1D8155" opacity="0.62"/><circle cx="436.0" cy="329.6" r="2.2" fill="#9A6614" opacity="0.75"/><circle cx="453.1" cy="316.1" r="2.6" fill="#9A6614" opacity="0.95"/><circle cx="431.6" cy="299.6" r="3.6" fill="#1D8155" opacity="0.63"/><circle cx="442.1" cy="332.9" r="2.2" fill="#6B9A3D" opacity="0.76"/><circle cx="448.7" cy="341.6" r="2.6" fill="#6B9A3D" opacity="0.67"/><circle cx="465.9" cy="339.2" r="2.2" fill="#1D8155" opacity="0.63"/><circle cx="462.2" cy="331.2" r="2.2" fill="#1D8155" opacity="0.75"/><circle cx="474.2" cy="303.1" r="2.6" fill="#1D8155" opacity="0.60"/><circle cx="450.5" cy="302.5" r="2.2" fill="#9A6614" opacity="0.78"/><circle cx="443.4" cy="346.2" r="3.1" fill="#1D8155" opacity="0.59"/><circle cx="438.7" cy="292.6" r="2.6" fill="#1D8155" opacity="0.61"/><circle cx="467.6" cy="327.1" r="2.2" fill="#9A6614" opacity="0.72"/><circle cx="461.7" cy="291.2" r="3.6" fill="#1D8155" opacity="0.59"/><circle cx="459.5" cy="340.9" r="2.2" fill="#1D8155" opacity="0.65"/><circle cx="420.8" cy="315.7" r="2.6" fill="#1D8155" opacity="0.59"/><circle cx="469.2" cy="308.4" r="3.1" fill="#9A6614" opacity="0.70"/><circle cx="208" cy="406" r="30" fill="rgba(88,67,108,.05)"/><circle cx="215.5" cy="430.4" r="3.6" fill="#1D8155" opacity="0.62"/><circle cx="194.3" cy="402.3" r="2.2" fill="#C4762A" opacity="0.79"/><circle cx="230.9" cy="399.5" r="4.4" fill="#1D8155" opacity="0.64"/><circle cx="206.4" cy="403.2" r="5.4" fill="#B23A2E" opacity="0.95"/><circle cx="226.9" cy="421.4" r="3.6" fill="#6B9A3D" opacity="0.63"/><circle cx="220.2" cy="385.5" r="3.1" fill="#6B9A3D" opacity="0.64"/><circle cx="194.7" cy="393.1" r="4.4" fill="#1D8155" opacity="0.72"/><circle cx="187.7" cy="410.6" r="4.4" fill="#6B9A3D" opacity="0.69"/><circle cx="201.9" cy="431.3" r="2.2" fill="#1D8155" opacity="0.61"/><circle cx="188.1" cy="422.5" r="4.4" fill="#6B9A3D" opacity="0.61"/><circle cx="219.4" cy="416.8" r="2.2" fill="#9A6614" opacity="0.76"/><circle cx="200.7" cy="420.4" r="2.2" fill="#6B9A3D" opacity="0.76"/><circle cx="212.8" cy="420.4" r="2.2" fill="#9A6614" opacity="0.77"/><circle cx="201.7" cy="411.5" r="2.2" fill="#9A6614" opacity="0.87"/><circle cx="218.4" cy="401.3" r="3.1" fill="#C4762A" opacity="0.83"/><circle cx="206.6" cy="386.7" r="2.2" fill="#6B9A3D" opacity="0.71"/><circle cx="211.4" cy="394.7" r="2.6" fill="#C4762A" opacity="0.82"/><circle cx="182.8" cy="402.1" r="2.6" fill="#1D8155" opacity="0.62"/><circle cx="212.6" cy="412.1" r="3.1" fill="#C4762A" opacity="0.89"/><circle cx="232.5" cy="416.4" r="2.2" fill="#1D8155" opacity="0.60"/><circle cx="221.5" cy="393.6" r="3.1" fill="#6B9A3D" opacity="0.73"/><circle cx="199.4" cy="384.2" r="2.6" fill="#1D8155" opacity="0.65"/><circle cx="208.0" cy="428.0" r="2.6" fill="#6B9A3D" opacity="0.67"/><circle cx="211.8" cy="379.2" r="4.4" fill="#1D8155" opacity="0.59"/><circle cx="206.9" cy="420.1" r="2.2" fill="#6B9A3D" opacity="0.79"/><circle cx="229.8" cy="390.6" r="2.6" fill="#1D8155" opacity="0.60"/><circle cx="195.7" cy="415.7" r="2.6" fill="#6B9A3D" opacity="0.77"/><circle cx="228.3" cy="408.5" r="3.1" fill="#9A6614" opacity="0.69"/><circle cx="189.0" cy="387.5" r="2.2" fill="#1D8155" opacity="0.60"/></g><text x="70" y="424" class="cap">2,802 files scored</text><text x="526" y="280" class="cap">OPEN FINDINGS</text><rect x="526" y="300" width="236" height="24" rx="6" fill="#F4EAE1"/><rect x="526" y="300" width="3" height="24" rx="1.5" fill="#B23A2E"/><text x="539" y="316" class="find">CHANGE ENTROPY</text><text x="752" y="316" class="find" fill="#B23A2E" text-anchor="end" font-weight="700">−3.0</text><rect x="526" y="328" width="236" height="24" rx="6" fill="#F4EAE1"/><rect x="526" y="328" width="3" height="24" rx="1.5" fill="#B23A2E"/><text x="539" y="344" class="find">NESTED COMPLEXITY</text><text x="752" y="344" class="find" fill="#B23A2E" text-anchor="end" font-weight="700">−2.4</text><rect x="526" y="356" width="236" height="24" rx="6" fill="#F4EAE1"/><rect x="526" y="356" width="3" height="24" rx="1.5" fill="#9A6614"/><text x="539" y="372" class="find">CHURN RISK</text><text x="752" y="372" class="find" fill="#9A6614" text-anchor="end" font-weight="700">−1.8</text><rect x="526" y="384" width="236" height="24" rx="6" fill="#F4EAE1"/><rect x="526" y="384" width="3" height="24" rx="1.5" fill="#9A6614"/><text x="539" y="400" class="find">CO-CHANGE SCATTER</text><text x="752" y="400" class="find" fill="#9A6614" text-anchor="end" font-weight="700">−1.5</text><rect x="526" y="412" width="236" height="24" rx="6" fill="#F4EAE1"/><rect x="526" y="412" width="3" height="24" rx="1.5" fill="#9A6614"/><text x="539" y="428" class="find">UNTESTED HOTSPOT</text><text x="752" y="428" class="find" fill="#9A6614" text-anchor="end" font-weight="700">−1.2</text><rect x="802" y="104" width="380" height="352" rx="14" fill="#FFFFFF" stroke="rgba(88,67,108,.12)" stroke-width="1"/><text x="824" y="135" class="lbl">DEPENDENCY GRAPH</text><text x="1160" y="135" class="kick" text-anchor="end">16 languages</text><rect x="820" y="150" width="344" height="240" rx="9" fill="#F4EAE1"/><clipPath id="gclip"><rect x="820" y="150" width="344" height="240" rx="9"/></clipPath><g clip-path="url(#gclip)"><line x1="944.0" y1="201.8" x2="912" y2="224" stroke="rgba(88,67,108,.15)" stroke-width="1"/><line x1="907.3" y1="185.9" x2="912" y2="224" stroke="rgba(88,67,108,.15)" stroke-width="1"/><line x1="913.1" y1="193.3" x2="912" y2="224" stroke="rgba(88,67,108,.15)" stroke-width="1"/><line x1="948.1" y1="258.8" x2="912" y2="224" stroke="rgba(88,67,108,.15)" stroke-width="1"/><line x1="901.9" y1="182.8" x2="912" y2="224" stroke="rgba(88,67,108,.15)" stroke-width="1"/><line x1="895.4" y1="193.7" x2="912" y2="224" stroke="rgba(88,67,108,.15)" stroke-width="1"/><line x1="888.8" y1="197.7" x2="912" y2="224" stroke="rgba(88,67,108,.15)" stroke-width="1"/><line x1="929.0" y1="238.4" x2="912" y2="224" stroke="rgba(88,67,108,.15)" stroke-width="1"/><line x1="951.9" y1="214.2" x2="912" y2="224" stroke="rgba(88,67,108,.15)" stroke-width="1"/><line x1="1040.2" y1="228.2" x2="1052" y2="212" stroke="rgba(88,67,108,.15)" stroke-width="1"/><line x1="1048.4" y1="190.3" x2="1052" y2="212" stroke="rgba(88,67,108,.15)" stroke-width="1"/><line x1="1068.5" y1="249.3" x2="1052" y2="212" stroke="rgba(88,67,108,.15)" stroke-width="1"/><line x1="1013.8" y1="237.1" x2="1052" y2="212" stroke="rgba(88,67,108,.15)" stroke-width="1"/><line x1="1009.6" y1="237.8" x2="1052" y2="212" stroke="rgba(88,67,108,.15)" stroke-width="1"/><line x1="1065.3" y1="185.2" x2="1052" y2="212" stroke="rgba(88,67,108,.15)" stroke-width="1"/><line x1="1017.4" y1="194.6" x2="1052" y2="212" stroke="rgba(88,67,108,.15)" stroke-width="1"/><line x1="1107.4" y1="209.0" x2="1052" y2="212" stroke="rgba(88,67,108,.15)" stroke-width="1"/><line x1="1061.0" y1="174.5" x2="1052" y2="212" stroke="rgba(88,67,108,.15)" stroke-width="1"/><line x1="944.8" y1="323.3" x2="982" y2="322" stroke="rgba(88,67,108,.15)" stroke-width="1"/><line x1="955.1" y1="338.1" x2="982" y2="322" stroke="rgba(88,67,108,.15)" stroke-width="1"/><line x1="1001.7" y1="341.5" x2="982" y2="322" stroke="rgba(88,67,108,.15)" stroke-width="1"/><line x1="1022.9" y1="322.0" x2="982" y2="322" stroke="rgba(88,67,108,.15)" stroke-width="1"/><line x1="961.2" y1="356.5" x2="982" y2="322" stroke="rgba(88,67,108,.15)" stroke-width="1"/><line x1="942.1" y1="301.8" x2="982" y2="322" stroke="rgba(88,67,108,.15)" stroke-width="1"/><line x1="943.8" y1="346.7" x2="982" y2="322" stroke="rgba(88,67,108,.15)" stroke-width="1"/><line x1="983.8" y1="296.5" x2="982" y2="322" stroke="rgba(88,67,108,.15)" stroke-width="1"/><line x1="949.1" y1="320.7" x2="982" y2="322" stroke="rgba(88,67,108,.15)" stroke-width="1"/><line x1="1131.2" y1="323.1" x2="1098" y2="310" stroke="rgba(88,67,108,.15)" stroke-width="1"/><line x1="1085.5" y1="327.7" x2="1098" y2="310" stroke="rgba(88,67,108,.15)" stroke-width="1"/><line x1="1106.7" y1="286.0" x2="1098" y2="310" stroke="rgba(88,67,108,.15)" stroke-width="1"/><line x1="1152.2" y1="302.0" x2="1098" y2="310" stroke="rgba(88,67,108,.15)" stroke-width="1"/><line x1="1068.9" y1="335.5" x2="1098" y2="310" stroke="rgba(88,67,108,.15)" stroke-width="1"/><line x1="1148.6" y1="288.2" x2="1098" y2="310" stroke="rgba(88,67,108,.15)" stroke-width="1"/><line x1="1064.8" y1="323.3" x2="1098" y2="310" stroke="rgba(88,67,108,.15)" stroke-width="1"/><line x1="1046.2" y1="319.4" x2="1098" y2="310" stroke="rgba(88,67,108,.15)" stroke-width="1"/><line x1="1137.2" y1="290.0" x2="1098" y2="310" stroke="rgba(88,67,108,.15)" stroke-width="1"/><line x1="912" y1="224" x2="1052" y2="212" stroke="rgba(88,67,108,.30)" stroke-width="1.6"/><line x1="912" y1="224" x2="982" y2="322" stroke="rgba(88,67,108,.30)" stroke-width="1.6"/><line x1="912" y1="224" x2="1098" y2="310" stroke="rgba(88,67,108,.30)" stroke-width="1.6"/><line x1="1052" y1="212" x2="982" y2="322" stroke="rgba(88,67,108,.30)" stroke-width="1.6"/><line x1="982" y1="322" x2="1098" y2="310" stroke="rgba(88,67,108,.30)" stroke-width="1.6"/><circle cx="912.0" cy="224.0" r="9.0" fill="#C0641A" stroke="#FFFFFF" stroke-width="1.5"/><circle cx="944.0" cy="201.8" r="4.0" fill="#F7A94D"/><circle cx="907.3" cy="185.9" r="3.3" fill="#F7A94D"/><circle cx="913.1" cy="193.3" r="2.8" fill="#F7A94D"/><circle cx="948.1" cy="258.8" r="3.8" fill="#F7A94D"/><circle cx="901.9" cy="182.8" r="3.1" fill="#F7A94D"/><circle cx="895.4" cy="193.7" r="3.4" fill="#F7A94D"/><circle cx="888.8" cy="197.7" r="3.3" fill="#F7A94D"/><circle cx="929.0" cy="238.4" r="3.8" fill="#F7A94D"/><circle cx="951.9" cy="214.2" r="3.6" fill="#F7A94D"/><circle cx="1052.0" cy="212.0" r="9.0" fill="#58436C" stroke="#FFFFFF" stroke-width="1.5"/><circle cx="1040.2" cy="228.2" r="3.4" fill="#826AA0"/><circle cx="1048.4" cy="190.3" r="3.5" fill="#826AA0"/><circle cx="1068.5" cy="249.3" r="3.5" fill="#826AA0"/><circle cx="1013.8" cy="237.1" r="3.7" fill="#826AA0"/><circle cx="1009.6" cy="237.8" r="3.8" fill="#826AA0"/><circle cx="1065.3" cy="185.2" r="4.1" fill="#826AA0"/><circle cx="1017.4" cy="194.6" r="4.2" fill="#826AA0"/><circle cx="1107.4" cy="209.0" r="3.7" fill="#826AA0"/><circle cx="1061.0" cy="174.5" r="3.3" fill="#826AA0"/><circle cx="982.0" cy="322.0" r="9.0" fill="#B23A2E" stroke="#FFFFFF" stroke-width="1.5"/><circle cx="944.8" cy="323.3" r="4.2" fill="#CF6A55"/><circle cx="955.1" cy="338.1" r="4.0" fill="#CF6A55"/><circle cx="1001.7" cy="341.5" r="3.2" fill="#CF6A55"/><circle cx="1022.9" cy="322.0" r="3.0" fill="#CF6A55"/><circle cx="961.2" cy="356.5" r="4.0" fill="#CF6A55"/><circle cx="942.1" cy="301.8" r="3.5" fill="#CF6A55"/><circle cx="943.8" cy="346.7" r="3.7" fill="#CF6A55"/><circle cx="983.8" cy="296.5" r="2.9" fill="#CF6A55"/><circle cx="949.1" cy="320.7" r="2.6" fill="#CF6A55"/><circle cx="1098.0" cy="310.0" r="9.0" fill="#6B7A3D" stroke="#FFFFFF" stroke-width="1.5"/><circle cx="1131.2" cy="323.1" r="3.2" fill="#90A05E"/><circle cx="1085.5" cy="327.7" r="4.3" fill="#90A05E"/><circle cx="1106.7" cy="286.0" r="2.9" fill="#90A05E"/><circle cx="1152.2" cy="302.0" r="3.9" fill="#90A05E"/><circle cx="1068.9" cy="335.5" r="2.8" fill="#90A05E"/><circle cx="1148.6" cy="288.2" r="4.2" fill="#90A05E"/><circle cx="1064.8" cy="323.3" r="3.6" fill="#90A05E"/><circle cx="1046.2" cy="319.4" r="4.0" fill="#90A05E"/><circle cx="1137.2" cy="290.0" r="2.9" fill="#90A05E"/></g><text x="832" y="380" class="cap">28,126 nodes · 11 communities</text><text x="824" y="414" class="out">Who calls this? What breaks if</text><text x="824" y="434" class="out">I change it?</text><rect x="1200" y="104" width="364" height="352" rx="14" fill="#FFFFFF" stroke="rgba(88,67,108,.12)" stroke-width="1"/><text x="1222" y="135" class="lbl">GIT HISTORY</text><text x="1542" y="135" class="kick" text-anchor="end">1,050 commits</text><rect x="1218" y="150" width="328" height="240" rx="9" fill="#F4EAE1"/><rect x="1230.0" y="312.2" width="16.0" height="13.8" rx="2.5" fill="#1D8155" opacity="0.82"/><rect x="1249.0" y="300.7" width="16.0" height="25.3" rx="2.5" fill="#1D8155" opacity="0.82"/><rect x="1268.0" y="282.3" width="16.0" height="43.7" rx="2.5" fill="#1D8155" opacity="0.82"/><rect x="1287.0" y="261.6" width="16.0" height="64.4" rx="2.5" fill="#1D8155" opacity="0.82"/><rect x="1306.0" y="231.7" width="16.0" height="94.3" rx="2.5" fill="#1D8155" opacity="0.82"/><rect x="1325.0" y="206.4" width="16.0" height="119.6" rx="2.5" fill="#1D8155" opacity="0.82"/><rect x="1344.0" y="181.1" width="16.0" height="144.9" rx="2.5" fill="#9A6614" opacity="0.82"/><rect x="1363.0" y="192.6" width="16.0" height="133.4" rx="2.5" fill="#9A6614" opacity="0.82"/><rect x="1382.0" y="217.9" width="16.0" height="108.1" rx="2.5" fill="#9A6614" opacity="0.82"/><rect x="1401.0" y="243.2" width="16.0" height="82.8" rx="2.5" fill="#9A6614" opacity="0.82"/><rect x="1420.0" y="263.9" width="16.0" height="62.1" rx="2.5" fill="#9A6614" opacity="0.82"/><rect x="1439.0" y="282.3" width="16.0" height="43.7" rx="2.5" fill="#B23A2E" opacity="0.82"/><rect x="1458.0" y="296.1" width="16.0" height="29.9" rx="2.5" fill="#B23A2E" opacity="0.82"/><rect x="1477.0" y="305.3" width="16.0" height="20.7" rx="2.5" fill="#B23A2E" opacity="0.82"/><rect x="1496.0" y="312.2" width="16.0" height="13.8" rx="2.5" fill="#B23A2E" opacity="0.82"/><rect x="1515.0" y="316.8" width="16.0" height="9.2" rx="2.5" fill="#B23A2E" opacity="0.82"/><line x1="1230" y1="327" x2="1534" y2="327" stroke="rgba(88,67,108,.20)"/><text x="1230" y="348" class="cap">low risk</text><text x="1534" y="348" class="cap" text-anchor="end">high risk</text><text x="1230" y="176" class="cap">CHANGE-RISK DISTRIBUTION</text><text x="1230" y="380" class="cap">hotspots · ownership · co-change · bus factor</text><text x="1222" y="414" class="out">Which of these 40 files is</text><text x="1222" y="434" class="out">actually dangerous?</text><rect x="36" y="474" width="496" height="302" rx="14" fill="#FFFFFF" stroke="rgba(88,67,108,.12)" stroke-width="1"/><text x="58" y="505" class="lbl">GENERATED DOCS</text><text x="510" y="505" class="kick" text-anchor="end">1,274 pages</text><rect x="54" y="520" width="460" height="176" rx="9" fill="#F4EAE1"/><rect x="70" y="536" width="265" height="11" rx="3" fill="#58436C" opacity=".85"/><rect x="70" y="562" width="394" height="6" rx="3" fill="rgba(88,67,108,.26)"/><rect x="70" y="578" width="368" height="6" rx="3" fill="rgba(88,67,108,.26)"/><rect x="70" y="594" width="304" height="6" rx="3" fill="rgba(88,67,108,.26)"/><rect x="70" y="607" width="188" height="9" rx="3" fill="#A16215" opacity=".55"/><rect x="70" y="630" width="377" height="6" rx="3" fill="rgba(88,67,108,.26)"/><rect x="70" y="646" width="338" height="6" rx="3" fill="rgba(88,67,108,.26)"/><rect x="70" y="662" width="235" height="6" rx="3" fill="rgba(88,67,108,.26)"/><rect x="416" y="530" width="86" height="20" rx="10" fill="rgba(29,129,85,.12)"/><text x="459" y="544" class="pill" fill="#1D8155" text-anchor="middle">fresh · v6</text><text x="58" y="730" class="out">Why does auth work this way?</text><text x="58" y="751" class="outsub">Answered from the wiki, not from 12 file reads.</text><rect x="550" y="474" width="496" height="302" rx="14" fill="#FFFFFF" stroke="rgba(88,67,108,.12)" stroke-width="1"/><text x="572" y="505" class="lbl">ARCHITECTURAL DECISIONS</text><text x="1024" y="505" class="kick" text-anchor="end">mined from 8 sources</text><rect x="568" y="520" width="460" height="176" rx="9" fill="#F4EAE1"/><text x="584" y="548" class="dec">Rebuild the KG skeleton only when the</text><text x="584" y="568" class="dec">graph fingerprint changes</text><rect x="584" y="582" width="392" height="1" fill="rgba(88,67,108,.15)"/><rect x="584" y="596" width="101" height="21" rx="10.5" fill="#FFFFFF" stroke="rgba(88,67,108,.18)"/><text x="635" y="610" class="pill" fill="#1D8155" text-anchor="middle">Verified quote</text><rect x="693.4" y="596" width="108" height="21" rx="10.5" fill="#FFFFFF" stroke="rgba(88,67,108,.18)"/><text x="747" y="610" class="pill" fill="#58436C" text-anchor="middle">supersedes #418</text><rect x="809.4" y="596" width="115" height="21" rx="10.5" fill="#FFFFFF" stroke="rgba(88,67,108,.18)"/><text x="867" y="610" class="pill" fill="#A16215" text-anchor="middle">governs 24 files</text><text x="584" y="660" class="cap">Delivered back at session start, and the moment</text><text x="584" y="676" class="cap">your agent edits a file the decision governs.</text><text x="572" y="730" class="out">Don’t re-litigate what we settled.</text><text x="572" y="751" class="outsub">Captured nowhere else.</text><rect x="1064" y="474" width="500" height="302" rx="14" fill="#FFFFFF" stroke="rgba(88,67,108,.12)" stroke-width="1"/><text x="1086" y="505" class="lbl">SERVED TO YOUR AGENT</text><text x="1542" y="505" class="kick" text-anchor="end">10 MCP tools</text><rect x="1084" y="528" width="95" height="24" rx="6" fill="#F4EAE1" stroke="rgba(88,67,108,.12)"/><text x="1131" y="544" class="mono" text-anchor="middle">get_overview</text><rect x="1187" y="528" width="81" height="24" rx="6" fill="#F4EAE1" stroke="rgba(88,67,108,.12)"/><text x="1227" y="544" class="mono" text-anchor="middle">get_answer</text><rect x="1276" y="528" width="88" height="24" rx="6" fill="#F4EAE1" stroke="rgba(88,67,108,.12)"/><text x="1320" y="544" class="mono" text-anchor="middle">get_context</text><rect x="1372" y="528" width="81" height="24" rx="6" fill="#F4EAE1" stroke="rgba(88,67,108,.12)"/><text x="1412" y="544" class="mono" text-anchor="middle">get_symbol</text><rect x="1084" y="558" width="116" height="24" rx="6" fill="#F4EAE1" stroke="rgba(88,67,108,.12)"/><text x="1142" y="574" class="mono" text-anchor="middle">search_codebase</text><rect x="1208" y="558" width="67" height="24" rx="6" fill="#F4EAE1" stroke="rgba(88,67,108,.12)"/><text x="1241" y="574" class="mono" text-anchor="middle">get_risk</text><rect x="1283" y="558" width="116" height="24" rx="6" fill="#F4EAE1" stroke="rgba(88,67,108,.12)"/><text x="1340" y="574" class="mono" text-anchor="middle">get_change_risk</text><rect x="1406" y="558" width="60" height="24" rx="6" fill="#F4EAE1" stroke="rgba(88,67,108,.12)"/><text x="1436" y="574" class="mono" text-anchor="middle">get_why</text><rect x="1084" y="588" width="102" height="24" rx="6" fill="#F4EAE1" stroke="rgba(88,67,108,.12)"/><text x="1135" y="604" class="mono" text-anchor="middle">get_dead_code</text><rect x="1194" y="588" width="81" height="24" rx="6" fill="#F4EAE1" stroke="rgba(88,67,108,.12)"/><text x="1234" y="604" class="mono" text-anchor="middle">get_health</text><text x="1084" y="636" class="cap">Every response carries index age, indexed commit, and a staleness warning.</text><rect x="1084" y="650" width="460" height="46" rx="9" fill="rgba(245,149,32,.10)" stroke="rgba(245,149,32,.28)"/><text x="1098" y="670" class="chipval" fill="#A16215" font-size="19">−96%<tspan class="chipunit" fill="#5E5360"> context tokens</tspan></text><text x="1098" y="688" class="cap">−70% tool calls · −89% file reads · fewer round-trips per answer</text><text x="1086" y="730" class="out">Claude Code, Codex, Cursor, VS Code.</text><text x="1086" y="751" class="outsub">Plus hooks, so it arrives unasked.</text></svg>\ No newline at end of file+</style><rect width="1600" height="812" fill="#FBF6F1"/><text x="36" y="50" class="h1">One index. Everything your agent and your team keep asking for.</text><text x="36" y="80" class="sub">Index once, in seconds, no LLM required. Every answer below is already computed.</text><rect x="36" y="104" width="748" height="352" rx="14" fill="#FFFFFF" stroke="rgba(88,67,108,.12)" stroke-width="1"/><text x="58" y="135" class="lbl">CODE HEALTH</text><text x="762" y="135" class="kick" text-anchor="end">49 detectors · zero LLM · <30s</text><rect x="58" y="150" width="146" height="58" rx="9" fill="#F4EAE1" stroke="rgba(88,67,108,.12)"/><text x="70" y="170" class="chiplbl">Defect risk</text><text x="70" y="197" class="chipval" fill="#9A6614">7.5<tspan class="chipunit">/10</tspan></text><rect x="214" y="150" width="146" height="58" rx="9" fill="#F4EAE1" stroke="rgba(88,67,108,.12)"/><text x="226" y="170" class="chiplbl">Maintainability</text><text x="226" y="197" class="chipval" fill="#1D8155">8.6<tspan class="chipunit">/10</tspan></text><rect x="370" y="150" width="146" height="58" rx="9" fill="#F4EAE1" stroke="rgba(88,67,108,.12)"/><text x="382" y="170" class="chiplbl">Perf risks</text><text x="382" y="197" class="chipval" fill="#58436C">268<tspan class="chipunit"></tspan></text><rect x="58" y="220" width="704" height="34" rx="8" fill="rgba(29,129,85,.09)" stroke="rgba(29,129,85,.22)"/><circle cx="76" cy="237" r="4.5" fill="#1D8155"/><text x="90" y="242" class="trust">17 of the 20 lowest-health files were bug-fixed in the last 6 months, <tspan font-weight="700">3.63× the 23% baseline</tspan></text><rect x="58" y="266" width="452" height="170" rx="9" fill="#F4EAE1"/><clipPath id="galaxy"><rect x="58" y="266" width="452" height="170" rx="9"/></clipPath><g clip-path="url(#galaxy)"><circle cx="150" cy="346" r="62" fill="rgba(88,67,108,.05)"/><circle cx="140.1" cy="365.8" r="5.4" fill="#6B9A3D" opacity="0.84"/><circle cx="157.6" cy="330.2" r="4.4" fill="#9A6614" opacity="0.87"/><circle cx="139.0" cy="345.5" r="3.6" fill="#C4762A" opacity="0.92"/><circle cx="152.5" cy="388.3" r="2.2" fill="#9A6614" opacity="0.69"/><circle cx="169.2" cy="364.9" r="5.4" fill="#9A6614" opacity="0.80"/><circle cx="190.4" cy="362.5" r="2.2" fill="#9A6614" opacity="0.68"/><circle cx="200.6" cy="361.2" r="3.1" fill="#1D8155" opacity="0.62"/><circle cx="108.3" cy="335.1" r="4.4" fill="#6B9A3D" opacity="0.69"/><circle cx="168.3" cy="385.4" r="5.4" fill="#1D8155" opacity="0.68"/><circle cx="189.4" cy="373.7" r="4.4" fill="#1D8155" opacity="0.65"/><circle cx="156.6" cy="307.6" r="3.6" fill="#6B9A3D" opacity="0.72"/><circle cx="150.2" cy="370.2" r="2.6" fill="#6B9A3D" opacity="0.82"/><circle cx="137.5" cy="384.2" r="3.1" fill="#6B9A3D" opacity="0.71"/><circle cx="136.7" cy="400.9" r="2.2" fill="#1D8155" opacity="0.59"/><circle cx="100.7" cy="372.4" r="2.2" fill="#1D8155" opacity="0.59"/><circle cx="102.1" cy="322.4" r="3.1" fill="#1D8155" opacity="0.61"/><circle cx="126.3" cy="378.5" r="3.6" fill="#1D8155" opacity="0.71"/><circle cx="142.8" cy="333.9" r="5.4" fill="#9A6614" opacity="0.90"/><circle cx="118.3" cy="391.4" r="3.1" fill="#1D8155" opacity="0.60"/><circle cx="160.3" cy="355.3" r="3.1" fill="#9A6614" opacity="0.90"/><circle cx="169.8" cy="323.3" r="3.6" fill="#C4762A" opacity="0.78"/><circle cx="135.6" cy="313.9" r="2.6" fill="#6B9A3D" opacity="0.74"/><circle cx="201.9" cy="349.9" r="2.6" fill="#1D8155" opacity="0.62"/><circle cx="186.9" cy="346.9" r="3.1" fill="#6B9A3D" opacity="0.73"/><circle cx="193.9" cy="331.8" r="5.4" fill="#1D8155" opacity="0.66"/><circle cx="190.7" cy="316.3" r="5.4" fill="#6B9A3D" opacity="0.63"/><circle cx="121.9" cy="368.5" r="2.2" fill="#6B9A3D" opacity="0.74"/><circle cx="129.8" cy="360.6" r="2.6" fill="#9A6614" opacity="0.82"/><circle cx="172.2" cy="346.0" r="2.2" fill="#C4762A" opacity="0.84"/><circle cx="118.4" cy="349.9" r="2.6" fill="#6B9A3D" opacity="0.77"/><circle cx="119.4" cy="382.2" r="2.2" fill="#6B9A3D" opacity="0.66"/><circle cx="178.5" cy="327.5" r="2.6" fill="#6B9A3D" opacity="0.75"/><circle cx="111.6" cy="312.9" r="2.6" fill="#6B9A3D" opacity="0.63"/><circle cx="170.4" cy="301.4" r="2.6" fill="#1D8155" opacity="0.64"/><circle cx="101.3" cy="348.2" r="2.2" fill="#6B9A3D" opacity="0.65"/><circle cx="205.6" cy="341.8" r="3.1" fill="#1D8155" opacity="0.60"/><circle cx="181.3" cy="369.4" r="3.1" fill="#1D8155" opacity="0.72"/><circle cx="160.0" cy="297.6" r="3.6" fill="#6B9A3D" opacity="0.64"/><circle cx="108.4" cy="364.3" r="2.2" fill="#6B9A3D" opacity="0.67"/><circle cx="159.8" cy="363.8" r="2.6" fill="#C4762A" opacity="0.85"/><circle cx="104.9" cy="356.0" r="4.4" fill="#6B9A3D" opacity="0.66"/><circle cx="99.9" cy="331.7" r="2.2" fill="#6B9A3D" opacity="0.62"/><circle cx="130.0" cy="330.2" r="3.1" fill="#9A6614" opacity="0.82"/><circle cx="133.2" cy="295.3" r="3.1" fill="#6B9A3D" opacity="0.61"/><circle cx="128.6" cy="318.9" r="3.1" fill="#9A6614" opacity="0.75"/><circle cx="176.2" cy="340.6" r="2.2" fill="#9A6614" opacity="0.81"/><circle cx="93.4" cy="350.5" r="2.6" fill="#1D8155" opacity="0.59"/><circle cx="120.0" cy="362.2" r="2.2" fill="#9A6614" opacity="0.75"/><circle cx="120.9" cy="317.3" r="2.2" fill="#1D8155" opacity="0.70"/><circle cx="173.8" cy="332.6" r="2.2" fill="#6B9A3D" opacity="0.80"/><circle cx="177.5" cy="308.0" r="3.1" fill="#1D8155" opacity="0.66"/><circle cx="149.1" cy="353.3" r="2.2" fill="#B23A2E" opacity="0.95"/><circle cx="143.4" cy="396.7" r="3.1" fill="#1D8155" opacity="0.63"/><circle cx="126.6" cy="310.9" r="3.6" fill="#9A6614" opacity="0.69"/><circle cx="147.2" cy="324.9" r="3.1" fill="#C4762A" opacity="0.85"/><circle cx="148.4" cy="317.2" r="2.6" fill="#6B9A3D" opacity="0.79"/><circle cx="140.6" cy="352.9" r="2.2" fill="#9A6614" opacity="0.92"/><circle cx="142.7" cy="306.6" r="3.1" fill="#6B9A3D" opacity="0.71"/><circle cx="150.8" cy="303.0" r="2.2" fill="#9A6614" opacity="0.69"/><circle cx="183.9" cy="337.2" r="3.6" fill="#6B9A3D" opacity="0.75"/><circle cx="198.2" cy="371.2" r="3.1" fill="#6B9A3D" opacity="0.60"/><circle cx="167.7" cy="399.6" r="3.6" fill="#1D8155" opacity="0.59"/><circle cx="95.0" cy="337.6" r="2.6" fill="#1D8155" opacity="0.60"/><circle cx="150.2" cy="293.8" r="2.2" fill="#6B9A3D" opacity="0.62"/><circle cx="109.0" cy="379.9" r="2.2" fill="#1D8155" opacity="0.61"/><circle cx="123.1" cy="337.7" r="2.6" fill="#6B9A3D" opacity="0.80"/><circle cx="159.3" cy="322.1" r="2.2" fill="#C4762A" opacity="0.81"/><circle cx="157.4" cy="374.7" r="2.2" fill="#9A6614" opacity="0.78"/><circle cx="184.6" cy="301.7" r="2.6" fill="#6B9A3D" opacity="0.59"/><circle cx="144.3" cy="377.6" r="2.6" fill="#6B9A3D" opacity="0.77"/><circle cx="130.5" cy="393.8" r="2.6" fill="#1D8155" opacity="0.63"/><circle cx="163.8" cy="341.2" r="2.6" fill="#9A6614" opacity="0.89"/><circle cx="177.0" cy="375.8" r="2.2" fill="#9A6614" opacity="0.71"/><circle cx="150.2" cy="360.5" r="2.2" fill="#C4762A" opacity="0.89"/><circle cx="155.7" cy="315.6" r="3.1" fill="#9A6614" opacity="0.78"/><circle cx="180.6" cy="390.0" r="4.4" fill="#1D8155" opacity="0.61"/><circle cx="176.3" cy="357.3" r="2.6" fill="#9A6614" opacity="0.79"/><circle cx="156.0" cy="396.9" r="5.4" fill="#1D8155" opacity="0.63"/><circle cx="153.7" cy="346.9" r="2.6" fill="#B23A2E" opacity="0.97"/><circle cx="135.1" cy="325.0" r="2.2" fill="#C4762A" opacity="0.81"/><circle cx="119.6" cy="329.8" r="2.6" fill="#6B9A3D" opacity="0.75"/><circle cx="116.4" cy="305.3" r="4.4" fill="#1D8155" opacity="0.62"/><circle cx="169.0" cy="374.5" r="2.6" fill="#6B9A3D" opacity="0.75"/><circle cx="141.5" cy="292.6" r="2.2" fill="#6B9A3D" opacity="0.61"/><circle cx="125.9" cy="345.5" r="2.2" fill="#9A6614" opacity="0.82"/><circle cx="163.4" cy="312.5" r="2.2" fill="#1D8155" opacity="0.74"/><circle cx="112.5" cy="321.6" r="2.6" fill="#1D8155" opacity="0.68"/><circle cx="199.4" cy="320.9" r="3.1" fill="#1D8155" opacity="0.60"/><circle cx="168.4" cy="334.5" r="2.2" fill="#6B9A3D" opacity="0.84"/><circle cx="132.9" cy="339.1" r="3.1" fill="#9A6614" opacity="0.87"/><circle cx="190.8" cy="356.0" r="2.6" fill="#6B9A3D" opacity="0.69"/><circle cx="149.5" cy="381.2" r="2.2" fill="#1D8155" opacity="0.74"/><circle cx="276" cy="328" r="50" fill="rgba(88,67,108,.05)"/><circle cx="250.0" cy="342.6" r="5.4" fill="#9A6614" opacity="0.73"/><circle cx="262.7" cy="320.7" r="3.1" fill="#C4762A" opacity="0.86"/><circle cx="307.0" cy="315.7" r="2.6" fill="#9A6614" opacity="0.70"/><circle cx="303.9" cy="357.4" r="5.4" fill="#1D8155" opacity="0.64"/><circle cx="251.5" cy="297.6" r="2.6" fill="#1D8155" opacity="0.65"/><circle cx="250.0" cy="361.6" r="3.1" fill="#6B9A3D" opacity="0.62"/><circle cx="258.5" cy="285.9" r="5.4" fill="#1D8155" opacity="0.59"/><circle cx="315.1" cy="328.7" r="3.1" fill="#1D8155" opacity="0.65"/><circle cx="261.3" cy="306.5" r="3.6" fill="#6B9A3D" opacity="0.77"/><circle cx="290.5" cy="348.7" r="3.6" fill="#6B9A3D" opacity="0.77"/><circle cx="296.6" cy="289.4" r="2.6" fill="#1D8155" opacity="0.61"/><circle cx="266.4" cy="362.6" r="4.4" fill="#1D8155" opacity="0.68"/><circle cx="280.2" cy="368.3" r="3.1" fill="#6B9A3D" opacity="0.64"/><circle cx="283.5" cy="356.2" r="4.4" fill="#6B9A3D" opacity="0.74"/><circle cx="303.8" cy="305.3" r="2.2" fill="#6B9A3D" opacity="0.68"/><circle cx="313.1" cy="348.0" r="5.4" fill="#6B9A3D" opacity="0.62"/><circle cx="253.7" cy="320.0" r="2.6" fill="#9A6614" opacity="0.79"/><circle cx="242.1" cy="333.7" r="3.6" fill="#6B9A3D" opacity="0.69"/><circle cx="265.2" cy="297.4" r="3.1" fill="#9A6614" opacity="0.71"/><circle cx="272.0" cy="322.7" r="2.2" fill="#B23A2E" opacity="0.94"/><circle cx="300.0" cy="317.0" r="2.2" fill="#9A6614" opacity="0.76"/><circle cx="232.6" cy="332.2" r="2.2" fill="#1D8155" opacity="0.61"/><circle cx="261.9" cy="336.3" r="2.2" fill="#C4762A" opacity="0.85"/><circle cx="234.6" cy="313.3" r="3.1" fill="#1D8155" opacity="0.60"/><circle cx="264.4" cy="345.2" r="3.6" fill="#C4762A" opacity="0.81"/><circle cx="300.0" cy="335.7" r="2.2" fill="#6B9A3D" opacity="0.77"/><circle cx="289.9" cy="299.5" r="5.4" fill="#1D8155" opacity="0.71"/><circle cx="279.7" cy="336.0" r="2.2" fill="#B23A2E" opacity="0.92"/><circle cx="288.8" cy="313.0" r="2.2" fill="#9A6614" opacity="0.82"/><circle cx="307.5" cy="324.2" r="3.6" fill="#6B9A3D" opacity="0.71"/><circle cx="303.4" cy="343.7" r="2.6" fill="#6B9A3D" opacity="0.72"/><circle cx="279.2" cy="343.7" r="3.6" fill="#6B9A3D" opacity="0.86"/><circle cx="271.7" cy="305.8" r="5.4" fill="#9A6614" opacity="0.80"/><circle cx="285.0" cy="322.5" r="2.2" fill="#9A6614" opacity="0.90"/><circle cx="239.6" cy="301.5" r="3.6" fill="#1D8155" opacity="0.59"/><circle cx="245.3" cy="354.8" r="3.1" fill="#6B9A3D" opacity="0.63"/><circle cx="258.4" cy="359.2" r="2.2" fill="#6B9A3D" opacity="0.68"/><circle cx="298.6" cy="324.6" r="3.6" fill="#9A6614" opacity="0.79"/><circle cx="275.6" cy="329.8" r="3.1" fill="#C4762A" opacity="0.98"/><circle cx="243.0" cy="320.7" r="3.6" fill="#9A6614" opacity="0.70"/><circle cx="288.2" cy="365.9" r="2.6" fill="#1D8155" opacity="0.64"/><circle cx="255.3" cy="327.2" r="2.6" fill="#6B9A3D" opacity="0.81"/><circle cx="236.0" cy="348.1" r="4.4" fill="#6B9A3D" opacity="0.60"/><circle cx="317.5" cy="313.7" r="2.2" fill="#6B9A3D" opacity="0.61"/><circle cx="310.7" cy="304.2" r="2.2" fill="#1D8155" opacity="0.62"/><circle cx="292.4" cy="334.0" r="2.2" fill="#C4762A" opacity="0.84"/><circle cx="281.5" cy="310.6" r="2.6" fill="#6B9A3D" opacity="0.84"/><circle cx="273.0" cy="357.4" r="2.6" fill="#1D8155" opacity="0.73"/><circle cx="281.0" cy="289.5" r="2.6" fill="#6B9A3D" opacity="0.65"/><circle cx="276.1" cy="316.6" r="2.2" fill="#B23A2E" opacity="0.90"/><circle cx="246.7" cy="307.5" r="2.2" fill="#1D8155" opacity="0.68"/><circle cx="312.0" cy="335.9" r="3.1" fill="#9A6614" opacity="0.67"/><circle cx="232.8" cy="324.8" r="3.6" fill="#1D8155" opacity="0.61"/><circle cx="250.9" cy="333.0" r="2.6" fill="#6B9A3D" opacity="0.77"/><circle cx="290.1" cy="319.6" r="2.2" fill="#6B9A3D" opacity="0.85"/><circle cx="257.1" cy="365.8" r="2.6" fill="#6B9A3D" opacity="0.62"/><circle cx="249.7" cy="314.7" r="2.2" fill="#1D8155" opacity="0.73"/><circle cx="318.7" cy="340.1" r="2.6" fill="#1D8155" opacity="0.60"/><circle cx="272.5" cy="287.9" r="2.2" fill="#6B9A3D" opacity="0.64"/><circle cx="264.0" cy="352.8" r="2.2" fill="#9A6614" opacity="0.75"/><circle cx="296.0" cy="363.5" r="3.1" fill="#6B9A3D" opacity="0.63"/><circle cx="267.5" cy="333.9" r="2.6" fill="#C4762A" opacity="0.91"/><circle cx="301.1" cy="295.8" r="3.6" fill="#1D8155" opacity="0.63"/><circle cx="319.1" cy="320.1" r="2.6" fill="#6B9A3D" opacity="0.61"/><circle cx="254.5" cy="353.6" r="3.1" fill="#9A6614" opacity="0.70"/><circle cx="272.2" cy="351.0" r="2.2" fill="#9A6614" opacity="0.79"/><circle cx="262.2" cy="371.8" r="2.6" fill="#6B9A3D" opacity="0.59"/><circle cx="294.7" cy="307.9" r="2.2" fill="#6B9A3D" opacity="0.75"/><circle cx="288.9" cy="340.6" r="3.1" fill="#9A6614" opacity="0.84"/><circle cx="376" cy="370" r="44" fill="rgba(88,67,108,.05)"/><circle cx="403.3" cy="359.1" r="3.1" fill="#6B9A3D" opacity="0.70"/><circle cx="411.6" cy="374.0" r="2.2" fill="#1D8155" opacity="0.63"/><circle cx="371.8" cy="391.9" r="4.4" fill="#9A6614" opacity="0.77"/><circle cx="388.9" cy="378.0" r="3.1" fill="#C4762A" opacity="0.85"/><circle cx="379.6" cy="331.7" r="4.4" fill="#6B9A3D" opacity="0.61"/><circle cx="354.0" cy="402.6" r="5.4" fill="#1D8155" opacity="0.60"/><circle cx="387.3" cy="391.1" r="3.6" fill="#6B9A3D" opacity="0.75"/><circle cx="375.2" cy="352.9" r="3.1" fill="#C4762A" opacity="0.82"/><circle cx="371.5" cy="343.1" r="5.4" fill="#6B9A3D" opacity="0.72"/><circle cx="402.7" cy="388.7" r="2.6" fill="#1D8155" opacity="0.67"/><circle cx="381.8" cy="360.5" r="5.4" fill="#C4762A" opacity="0.89"/><circle cx="384.6" cy="384.3" r="2.2" fill="#9A6614" opacity="0.83"/><circle cx="396.7" cy="338.4" r="3.6" fill="#1D8155" opacity="0.61"/><circle cx="375.1" cy="383.2" r="2.6" fill="#9A6614" opacity="0.86"/><circle cx="346.4" cy="370.7" r="2.2" fill="#1D8155" opacity="0.70"/><circle cx="376.8" cy="400.5" r="2.2" fill="#6B9A3D" opacity="0.69"/><circle cx="363.9" cy="383.4" r="3.6" fill="#9A6614" opacity="0.82"/><circle cx="395.8" cy="349.3" r="2.2" fill="#1D8155" opacity="0.71"/><circle cx="345.6" cy="355.6" r="2.6" fill="#1D8155" opacity="0.66"/><circle cx="393.6" cy="402.7" r="2.2" fill="#1D8155" opacity="0.62"/><circle cx="366.5" cy="360.6" r="4.4" fill="#C4762A" opacity="0.86"/><circle cx="397.1" cy="380.4" r="2.2" fill="#1D8155" opacity="0.76"/><circle cx="351.2" cy="347.2" r="3.1" fill="#1D8155" opacity="0.66"/><circle cx="386.2" cy="344.5" r="2.6" fill="#1D8155" opacity="0.72"/><circle cx="353.0" cy="366.3" r="2.2" fill="#9A6614" opacity="0.76"/><circle cx="359.7" cy="339.1" r="2.6" fill="#1D8155" opacity="0.64"/><circle cx="404.3" cy="370.8" r="2.2" fill="#9A6614" opacity="0.71"/><circle cx="405.3" cy="343.2" r="3.1" fill="#1D8155" opacity="0.59"/><circle cx="373.0" cy="372.4" r="2.6" fill="#9A6614" opacity="0.96"/><circle cx="406.5" cy="382.0" r="3.6" fill="#1D8155" opacity="0.67"/><circle cx="394.2" cy="386.9" r="3.1" fill="#9A6614" opacity="0.75"/><circle cx="337.9" cy="380.1" r="3.6" fill="#1D8155" opacity="0.60"/><circle cx="412.6" cy="358.9" r="2.6" fill="#1D8155" opacity="0.61"/><circle cx="387.1" cy="404.6" r="2.2" fill="#1D8155" opacity="0.63"/><circle cx="348.6" cy="389.1" r="3.1" fill="#1D8155" opacity="0.66"/><circle cx="395.5" cy="360.4" r="2.6" fill="#6B9A3D" opacity="0.78"/><circle cx="337.6" cy="369.9" r="2.2" fill="#6B9A3D" opacity="0.61"/><circle cx="388.2" cy="350.4" r="2.2" fill="#6B9A3D" opacity="0.76"/><circle cx="365.3" cy="407.7" r="3.1" fill="#1D8155" opacity="0.60"/><circle cx="381.6" cy="407.3" r="2.2" fill="#6B9A3D" opacity="0.61"/><circle cx="363.6" cy="373.6" r="2.2" fill="#C4762A" opacity="0.87"/><circle cx="414.6" cy="367.7" r="3.1" fill="#1D8155" opacity="0.60"/><circle cx="354.6" cy="357.3" r="2.2" fill="#9A6614" opacity="0.75"/><circle cx="387.7" cy="368.0" r="2.2" fill="#9A6614" opacity="0.88"/><circle cx="394.3" cy="370.8" r="3.1" fill="#9A6614" opacity="0.81"/><circle cx="406.0" cy="352.6" r="2.2" fill="#6B9A3D" opacity="0.65"/><circle cx="399.9" cy="396.8" r="3.1" fill="#1D8155" opacity="0.63"/><circle cx="380.0" cy="376.5" r="3.1" fill="#C4762A" opacity="0.92"/><circle cx="370.7" cy="378.8" r="2.2" fill="#B23A2E" opacity="0.90"/><circle cx="364.1" cy="350.8" r="3.1" fill="#6B9A3D" opacity="0.77"/><circle cx="342.3" cy="362.5" r="3.1" fill="#1D8155" opacity="0.65"/><circle cx="387.1" cy="334.7" r="2.2" fill="#6B9A3D" opacity="0.62"/><circle cx="367.7" cy="331.3" r="3.1" fill="#1D8155" opacity="0.59"/><circle cx="353.7" cy="373.3" r="2.2" fill="#6B9A3D" opacity="0.77"/><circle cx="347.0" cy="377.8" r="2.2" fill="#1D8155" opacity="0.69"/><circle cx="361.0" cy="389.9" r="2.2" fill="#9A6614" opacity="0.75"/><circle cx="383.9" cy="397.7" r="2.2" fill="#9A6614" opacity="0.71"/><circle cx="364.2" cy="398.2" r="2.2" fill="#6B9A3D" opacity="0.69"/><circle cx="373.8" cy="405.5" r="2.2" fill="#1D8155" opacity="0.64"/><circle cx="450" cy="318" r="32" fill="rgba(88,67,108,.05)"/><circle cx="472.5" cy="330.2" r="2.2" fill="#1D8155" opacity="0.64"/><circle cx="463.4" cy="312.4" r="2.6" fill="#6B9A3D" opacity="0.80"/><circle cx="424.1" cy="329.8" r="2.2" fill="#1D8155" opacity="0.60"/><circle cx="439.2" cy="323.2" r="2.2" fill="#9A6614" opacity="0.83"/><circle cx="475.2" cy="316.4" r="5.4" fill="#1D8155" opacity="0.65"/><circle cx="445.0" cy="313.0" r="4.4" fill="#9A6614" opacity="0.90"/><circle cx="445.6" cy="297.4" r="2.6" fill="#1D8155" opacity="0.70"/><circle cx="453.8" cy="332.5" r="3.6" fill="#6B9A3D" opacity="0.79"/><circle cx="447.2" cy="323.0" r="3.6" fill="#C4762A" opacity="0.92"/><circle cx="462.1" cy="322.5" r="3.1" fill="#C4762A" opacity="0.82"/><circle cx="431.8" cy="310.4" r="2.2" fill="#9A6614" opacity="0.72"/><circle cx="458.0" cy="299.4" r="2.6" fill="#1D8155" opacity="0.72"/><circle cx="455.2" cy="346.2" r="3.1" fill="#1D8155" opacity="0.60"/><circle cx="455.5" cy="307.5" r="3.6" fill="#9A6614" opacity="0.83"/><circle cx="441.9" cy="305.4" r="2.2" fill="#6B9A3D" opacity="0.79"/><circle cx="432.4" cy="334.4" r="2.2" fill="#1D8155" opacity="0.66"/><circle cx="433.0" cy="319.8" r="3.1" fill="#6B9A3D" opacity="0.76"/><circle cx="422.7" cy="307.4" r="4.4" fill="#1D8155" opacity="0.59"/><circle cx="425.0" cy="320.8" r="2.6" fill="#6B9A3D" opacity="0.65"/><circle cx="466.1" cy="299.1" r="3.6" fill="#1D8155" opacity="0.65"/><circle cx="450.7" cy="292.8" r="2.6" fill="#6B9A3D" opacity="0.65"/><circle cx="436.7" cy="341.7" r="2.2" fill="#1D8155" opacity="0.62"/><circle cx="436.0" cy="329.6" r="2.2" fill="#9A6614" opacity="0.75"/><circle cx="453.1" cy="316.1" r="2.6" fill="#9A6614" opacity="0.95"/><circle cx="431.6" cy="299.6" r="3.6" fill="#1D8155" opacity="0.63"/><circle cx="442.1" cy="332.9" r="2.2" fill="#6B9A3D" opacity="0.76"/><circle cx="448.7" cy="341.6" r="2.6" fill="#6B9A3D" opacity="0.67"/><circle cx="465.9" cy="339.2" r="2.2" fill="#1D8155" opacity="0.63"/><circle cx="462.2" cy="331.2" r="2.2" fill="#1D8155" opacity="0.75"/><circle cx="474.2" cy="303.1" r="2.6" fill="#1D8155" opacity="0.60"/><circle cx="450.5" cy="302.5" r="2.2" fill="#9A6614" opacity="0.78"/><circle cx="443.4" cy="346.2" r="3.1" fill="#1D8155" opacity="0.59"/><circle cx="438.7" cy="292.6" r="2.6" fill="#1D8155" opacity="0.61"/><circle cx="467.6" cy="327.1" r="2.2" fill="#9A6614" opacity="0.72"/><circle cx="461.7" cy="291.2" r="3.6" fill="#1D8155" opacity="0.59"/><circle cx="459.5" cy="340.9" r="2.2" fill="#1D8155" opacity="0.65"/><circle cx="420.8" cy="315.7" r="2.6" fill="#1D8155" opacity="0.59"/><circle cx="469.2" cy="308.4" r="3.1" fill="#9A6614" opacity="0.70"/><circle cx="208" cy="406" r="30" fill="rgba(88,67,108,.05)"/><circle cx="215.5" cy="430.4" r="3.6" fill="#1D8155" opacity="0.62"/><circle cx="194.3" cy="402.3" r="2.2" fill="#C4762A" opacity="0.79"/><circle cx="230.9" cy="399.5" r="4.4" fill="#1D8155" opacity="0.64"/><circle cx="206.4" cy="403.2" r="5.4" fill="#B23A2E" opacity="0.95"/><circle cx="226.9" cy="421.4" r="3.6" fill="#6B9A3D" opacity="0.63"/><circle cx="220.2" cy="385.5" r="3.1" fill="#6B9A3D" opacity="0.64"/><circle cx="194.7" cy="393.1" r="4.4" fill="#1D8155" opacity="0.72"/><circle cx="187.7" cy="410.6" r="4.4" fill="#6B9A3D" opacity="0.69"/><circle cx="201.9" cy="431.3" r="2.2" fill="#1D8155" opacity="0.61"/><circle cx="188.1" cy="422.5" r="4.4" fill="#6B9A3D" opacity="0.61"/><circle cx="219.4" cy="416.8" r="2.2" fill="#9A6614" opacity="0.76"/><circle cx="200.7" cy="420.4" r="2.2" fill="#6B9A3D" opacity="0.76"/><circle cx="212.8" cy="420.4" r="2.2" fill="#9A6614" opacity="0.77"/><circle cx="201.7" cy="411.5" r="2.2" fill="#9A6614" opacity="0.87"/><circle cx="218.4" cy="401.3" r="3.1" fill="#C4762A" opacity="0.83"/><circle cx="206.6" cy="386.7" r="2.2" fill="#6B9A3D" opacity="0.71"/><circle cx="211.4" cy="394.7" r="2.6" fill="#C4762A" opacity="0.82"/><circle cx="182.8" cy="402.1" r="2.6" fill="#1D8155" opacity="0.62"/><circle cx="212.6" cy="412.1" r="3.1" fill="#C4762A" opacity="0.89"/><circle cx="232.5" cy="416.4" r="2.2" fill="#1D8155" opacity="0.60"/><circle cx="221.5" cy="393.6" r="3.1" fill="#6B9A3D" opacity="0.73"/><circle cx="199.4" cy="384.2" r="2.6" fill="#1D8155" opacity="0.65"/><circle cx="208.0" cy="428.0" r="2.6" fill="#6B9A3D" opacity="0.67"/><circle cx="211.8" cy="379.2" r="4.4" fill="#1D8155" opacity="0.59"/><circle cx="206.9" cy="420.1" r="2.2" fill="#6B9A3D" opacity="0.79"/><circle cx="229.8" cy="390.6" r="2.6" fill="#1D8155" opacity="0.60"/><circle cx="195.7" cy="415.7" r="2.6" fill="#6B9A3D" opacity="0.77"/><circle cx="228.3" cy="408.5" r="3.1" fill="#9A6614" opacity="0.69"/><circle cx="189.0" cy="387.5" r="2.2" fill="#1D8155" opacity="0.60"/></g><text x="70" y="424" class="cap">2,802 files scored</text><text x="526" y="280" class="cap">OPEN FINDINGS</text><rect x="526" y="300" width="236" height="24" rx="6" fill="#F4EAE1"/><rect x="526" y="300" width="3" height="24" rx="1.5" fill="#B23A2E"/><text x="539" y="316" class="find">CHANGE ENTROPY</text><text x="752" y="316" class="find" fill="#B23A2E" text-anchor="end" font-weight="700">−3.0</text><rect x="526" y="328" width="236" height="24" rx="6" fill="#F4EAE1"/><rect x="526" y="328" width="3" height="24" rx="1.5" fill="#B23A2E"/><text x="539" y="344" class="find">NESTED COMPLEXITY</text><text x="752" y="344" class="find" fill="#B23A2E" text-anchor="end" font-weight="700">−2.4</text><rect x="526" y="356" width="236" height="24" rx="6" fill="#F4EAE1"/><rect x="526" y="356" width="3" height="24" rx="1.5" fill="#9A6614"/><text x="539" y="372" class="find">CHURN RISK</text><text x="752" y="372" class="find" fill="#9A6614" text-anchor="end" font-weight="700">−1.8</text><rect x="526" y="384" width="236" height="24" rx="6" fill="#F4EAE1"/><rect x="526" y="384" width="3" height="24" rx="1.5" fill="#9A6614"/><text x="539" y="400" class="find">CO-CHANGE SCATTER</text><text x="752" y="400" class="find" fill="#9A6614" text-anchor="end" font-weight="700">−1.5</text><rect x="526" y="412" width="236" height="24" rx="6" fill="#F4EAE1"/><rect x="526" y="412" width="3" height="24" rx="1.5" fill="#9A6614"/><text x="539" y="428" class="find">UNTESTED HOTSPOT</text><text x="752" y="428" class="find" fill="#9A6614" text-anchor="end" font-weight="700">−1.2</text><rect x="802" y="104" width="380" height="352" rx="14" fill="#FFFFFF" stroke="rgba(88,67,108,.12)" stroke-width="1"/><text x="824" y="135" class="lbl">DEPENDENCY GRAPH</text><text x="1160" y="135" class="kick" text-anchor="end">16 languages</text><rect x="820" y="150" width="344" height="240" rx="9" fill="#F4EAE1"/><clipPath id="gclip"><rect x="820" y="150" width="344" height="240" rx="9"/></clipPath><g clip-path="url(#gclip)"><line x1="944.0" y1="201.8" x2="912" y2="224" stroke="rgba(88,67,108,.15)" stroke-width="1"/><line x1="907.3" y1="185.9" x2="912" y2="224" stroke="rgba(88,67,108,.15)" stroke-width="1"/><line x1="913.1" y1="193.3" x2="912" y2="224" stroke="rgba(88,67,108,.15)" stroke-width="1"/><line x1="948.1" y1="258.8" x2="912" y2="224" stroke="rgba(88,67,108,.15)" stroke-width="1"/><line x1="901.9" y1="182.8" x2="912" y2="224" stroke="rgba(88,67,108,.15)" stroke-width="1"/><line x1="895.4" y1="193.7" x2="912" y2="224" stroke="rgba(88,67,108,.15)" stroke-width="1"/><line x1="888.8" y1="197.7" x2="912" y2="224" stroke="rgba(88,67,108,.15)" stroke-width="1"/><line x1="929.0" y1="238.4" x2="912" y2="224" stroke="rgba(88,67,108,.15)" stroke-width="1"/><line x1="951.9" y1="214.2" x2="912" y2="224" stroke="rgba(88,67,108,.15)" stroke-width="1"/><line x1="1040.2" y1="228.2" x2="1052" y2="212" stroke="rgba(88,67,108,.15)" stroke-width="1"/><line x1="1048.4" y1="190.3" x2="1052" y2="212" stroke="rgba(88,67,108,.15)" stroke-width="1"/><line x1="1068.5" y1="249.3" x2="1052" y2="212" stroke="rgba(88,67,108,.15)" stroke-width="1"/><line x1="1013.8" y1="237.1" x2="1052" y2="212" stroke="rgba(88,67,108,.15)" stroke-width="1"/><line x1="1009.6" y1="237.8" x2="1052" y2="212" stroke="rgba(88,67,108,.15)" stroke-width="1"/><line x1="1065.3" y1="185.2" x2="1052" y2="212" stroke="rgba(88,67,108,.15)" stroke-width="1"/><line x1="1017.4" y1="194.6" x2="1052" y2="212" stroke="rgba(88,67,108,.15)" stroke-width="1"/><line x1="1107.4" y1="209.0" x2="1052" y2="212" stroke="rgba(88,67,108,.15)" stroke-width="1"/><line x1="1061.0" y1="174.5" x2="1052" y2="212" stroke="rgba(88,67,108,.15)" stroke-width="1"/><line x1="944.8" y1="323.3" x2="982" y2="322" stroke="rgba(88,67,108,.15)" stroke-width="1"/><line x1="955.1" y1="338.1" x2="982" y2="322" stroke="rgba(88,67,108,.15)" stroke-width="1"/><line x1="1001.7" y1="341.5" x2="982" y2="322" stroke="rgba(88,67,108,.15)" stroke-width="1"/><line x1="1022.9" y1="322.0" x2="982" y2="322" stroke="rgba(88,67,108,.15)" stroke-width="1"/><line x1="961.2" y1="356.5" x2="982" y2="322" stroke="rgba(88,67,108,.15)" stroke-width="1"/><line x1="942.1" y1="301.8" x2="982" y2="322" stroke="rgba(88,67,108,.15)" stroke-width="1"/><line x1="943.8" y1="346.7" x2="982" y2="322" stroke="rgba(88,67,108,.15)" stroke-width="1"/><line x1="983.8" y1="296.5" x2="982" y2="322" stroke="rgba(88,67,108,.15)" stroke-width="1"/><line x1="949.1" y1="320.7" x2="982" y2="322" stroke="rgba(88,67,108,.15)" stroke-width="1"/><line x1="1131.2" y1="323.1" x2="1098" y2="310" stroke="rgba(88,67,108,.15)" stroke-width="1"/><line x1="1085.5" y1="327.7" x2="1098" y2="310" stroke="rgba(88,67,108,.15)" stroke-width="1"/><line x1="1106.7" y1="286.0" x2="1098" y2="310" stroke="rgba(88,67,108,.15)" stroke-width="1"/><line x1="1152.2" y1="302.0" x2="1098" y2="310" stroke="rgba(88,67,108,.15)" stroke-width="1"/><line x1="1068.9" y1="335.5" x2="1098" y2="310" stroke="rgba(88,67,108,.15)" stroke-width="1"/><line x1="1148.6" y1="288.2" x2="1098" y2="310" stroke="rgba(88,67,108,.15)" stroke-width="1"/><line x1="1064.8" y1="323.3" x2="1098" y2="310" stroke="rgba(88,67,108,.15)" stroke-width="1"/><line x1="1046.2" y1="319.4" x2="1098" y2="310" stroke="rgba(88,67,108,.15)" stroke-width="1"/><line x1="1137.2" y1="290.0" x2="1098" y2="310" stroke="rgba(88,67,108,.15)" stroke-width="1"/><line x1="912" y1="224" x2="1052" y2="212" stroke="rgba(88,67,108,.30)" stroke-width="1.6"/><line x1="912" y1="224" x2="982" y2="322" stroke="rgba(88,67,108,.30)" stroke-width="1.6"/><line x1="912" y1="224" x2="1098" y2="310" stroke="rgba(88,67,108,.30)" stroke-width="1.6"/><line x1="1052" y1="212" x2="982" y2="322" stroke="rgba(88,67,108,.30)" stroke-width="1.6"/><line x1="982" y1="322" x2="1098" y2="310" stroke="rgba(88,67,108,.30)" stroke-width="1.6"/><circle cx="912.0" cy="224.0" r="9.0" fill="#C0641A" stroke="#FFFFFF" stroke-width="1.5"/><circle cx="944.0" cy="201.8" r="4.0" fill="#F7A94D"/><circle cx="907.3" cy="185.9" r="3.3" fill="#F7A94D"/><circle cx="913.1" cy="193.3" r="2.8" fill="#F7A94D"/><circle cx="948.1" cy="258.8" r="3.8" fill="#F7A94D"/><circle cx="901.9" cy="182.8" r="3.1" fill="#F7A94D"/><circle cx="895.4" cy="193.7" r="3.4" fill="#F7A94D"/><circle cx="888.8" cy="197.7" r="3.3" fill="#F7A94D"/><circle cx="929.0" cy="238.4" r="3.8" fill="#F7A94D"/><circle cx="951.9" cy="214.2" r="3.6" fill="#F7A94D"/><circle cx="1052.0" cy="212.0" r="9.0" fill="#58436C" stroke="#FFFFFF" stroke-width="1.5"/><circle cx="1040.2" cy="228.2" r="3.4" fill="#826AA0"/><circle cx="1048.4" cy="190.3" r="3.5" fill="#826AA0"/><circle cx="1068.5" cy="249.3" r="3.5" fill="#826AA0"/><circle cx="1013.8" cy="237.1" r="3.7" fill="#826AA0"/><circle cx="1009.6" cy="237.8" r="3.8" fill="#826AA0"/><circle cx="1065.3" cy="185.2" r="4.1" fill="#826AA0"/><circle cx="1017.4" cy="194.6" r="4.2" fill="#826AA0"/><circle cx="1107.4" cy="209.0" r="3.7" fill="#826AA0"/><circle cx="1061.0" cy="174.5" r="3.3" fill="#826AA0"/><circle cx="982.0" cy="322.0" r="9.0" fill="#B23A2E" stroke="#FFFFFF" stroke-width="1.5"/><circle cx="944.8" cy="323.3" r="4.2" fill="#CF6A55"/><circle cx="955.1" cy="338.1" r="4.0" fill="#CF6A55"/><circle cx="1001.7" cy="341.5" r="3.2" fill="#CF6A55"/><circle cx="1022.9" cy="322.0" r="3.0" fill="#CF6A55"/><circle cx="961.2" cy="356.5" r="4.0" fill="#CF6A55"/><circle cx="942.1" cy="301.8" r="3.5" fill="#CF6A55"/><circle cx="943.8" cy="346.7" r="3.7" fill="#CF6A55"/><circle cx="983.8" cy="296.5" r="2.9" fill="#CF6A55"/><circle cx="949.1" cy="320.7" r="2.6" fill="#CF6A55"/><circle cx="1098.0" cy="310.0" r="9.0" fill="#6B7A3D" stroke="#FFFFFF" stroke-width="1.5"/><circle cx="1131.2" cy="323.1" r="3.2" fill="#90A05E"/><circle cx="1085.5" cy="327.7" r="4.3" fill="#90A05E"/><circle cx="1106.7" cy="286.0" r="2.9" fill="#90A05E"/><circle cx="1152.2" cy="302.0" r="3.9" fill="#90A05E"/><circle cx="1068.9" cy="335.5" r="2.8" fill="#90A05E"/><circle cx="1148.6" cy="288.2" r="4.2" fill="#90A05E"/><circle cx="1064.8" cy="323.3" r="3.6" fill="#90A05E"/><circle cx="1046.2" cy="319.4" r="4.0" fill="#90A05E"/><circle cx="1137.2" cy="290.0" r="2.9" fill="#90A05E"/></g><text x="832" y="380" class="cap">28,126 nodes · 11 communities</text><text x="824" y="414" class="out">Who calls this? What breaks if</text><text x="824" y="434" class="out">I change it?</text><rect x="1200" y="104" width="364" height="352" rx="14" fill="#FFFFFF" stroke="rgba(88,67,108,.12)" stroke-width="1"/><text x="1222" y="135" class="lbl">GIT HISTORY</text><text x="1542" y="135" class="kick" text-anchor="end">1,050 commits</text><rect x="1218" y="150" width="328" height="240" rx="9" fill="#F4EAE1"/><rect x="1230.0" y="312.2" width="16.0" height="13.8" rx="2.5" fill="#1D8155" opacity="0.82"/><rect x="1249.0" y="300.7" width="16.0" height="25.3" rx="2.5" fill="#1D8155" opacity="0.82"/><rect x="1268.0" y="282.3" width="16.0" height="43.7" rx="2.5" fill="#1D8155" opacity="0.82"/><rect x="1287.0" y="261.6" width="16.0" height="64.4" rx="2.5" fill="#1D8155" opacity="0.82"/><rect x="1306.0" y="231.7" width="16.0" height="94.3" rx="2.5" fill="#1D8155" opacity="0.82"/><rect x="1325.0" y="206.4" width="16.0" height="119.6" rx="2.5" fill="#1D8155" opacity="0.82"/><rect x="1344.0" y="181.1" width="16.0" height="144.9" rx="2.5" fill="#9A6614" opacity="0.82"/><rect x="1363.0" y="192.6" width="16.0" height="133.4" rx="2.5" fill="#9A6614" opacity="0.82"/><rect x="1382.0" y="217.9" width="16.0" height="108.1" rx="2.5" fill="#9A6614" opacity="0.82"/><rect x="1401.0" y="243.2" width="16.0" height="82.8" rx="2.5" fill="#9A6614" opacity="0.82"/><rect x="1420.0" y="263.9" width="16.0" height="62.1" rx="2.5" fill="#9A6614" opacity="0.82"/><rect x="1439.0" y="282.3" width="16.0" height="43.7" rx="2.5" fill="#B23A2E" opacity="0.82"/><rect x="1458.0" y="296.1" width="16.0" height="29.9" rx="2.5" fill="#B23A2E" opacity="0.82"/><rect x="1477.0" y="305.3" width="16.0" height="20.7" rx="2.5" fill="#B23A2E" opacity="0.82"/><rect x="1496.0" y="312.2" width="16.0" height="13.8" rx="2.5" fill="#B23A2E" opacity="0.82"/><rect x="1515.0" y="316.8" width="16.0" height="9.2" rx="2.5" fill="#B23A2E" opacity="0.82"/><line x1="1230" y1="327" x2="1534" y2="327" stroke="rgba(88,67,108,.20)"/><text x="1230" y="348" class="cap">low risk</text><text x="1534" y="348" class="cap" text-anchor="end">high risk</text><text x="1230" y="176" class="cap">CHANGE-RISK DISTRIBUTION</text><text x="1230" y="380" class="cap">hotspots · ownership · co-change · bus factor</text><text x="1222" y="414" class="out">Which of these 40 files is</text><text x="1222" y="434" class="out">actually dangerous?</text><rect x="36" y="474" width="496" height="302" rx="14" fill="#FFFFFF" stroke="rgba(88,67,108,.12)" stroke-width="1"/><text x="58" y="505" class="lbl">GENERATED DOCS</text><text x="510" y="505" class="kick" text-anchor="end">1,274 pages</text><rect x="54" y="520" width="460" height="176" rx="9" fill="#F4EAE1"/><rect x="70" y="536" width="265" height="11" rx="3" fill="#58436C" opacity=".85"/><rect x="70" y="562" width="394" height="6" rx="3" fill="rgba(88,67,108,.26)"/><rect x="70" y="578" width="368" height="6" rx="3" fill="rgba(88,67,108,.26)"/><rect x="70" y="594" width="304" height="6" rx="3" fill="rgba(88,67,108,.26)"/><rect x="70" y="607" width="188" height="9" rx="3" fill="#A16215" opacity=".55"/><rect x="70" y="630" width="377" height="6" rx="3" fill="rgba(88,67,108,.26)"/><rect x="70" y="646" width="338" height="6" rx="3" fill="rgba(88,67,108,.26)"/><rect x="70" y="662" width="235" height="6" rx="3" fill="rgba(88,67,108,.26)"/><rect x="416" y="530" width="86" height="20" rx="10" fill="rgba(29,129,85,.12)"/><text x="459" y="544" class="pill" fill="#1D8155" text-anchor="middle">fresh · v6</text><text x="58" y="730" class="out">Why does auth work this way?</text><text x="58" y="751" class="outsub">Answered from the wiki, not from 12 file reads.</text><rect x="550" y="474" width="496" height="302" rx="14" fill="#FFFFFF" stroke="rgba(88,67,108,.12)" stroke-width="1"/><text x="572" y="505" class="lbl">ARCHITECTURAL DECISIONS</text><text x="1024" y="505" class="kick" text-anchor="end">mined from 8 sources</text><rect x="568" y="520" width="460" height="176" rx="9" fill="#F4EAE1"/><text x="584" y="548" class="dec">Rebuild the KG skeleton only when the</text><text x="584" y="568" class="dec">graph fingerprint changes</text><rect x="584" y="582" width="392" height="1" fill="rgba(88,67,108,.15)"/><rect x="584" y="596" width="101" height="21" rx="10.5" fill="#FFFFFF" stroke="rgba(88,67,108,.18)"/><text x="635" y="610" class="pill" fill="#1D8155" text-anchor="middle">Verified quote</text><rect x="693.4" y="596" width="108" height="21" rx="10.5" fill="#FFFFFF" stroke="rgba(88,67,108,.18)"/><text x="747" y="610" class="pill" fill="#58436C" text-anchor="middle">supersedes #418</text><rect x="809.4" y="596" width="115" height="21" rx="10.5" fill="#FFFFFF" stroke="rgba(88,67,108,.18)"/><text x="867" y="610" class="pill" fill="#A16215" text-anchor="middle">governs 24 files</text><text x="584" y="660" class="cap">Delivered back at session start, and the moment</text><text x="584" y="676" class="cap">your agent edits a file the decision governs.</text><text x="572" y="730" class="out">Don’t re-litigate what we settled.</text><text x="572" y="751" class="outsub">Captured nowhere else.</text><rect x="1064" y="474" width="500" height="302" rx="14" fill="#FFFFFF" stroke="rgba(88,67,108,.12)" stroke-width="1"/><text x="1086" y="505" class="lbl">SERVED TO YOUR AGENT</text><text x="1542" y="505" class="kick" text-anchor="end">10 MCP tools</text><rect x="1084" y="528" width="95" height="24" rx="6" fill="#F4EAE1" stroke="rgba(88,67,108,.12)"/><text x="1131" y="544" class="mono" text-anchor="middle">get_overview</text><rect x="1187" y="528" width="81" height="24" rx="6" fill="#F4EAE1" stroke="rgba(88,67,108,.12)"/><text x="1227" y="544" class="mono" text-anchor="middle">get_answer</text><rect x="1276" y="528" width="88" height="24" rx="6" fill="#F4EAE1" stroke="rgba(88,67,108,.12)"/><text x="1320" y="544" class="mono" text-anchor="middle">get_context</text><rect x="1372" y="528" width="81" height="24" rx="6" fill="#F4EAE1" stroke="rgba(88,67,108,.12)"/><text x="1412" y="544" class="mono" text-anchor="middle">get_symbol</text><rect x="1084" y="558" width="116" height="24" rx="6" fill="#F4EAE1" stroke="rgba(88,67,108,.12)"/><text x="1142" y="574" class="mono" text-anchor="middle">search_codebase</text><rect x="1208" y="558" width="67" height="24" rx="6" fill="#F4EAE1" stroke="rgba(88,67,108,.12)"/><text x="1241" y="574" class="mono" text-anchor="middle">get_risk</text><rect x="1283" y="558" width="116" height="24" rx="6" fill="#F4EAE1" stroke="rgba(88,67,108,.12)"/><text x="1340" y="574" class="mono" text-anchor="middle">get_change_risk</text><rect x="1406" y="558" width="60" height="24" rx="6" fill="#F4EAE1" stroke="rgba(88,67,108,.12)"/><text x="1436" y="574" class="mono" text-anchor="middle">get_why</text><rect x="1084" y="588" width="102" height="24" rx="6" fill="#F4EAE1" stroke="rgba(88,67,108,.12)"/><text x="1135" y="604" class="mono" text-anchor="middle">get_dead_code</text><rect x="1194" y="588" width="81" height="24" rx="6" fill="#F4EAE1" stroke="rgba(88,67,108,.12)"/><text x="1234" y="604" class="mono" text-anchor="middle">get_health</text><text x="1084" y="636" class="cap">Every response carries index age, indexed commit, and a staleness warning.</text><rect x="1084" y="650" width="460" height="46" rx="9" fill="rgba(245,149,32,.10)" stroke="rgba(245,149,32,.28)"/><text x="1098" y="670" class="chipval" fill="#A16215" font-size="19">−96%<tspan class="chipunit" fill="#5E5360"> context tokens</tspan></text><text x="1098" y="688" class="cap">−70% tool calls · −89% file reads · fewer round-trips per answer</text><text x="1086" y="730" class="out">Claude Code, Codex, Cursor, VS Code.</text><text x="1086" y="751" class="outsub">Plus hooks, so it arrives unasked.</text></svg>\ No newline at end of filediff --git a/README.md b/README.mdindex 9baf73e64..f15db1dfc 100644--- a/README.md+++ b/README.md@@ -43,12 +43,14 @@### Your AI agent burns most of its budget rediscovering your codebase. Index it once, and it never has to again.-### −97% tokens to load a commit's context--<sub>13,984 tokens of raw file reads become 393, measured over 30 commits, raw CSV published.-We are also the only tool in this category that publishes what that is worth inside a real-agent loop, where the honest number is −16% of the agent's own output tokens (n=15, p=0.035)-and the only significant result in the field. <a href="docs/BENCHMARKS.md"><strong>See how the others did →</strong></a><br />+### One index. A dependency graph, a generated wiki, mined git history, architectural decisions, and a defect-validated health score.++<sub><strong>−31.6%</strong> of an agent's own output tokens across 43 questions (p<0.0001), reached in+3.8 tool calls where a bare agent needed 7.2 · loading one commit's context costs 393 tokens+instead of 13,984 raw · defect-risk <strong>ROC AUC 0.737</strong> across 21 repos and 9 languages, scored+leakage-free, and 0.76–0.78 on a public dataset that played no part in calibration · every+layer computed with <strong>zero LLM calls</strong>. We publish the rows we lose.+<a href="docs/BENCHMARKS.md"><strong>See how the others did →</strong></a><br />Free and self-hosted, runs on your machine, and the first index needs no API key.</sub><picture>@@ -86,9 +88,12 @@ repository, no API key and nothing uploaded.</sub>Because the exploration work is already done, that phase mostly disappears. Loadingone commit's context through `get_context` costs **393 tokens instead of 13,984**-raw, 35.6x fewer. In a measured agent loop the saving is more modest and more-honest: **-15.9% of the agent's own output tokens**, and it grows with how much-of the codebase the task touches.+raw, 35.6x fewer. In a measured agent loop, across 43 questions on `django/django`,+that is worth **-31.6% of the agent's own output tokens** (p<0.0001), reached in+**3.8 tool calls against a bare agent's 7.2** — roughly one answered question+replacing six greps. The saving grows with how much of the codebase the task+touches. CodeGraph is a genuine second here at -24.4%: we lead a field in which+more than one tool works.**And it arrives without being asked.** Optional [hooks](docs/agent/HOOKS.md) pushcontext into the session at the moment it matters: the governing architectural@@ -112,11 +117,11 @@ queryable from the CLI, the MCP tools, and the local dashboard.| Layer | What it gives you | Edge ||---|---|---|-| **◈ Graph** | Dependency graph across 17 languages · file + symbol nodes · 3-tier call resolution · Leiden communities · PageRank and execution flows · framework-aware route→handler edges | A real graph most tools never build |-| **◈ Git** | Hotspots (churn × complexity) · ownership % · co-change pairs (hidden coupling) · bus factor · which files actually get bug-fixed, and how recently | Behavioural signals static analysis cannot see |+| **◈ Graph** | Dependency graph across 18 languages · file + symbol nodes · 3-tier call resolution · Leiden communities · PageRank and execution flows · route→handler edges across 22 frameworks | A real graph most tools never build |+| **◈ Git** | Hotspots (decayed churn + activity floors) · ownership % · co-change pairs (hidden coupling) · bus factor · which files actually get bug-fixed, and how recently | Behavioural signals static analysis cannot see || **◈ Docs** | A generated wiki page per module and file · rebuilt incrementally every commit · freshness and confidence scoring · hybrid search (full-text + vector) · selectable style and output language | Stays current instead of rotting |-| **◈ Decisions** | Architectural decisions mined from eight sources, evidence-backed, linked to the graph nodes they govern, tracked for staleness | **★ Captured nowhere else** |-| **★ Code health** | **25 deterministic markers**, 1 to 10 per file · three signals: defect risk · maintainability · performance · coverage ingestion · concrete refactoring plans (Extract Class / Helper, Move Method, Break Cycle, Split File) · **zero LLM, under 30s** | **★ Defect-validated, with the fix attached** |+| **◈ Decisions** | Architectural decisions mined from five sources, evidence-backed, each traced to a verbatim source span and stamped exact / fuzzy / unverified | **★ Captured nowhere else** |+| **★ Code health** | **49 deterministic detectors**, of which only 26 may move the number · 1 to 10 per file · three signals: defect risk · maintainability · performance · concrete refactoring plans (Extract Class / Method / Helper, Move Method, Break Cycle, Split File) · **zero LLM, under 30s** | **★ Defect-validated, with the fix attached** |**The whole wiki is generated with no LLM, then upgraded to model-written prose ondemand.** `repowise init --no-prose` builds the graph, git, decision and health@@ -227,24 +232,32 @@ A score that says *"this file is risky"* is where most tools stop. repowise scorevery file, locates where the risk concentrates, and then names the specific fix.<div align="center">-<img src=".github/assets/health-loop.svg" alt="repowise code-health loop: 25 deterministic markers fan into three signals, the graph and git history locate where risk concentrates, and refactoring intelligence emits concrete plans your agent executes" width="100%" />+<img src=".github/assets/health-loop.svg" alt="repowise code-health loop: deterministic markers fan into three signals, the graph and git history locate where risk concentrates, and refactoring intelligence emits concrete plans your agent executes" width="100%" /></div>-Every file is scored 1-10 from **25 deterministic markers** (McCabe complexity, brain+Every file is scored 1-10 by **49 deterministic detectors** (McCabe complexity, brainmethods, LCOM4 cohesion, god classes, native Rabin-Karp clone detection, untestedhotspots, change entropy, prior-defect history and more), split into three lenses:-**defect risk**, **maintainability**, and **performance** (static N+1 and I/O-in-loop-risk traced *across* files through the call graph, where file-local linters found 0 of-the cross-function cases repowise surfaced 557 of).+**defect risk**, **maintainability**, and **performance** — static N+1 and I/O-in-loop+risk traced *across* files through the call graph, where file-local linters found **0**+of the cross-function cases and repowise surfaced ~90. Only **26** of the 49 are+permitted to move the defect number, because that is the number carrying published+accuracy claims.> **Zero LLM calls, zero cloud, zero new runtime dependencies.** Pure Python over-> tree-sitter and git data, **under 30 seconds** on a 3,000-file repo, with marker-> weights **calibrated against a real defect corpus, not hand-tuned**.+> tree-sitter and git data, **under 30 seconds** on a 3,000-file repo — a budget+> enforced by a CI test, not an estimate. Marker weights are **calibrated against a+> real defect corpus, not hand-tuned**: every file scored at a commit preceding the+> bug window so nothing leaks backward, and an L2-logistic fit with file size as an+> explicit control, so a marker only earns weight for defect lift *beyond* being big.+> Only the learned constants ship.**It proves itself on your repo, not just on a benchmark.** After every index,repowise checks its own flags against your git history and reports what it found:-*"17 of the 20 lowest-health files had a bug fix in the last 6 months, 3.6x the 23%-baseline."* If that number is bad on your codebase, you will see it.+*"16 of the 20 lowest-health files had a bug fix in the last 6 months, 3.3x the 24%+baseline."* If that number is bad on your codebase, you will see it. (It is an+association on your indexed history, not a forward prediction — the leakage-free+version is [in the benchmarks](docs/BENCHMARKS.md#5-code-health-predicts-defects).)Then it names the fix. Not "this class is too big", but **Extract Class**, **ExtractHelper**, **Move Method**, **Break Cycle**, **Split File**, or **Extract Method**, with@@ -501,7 +514,7 @@ so your agent always knows how much to trust what it just read.| `search_codebase(query, kind?)` | Semantic search over the wiki, filterable by kind (implementation / test / config / doc), tagging each result's `search_method`. || `get_risk(targets, changed_files?)` | Hotspots, dependents, co-change partners, ownership, test gaps, bug history. Pass `changed_files` for PR mode and get a `directive` block back. || `get_change_risk(revspec)` | Pre-merge defect score for a whole commit or range from the shape of the diff, ranked as a percentile against recent commits, plus the tests coverage proves it touches. |-| `get_why(query?, targets?)` | Architectural decisions, their evidence spans and the supersession lineage. Falls back to git archaeology when no decisions exist. |+| `get_why(query?, targets?)` | Architectural decisions and their verbatim evidence spans, stamped exact / fuzzy / unverified. Falls back to git archaeology when no decisions exist. || `get_dead_code(...)` | Unreachable code by confidence tier with cleanup-impact estimates, and cross-repo consumer detection in workspace mode. || `get_health(targets?, include?)` | Per-file marker scores across all three signals. `include` opens coverage, trends, per-file signals, the accuracy self-check, and structured refactoring plans. |@@ -523,11 +536,13 @@ the rows we lose beside the rows we win.1 loss per instance. Deterministic grading, no LLM judge. *n=42, sign testp=0.00004.* CodeGraph scores the same on both halves to three decimals, soneither half is the easy one.-- **Less work in a real agent loop.** -15.9% output tokens against a bare agent,- leaner on 12 of 15 questions, and the only tool in the field to reach- significance. *n=15, sign test p=0.035.*-- **Agents actually call it.** Adopted in 15 of 15 cells, against CodeGraph 13,- Serena 4, Graphify 3, and code-review-graph 0 with 30 tools advertised.+- **Less work in a real agent loop.** -31.6% output tokens against a bare agent,+ leaner on 37 of 44 questions. *n=43, p<0.0001.* CodeGraph is a genuine second+ at -24.4%: more than one tool here works, and we lead the field rather than+ being alone in it.+- **Fewer steps to get there.** 3.8 tool calls where the bare agent needed 7.2,+ and 3.0 files opened instead of 7.2 — the mechanism behind the token saving,+ visible directly rather than inferred.**[The full results, the methodology, and the rows we lose →](docs/BENCHMARKS.md)**@@ -551,8 +566,9 @@ agent over MCP.| Private repo, no cloud | ✅ | ✅ | ✅ | ❌ OSS forks only || MCP tools served | 10 | 1 | 29 | 3 || **Finds the gold files** *([measured](docs/BENCHMARKS.md#1-finding-the-right-files), n=42 sealed)* | ✅ **0.876** | 0.610 | not in this run | not measured |-| **The agent actually calls it** *([measured](docs/BENCHMARKS.md#2-what-changes-in-a-real-agent-loop), n=15)* | ✅ **15/15** | 13/15 | 4/15 | not measured |+| **Output tokens vs a bare agent** *([measured](docs/BENCHMARKS.md#2-what-changes-in-a-real-agent-loop), n=43)* | ✅ **-31.6%** | -24.4% | -14.8% | not measured || **Index time, django** *([measured](docs/BENCHMARKS.md#6-indexing-time-the-row-we-lose))* | ⚠️ **366.8s**, slowest here | ✅ **16.4s** | not measured | n/a, cloud |+| | *one-time; updates after it are incremental* | | | || Generated documentation | ✅ | ❌ | ❌ | ✅ || Proactive agent hooks | ✅ Claude + Codex | ❌ | ❌ | ❌ || Auto-generated AI instructions (`CLAUDE.md`, `AGENTS.md`) | ✅ | ❌ | ❌ | ❌ |@@ -562,20 +578,27 @@ agent over MCP.| Multi-repo workspace intelligence | ✅ contracts, co-change, federated MCP | ❌ | ❌ | ❌ |CodeGraph builds its index **22x faster than we do**, and if a call graph is all-you need, that is the right trade. Graphify and code-review-graph were in the same-measured field and are on the benchmarks page.+you need, that is the right trade. With prose generation on, which is what a+default `repowise init` actually costs, it is **135x**. Graphify and+code-review-graph were in the same measured field and are on the benchmarks page.++<sub>Measured against CodeGraph 1.5.0, Graphify 0.9.31, Serena 1.6.2.dev0,+code-review-graph 2.3.7, on repowise `081a59fa` (between v0.37.0 and v0.38.0),+August 2026. Unmarked rows are capability presence, not measurements.</sub>### As a code health tool| | repowise | CodeScene ||---|---|---|-| Self-hostable, open source | ✅ AGPL-3.0 | ✅ Docker |-| Code health score (1-10) | ✅ 25 markers | ✅ 25-30 |+| Self-hostable, open source | ✅ AGPL-3.0 | ⚠️ on-prem Docker, proprietary |+| Code health score (1-10) | ✅ 49 detectors, 26 scoring | ✅ 25-30 || Brain Method / LCOM4 / god class | ✅ | ✅ || **Defects found at a 20% review budget** *([measured](docs/BENCHMARKS.md#5-code-health-predicts-defects), 2,770 files)* | ✅ **0.173** | 0.074 || **Effort-aware ranking, Popt** *(measured, p=0.003)* | ✅ **0.607** | 0.462 || **Precision at that budget** *(measured)* | 0.580 | ✅ **0.636**, a shorter list |-| Defect-prediction AUC, published and reproducible | ✅ 0.74 over 21 repos | ✅ Code Red study |+| **Discrimination, ROC AUC** *(measured, paired)* | 0.731 | 0.705 — *p=0.054, not significant* |+| Defect-prediction AUC, published and reproducible | ✅ 0.737 over 21 repos, held-out 0.76-0.78 | ✅ Code Red study |+| Business impact (resolution time) | ❌ *we could not replicate this on open data* | ✅ Code Red study || Git intelligence (hotspots, ownership, co-change) | ✅ | ✅ || Pre-merge change-risk scoring | ✅ 0-10 + directives | ✅ || Health trend + declining alerts | ✅ rolling snapshots | ✅ |diff --git a/docs/BENCHMARKS.md b/docs/BENCHMARKS.mdindex 04bfaebdb..87a9c6c5b 100644--- a/docs/BENCHMARKS.md+++ b/docs/BENCHMARKS.md@@ -442,11 +442,20 @@ A health score is worth something only if the files it flags are the files thatbreak. Scores are taken at a historical commit, bug fixes are counted over thefollowing six months, and nothing after the scoring commit feeds the score.-Across **21 repositories, 9 languages, 2,826 files**: **ROC AUC 0.74**-(95% CI 0.68 to 0.79), reaching 0.90 on individual repos. It survives controlling-for file size, so it is not simply flagging the big files, and it beats recent-churn by +0.10 AUC and prior-defect history by +0.12 (DeLong p < 1e-9). On-PROMISE/jEdit, a dataset it never saw, it holds at 0.76 to 0.78.+Across **21 repositories, 9 languages, 2,826 files**: **ROC AUC 0.737**+(95% CI 0.683 to 0.787), ranging from 0.55 to 0.86 across individual repos. It+beats recent churn by +0.100 AUC and prior-defect history by +0.117 (DeLong+p < 1e-9). On PROMISE/jEdit, a dataset it never saw and which carries no git+history at all, it holds at 0.76 to 0.78.++It is **not** better than raw file size at discrimination: LOC-only scores 0.742+against our 0.737 (p = 0.92, a tie). Where it wins is effort-aware ranking,+Popt +0.134 (95% CI +0.080 to +0.198) — same discrimination as counting lines,+much better at ordering a fixed review budget, and unlike a line count it says+why. Holding size fixed, within-band AUC runs 0.525 / 0.572 / 0.593 / 0.718+across NLOC quartiles: the signal survives cleanly only in the largest quartile,+and a purpose-built positive control confirms that collapse is a real absence+rather than too few positives to detect one.**Against CodeScene**, the closest commercial product and the only other vendorin this category with a published empirical defect study. Both tools scored the@@ -471,7 +480,23 @@ through. If what you want is a handful of files to fix this quarter rather thanthe ranking that catches the most defects, that operating point is the betterone, and it is a deliberate design choice rather than a weaker model. Our AUCedge is also **marginal**, not significant at 0.05, and we would rather say so-than round it up.+than round it up. So is our raw defect-density lead: 16.9x against 14.2x, but+p = 0.65 on a heavy tail. The size-normalized version of that row, 2.18x against+0.56x, is the one that reaches significance and the one to cite.++**The axis we lost outright.** CodeScene's "Code Red" study reports a Pearson+correlation of **−0.58** between Code Health and mean issue-resolution time, on+proprietary Jira cycle-time data. We tried to replicate that on open data and+could not. Across 17 repositories and 271 files, GitHub PR merge time correlated+with health at **−0.09** (95% CI −0.19 to +0.14), and six queue-independent+effort signals — commit span, commit count, review rounds, changes-requested,+commits after first review, review-comment density — all came back flat with+every interval spanning zero. The likely reason is structural: GitHub merge time+largely measures maintainer review-queue availability, not how hard a change was.+An early three-repo slice did show a review-rounds correlation of about −0.29,+and it did not survive expanding the corpus, so we treat it as small-sample+noise. **The business-impact axis remains CodeScene's, unreplicated on open+data.**Reports:**[BENCHMARK\_REPORT.md](https://github.com/repowise-dev/repowise-bench/blob/master/health-defect/BENCHMARK_REPORT.md)** ·diff --git a/docs/README.md b/docs/README.mdindex d2269a06d..c2d363bb9 100644--- a/docs/README.md+++ b/docs/README.md@@ -49,7 +49,7 @@ your agent in under five minutes, with no API key.| Doc | What it covers ||-----|----------------|| [layers/INTELLIGENCE_LAYERS.md](layers/INTELLIGENCE_LAYERS.md) | Overview of the five layers: graph, git, docs, decisions, code health |-| [layers/CODE_HEALTH.md](layers/CODE_HEALTH.md) | Defect risk, maintainability, and performance from 25 deterministic markers |+| [layers/CODE_HEALTH.md](layers/CODE_HEALTH.md) | Defect risk, maintainability, and performance from 49 deterministic detectors || [layers/REFACTORING.md](layers/REFACTORING.md) | Concrete, graph-aware refactoring plans (Extract Class, Move Method, Break Cycle) || [layers/CHANGE_RISK.md](layers/CHANGE_RISK.md) | Score any commit or `base..HEAD` range 0-10 for defect risk || [layers/BUG_HISTORY.md](layers/BUG_HISTORY.md) | Which files and symbols actually get bug-fixed, and how recently |diff --git a/docs/architecture/ARCHITECTURE.md b/docs/architecture/ARCHITECTURE.mdindex a3aed3f6e..84c1d364d 100644--- a/docs/architecture/ARCHITECTURE.md+++ b/docs/architecture/ARCHITECTURE.md@@ -851,7 +851,7 @@ tracked file, it computes:- **Timeline** — first and last commit dates (file age)- **Ownership** — primary owner from `git blame` (who wrote the most lines),top 3 contributors by commit count-- **Significant commits** — last 10 meaningful commit messages (filtered: no merges,+- **Significant commits** — up to 50 meaningful commit messages (filtered: no merges,no dependency bumps, no chore/ci, messages > 20 chars). These explain *why*the code evolved this way and are included in generation prompts.- **Co-change partners** — files that changed in the same commit >= 3 times, evendiff --git a/docs/architecture/code-health.md b/docs/architecture/code-health.mdindex 66d9ea1da..a077337ac 100644--- a/docs/architecture/code-health.md+++ b/docs/architecture/code-health.md@@ -281,24 +281,55 @@ than dormant ones.----## 5. The 26 markers and their categories+## 5. The markers and their categoriesEach marker is a stateless class implementing the `Biomarker` Protocol from`biomarkers/base.py`: a `name` (`"brain_method"`, `"nested_complexity"`, ...), a`category` (see `scoring.CATEGORY_CAPS`), and a `detect(ctx: FileContext)` methodreturning a list of `BiomarkerResult`s.+### The full roster++`biomarkers/registry.py` registers **49 detectors**; counting the three+governance findings written by the additive pass (`governance.py`) there+are **52 marker ids**. They divide by what each is permitted to affect:++| Group | Count | Scores into |+|---|---:|---|+| Defect-scoring | 26 | `defect` (8 of them also `maintainability`) |+| Performance | 20 | `performance` only |+| SQL | 3 | `maintainability` only |+| Governance | 3 | nothing — the finding surfaces, the score is untouched |++The authority is `scoring._BIOMARKER_DIMENSIONS`. Any biomarker **not** listed+there defaults into `defect`, which is why every `sql_*` and every performance+name must be listed explicitly: an omission would silently break the defect+golden guarantee (§6).++### Defect categories and caps+| Category | Cap | Markers ||------------------------|------|------------|-| Organizational | −3.5 | developer_congestion, knowledge_loss, hidden_coupling, function_hotspot, code_age_volatility, ownership_risk, churn_risk, change_entropy, co_change_scatter, prior_defect |+| Organizational | −3.5 | developer_congestion, knowledge_loss, hidden_coupling, function_hotspot, code_age_volatility, ownership_risk, churn_risk, change_entropy, co_change_scatter, prior_defect, ungoverned_hotspot†, stale_governance†, contradictory_decision† || Structural complexity | −2.5 | brain_method, low_cohesion, god_class, nested_complexity, bumpy_road, complex_conditional || Test coverage | −2.0 | untested_hotspot, coverage_gap |-| Test coverage (cont.) | −2.0 | coverage_gradient |+| Test coverage gradient | −2.0 | coverage_gradient || Size & complexity | −1.5 | complex_method, large_method, primitive_obsession || Duplication | −1.0 | dry_violation || Test quality | −0.5 | large_assertion_block, duplicated_assertion_block || Error handling | −0.5 | error_handling |+† The three governance markers carry a category and a weight, but the pass that+writes them runs *after* scoring completes and never touches+`HealthFileMetric.score` — so in practice they never deduct. They are counted+in the table above because `scoring.py` maps them, not because they move a+number.++The maintainability dimension has its own independent tables+(`_MAINTAINABILITY_CATEGORY`, caps: structural_complexity 4.0,+size_and_complexity 2.0, duplication 2.0, error_handling 2.0, sql 2.0), and the+performance dimension a single `performance` category capped at 2.0. See §6.+`large_assertion_block` and `duplicated_assertion_block` are the two**test-quality** smells (see §5.3). They fire only on test files and sit ina deliberately small category so a noisy test can't dominate its own score.diff --git a/docs/business/COMMERCIAL.md b/docs/business/COMMERCIAL.mdindex 5cea23ee8..82c1547ed 100644--- a/docs/business/COMMERCIAL.md+++ b/docs/business/COMMERCIAL.md@@ -47,7 +47,7 @@ All of the following ship in `pip install repowise` today, free for internal usefactor, significant commits, contributor profiles, module health), Documentation(LLM-generated wiki, freshness scoring, RAG search), Decision (architecturaldecision records linked to graph nodes, staleness tracking), and Code Health- (25 deterministic markers, 1–10 score per file, coverage ingestion, trend+ (49 deterministic detectors, 1–10 score per file, coverage ingestion, trendalerts).- **Ten task-shaped MCP tools** — `get_overview`, `get_answer`, `get_context`,`get_symbol`, `search_codebase`, `get_risk`, `get_change_risk`, `get_why`,diff --git a/docs/layers/CODE_HEALTH.md b/docs/layers/CODE_HEALTH.mdindex 3285dfa46..bbf061938 100644--- a/docs/layers/CODE_HEALTH.md+++ b/docs/layers/CODE_HEALTH.md@@ -1,31 +1,26 @@# Code Health-Repowise computes a 1-10 health score for every file in your repo from twenty-five-deterministic markers: McCabe complexity, deep nesting, brain methods,-class cohesion (LCOM4), god classes, clone detection, untested hotspots,-function-level churn, code-age volatility, ownership dispersion, relative churn,-change entropy, co-change scatter, recent defect history, test-quality smells,-and more. **No LLM calls, no cloud requirement.** Pure Python over tree-sitter-and git data, designed to finish in under 30 seconds on a 3,000-file repo.+Repowise scores every file in your repo from 1 to 10 on three signals: **defect+risk**, **maintainability**, and **performance risk**. The score is computed by+static analysis over tree-sitter ASTs and git history, with **no LLM calls and+no cloud requirement**, and the defect signal's weights are calibrated against a+real bug corpus rather than hand-tuned.<div align="center">-<img src="../../.github/assets/health-loop.svg" alt="repowise code-health loop: 25 deterministic markers fan into three signals (defect risk, maintainability, performance), the graph and git history locate where risk concentrates, and refactoring intelligence emits concrete plans (Extract Class, Extract Helper, Move Method, Break Cycle, Split File) your agent executes" width="100%" />+<img src="../../.github/assets/health-loop.svg" alt="repowise code-health loop: deterministic markers fan into three signals (defect risk, maintainability, performance), the graph and git history locate where risk concentrates, and refactoring intelligence emits concrete plans your agent executes" width="100%" /></div>-Code health runs as a loop: **measure** every file across three signals,-**locate** where the risk concentrates through the dependency graph and git-history, then **fix** it with a concrete refactoring plan an agent can execute-(see [Refactoring targets](#refactoring-targets) and-[docs/layers/REFACTORING.md](REFACTORING.md)).--Linters check patterns. The health score predicts risk. A linter flags a line-that matches a known-bad shape; the score estimates which files are likely to-harbor the next bug, ranks them, and is calibrated against real defect history-(see [Does the score find the bugs?](#does-the-score-find-the-bugs)). It uses-signals a linter has no access to: churn, ownership, co-change, blast radius,-hotspots. The [performance signal](#performance-static-performance-risk) is the-one part that overlaps a linter, and even it follows the call graph across files,-which a file-local linter cannot.+The loop is **measure → locate → fix**: score every file, use the dependency+graph and git history to find where risk concentrates, then emit a concrete+refactoring plan an agent can execute.++**Linters check patterns. This predicts risk.** A linter flags a line matching a+known-bad shape. The health score estimates which files are likely to harbor the+next bug, ranks them, and has been validated forward in time against real defect+history. It uses signals a linter has no access to — churn, ownership,+co-change, blast radius, hotspots — and even its one linter-adjacent pillar+(performance) follows the call graph *across* files, which a file-local linter+cannot do.## Quick start@@ -35,628 +30,384 @@ repowise health # KPIs + 20 worst-scoring files + top findingsrepowise update # re-score only changed files on each subsequent run```-Copy-paste walkthrough (health + coverage + impacted-tests):-[examples/health-coverage/](../../examples/health-coverage/).--Open `http://localhost:7777/repos/<id>/health` for the dashboard once the-local server is running (`repowise serve`).+Open `http://localhost:7777/repos/<id>/health` for the dashboard once+`repowise serve` is running.++## How a marker is actually computed++Every marker is a deterministic function of the parse tree, the dependency+graph, and git metadata. Two worked examples, one simple and one that shows why+this is not a linter.++**`nested_complexity` — pure AST.** tree-sitter parses the file into a syntax+tree. A walker descends it tracking control-flow depth, incrementing on each+`if` / `for` / `while` / `try` / `switch` node. If any function reaches depth 4+or more, the marker fires, and severity scales with the depth it reached. No+heuristics about intent, no model, no sampling. The same commit produces the+same finding forever.++**`io_in_loop` — AST plus call graph.** The walker finds a call inside a loop+body and resolves the callee through the same resolver the dependency graph+uses. The resolved target is classified against an I/O-boundary lexicon into+`db` / `network` / `filesystem` / `subprocess` / `lock`, and the marker fires+only on an actual round-trip (a `.execute`, an awaited HTTP call, a+`subprocess.run`) — never on a query-builder chain or a same-named pure helper.++The part a linter cannot reach: **the loop and the I/O call do not have to be in+the same function.** If the sink is not local, Repowise walks the resolved call+graph backwards up to three hops and reports the finding with its full+`caller → … → sink` path attached. clippy, ruff's `PERF` rules, ESLint and+golangci-lint all analyze one function at a time, so a loop in one file whose+database call lives in another is structurally invisible to them.++**Why deterministic is a feature, not a limitation.** The score is reproducible,+diffable between commits, and safe to gate CI on. It is also the precondition+for everything in the validation sections below: a score that returns a+different number each time it runs cannot be calibrated against a defect corpus+at all. An LLM asked to rate the same file twice gives two answers, and neither+can be checked against what actually broke.++The full analysis is budgeted to finish in **under 30 seconds on a 3,000-file+repo** — enforced by a CI test+(`tests/integration/test_health_perf_benchmark.py`), including a variant with a+per-line blame index built for every file.++## The markers, and what each is allowed to do++Repowise ships **49 registered detectors (52 marker ids)**, but only **26 are+permitted to move the headline number**. That restriction is deliberate: the+defect score carries published accuracy claims, so only markers that earned+their weight against a bug corpus may affect it.++| Tier | Markers | What it may do |+|---|---:|---|+| **Defect-scoring** | **26** | Calibrated weights; moves the 1-10 score |+| **Performance** | **20** | Own pillar, own cap; never touches the defect score |+| **Maintainability-only (SQL)** | **3** | Maintainability only |+| **Governance** | **3** | Surfaces as a finding; never deducts |++Nothing is inert, but "doesn't move the number" means three different things:++- **Floored (8 markers).** `developer_congestion`, `low_cohesion`,+ `brain_method`, `bumpy_road`, `primitive_obsession`, `dry_violation` and+ `error_handling` at ×0.5; `knowledge_loss` at ×0.4. These fire widely and+ matter for readability, but proved weak defect predictors under leakage-free+ calibration, so they were demoted rather than deleted — and most of them earn+ **full weight in maintainability**, where they belong.+- **Advisory (19 of the 20 performance markers).** Sub-1.0 weights inside a+ category capped at 2.0. Only `io_in_loop` carries full weight.+- **Non-scoring (3 governance markers).** `ungoverned_hotspot`,+ `stale_governance` and `contradictory_decision` are written by an additive+ pass that runs *after* scoring completes, so they surface in `get_risk`,+ `get_context` and `CLAUDE.md` without ever deducting.++### Markers that count toward two signals++Maintainability is not a separate detector set. It is a **second weighting of+the same evidence**, with its own category table and its own caps.++| Markers | In defect | In maintainability |+|---|---|---|+| `low_cohesion`, `brain_method`, `primitive_obsession`, `dry_violation`, `error_handling` | ×0.5 (floored — weak bug predictors) | **×1.0 (full weight)** |+| `god_class`, `large_method`, `nested_complexity` | ×1.13 / ×1.25 / ×1.34 (calibrated) | **×1.0** |+| `sql_high_complexity`, `sql_select_star`, `sql_update_delete_without_where` | — | ×0.7 |++So **11 markers feed maintainability: the 8 shared above plus the 3 SQL-only+ones.** The first group is the design in miniature — a smell that doesn't+predict bugs gets demoted in the pillar that claims to predict bugs, and paid in+full in the pillar that claims to measure how hard code is to work with.++One display nuance: scoring membership and a finding's displayed "home" pillar+are different. `god_class`, `large_method` and `nested_complexity` score in both+pillars but display as defect findings, so a pillar-filtered view shows 8+maintainability findings while 11 markers actually feed the number.++The complete marker roster, per-marker definitions, and the full weight tables+live in+[`docs/architecture/code-health.md`](../architecture/code-health.md#5-the-markers-and-their-categories).## The score-Each file starts at 10.0. Marker findings deduct from the score; deductions-are capped per category so any one category can drive the score down by at-most:--| Category | Cap | Markers |-|------------------------|-------|------------|-| Organizational | −3.5 | developer_congestion, knowledge_loss, hidden_coupling, function_hotspot, code_age_volatility, ownership_risk, churn_risk, change_entropy, co_change_scatter, prior_defect |-| Structural complexity | −2.5 | brain_method, low_cohesion, god_class, nested_complexity, bumpy_road, complex_conditional |-| Test coverage | −2.0 | untested_hotspot, coverage_gap |-| Test coverage (cont.) | −2.0 | coverage_gradient |-| Size & complexity | −1.5 | complex_method, large_method, primitive_obsession |-| Duplication | −1.0 | dry_violation |-| Test quality | −0.5 | large_assertion_block, duplicated_assertion_block |-| Error handling | −0.5 | error_handling |--Twenty-five markers across the categories above. `function_hotspot` and-`code_age_volatility` are blame-based and sit in the organizational bucket: both-are tier-aware and stay silent on ESSENTIAL-tier repos until the per-line blame-index is built.--Per-marker weight multipliers let the strongest empirical predictors deduct-more than the uniform severity table alone allows. The weights are **calibrated-offline against a defect corpus, not hand-tuned**: each file is scored at the-pre-window commit (T0, no leakage) and a logistic regression with NLOC as an-explicit control fits each marker's defect lift beyond file size. The runtime-stays deterministic; only the learned constants ship. The strongest predictors-are `co_change_scatter`, `change_entropy`, `ownership_risk`, and-`nested_complexity`; widely-firing smells that proved weak under leakage-free-scoring are floored. The full per-marker table lives in the-[architecture doc](../architecture/code-health.md#61-calibrated-weight-multipliers),-and the calibration with confidence intervals is in the-[benchmark report](https://github.com/repowise-dev/repowise-bench/blob/master/health-defect/BENCHMARK_REPORT.md).--The final score is clamped to `[1.0, 10.0]`. The three repo-level KPIs:--- **Hotspot Health**: NLOC-weighted average over the files classified as- hotspots by the git layer (high churn percentile plus minimum-activity- floors), not a fixed top-N slice.-- **Average Health**: NLOC-weighted average over all files.-- **Worst Performer**: single lowest-scoring file.--## Three health signals: defect risk, maintainability, and performance+Each file starts at 10.0 and marker findings deduct from it. Deductions are+capped per category, so no single category can dominate:-Repowise surfaces three orthogonal health signals computed from the same-marker stream by one shared scoring kernel: **defect risk** (the calibrated,-overall number), **maintainability**, and **performance**. They are co-equal-views, never blended into one number (the section on the overall score below-explains why).--The score above is the defect-risk signal: it is calibrated against a defect-corpus, the bands are calibrated to it (Alert files carry roughly 17x the defect-rate of Healthy files), and it is the overall number surfaced everywhere. But-not every code smell predicts bugs. A handful of markers fire widely and-matter a lot for how hard code is to read and change, yet proved weak as defect-predictors under leakage-free scoring, so the defect calibration floors them to-0.5 (`low_cohesion`, `brain_method`, `primitive_obsession`, `dry_violation`,-`error_handling`). Floored inside a defect-framed score they do two unhelpful-things at once: they still nudge the number a little (noise against the-calibrated signal) and they get no credit for the real problem they describe-(maintainability).--Repowise therefore computes a second, parallel signal, maintainability, from the-same marker stream:--- The floored smells above deduct at full weight (1.0) in maintainability- instead of the 0.5 the defect calibration imposes. The defect calibration does- not apply to a non-defect signal, so the maintainability weights are expert-set- and tuned only against the maintainability pillar's own per-category caps.-- The structural smells that are genuine defect predictors and core- maintainability concerns (`god_class`, `large_method`, `nested_complexity`)- count toward both dimensions.-- Pure defect and organizational predictors (`change_entropy`, `ownership_risk`,- `co_change_scatter`, and the like) stay out of maintainability entirely.-- SQL markers (`sql_high_complexity` for branchy stored routines,- `sql_select_star` in views/routines, `sql_update_delete_without_where`) are- maintainability-only by construction: no defect corpus covers procedural SQL,- so they carry an advisory weight under their own capped `sql` category and- never touch the defect score. Routine complexity is counted from the decision- keywords in the body text (procedural bodies do not parse into an AST);- statement smells come from the sqlglot AST and stay silent on dbt/Jinja- templates and on statements whose parse looks garbled.--The two signals are computed by the single shared scoring kernel-(`scoring.score_file`) against independent weight/category/cap tables, and they-never feed back into each other. The overall, surfaced score remains exactly the-defect score (byte-for-byte; a golden test locks this) until a later, deliberate-decision to blend. Maintainability is surfaced alongside it as a co-equal-headline:--- **REST/overview**: `summary.maintainability_average` plus a per-file- `maintainability_score` on every metric row.-- **MCP `get_health`**: `kpis.maintainability_average` and per-file- `defect_score` / `maintainability_score` / `performance_score`.-- **CLAUDE.md** and the CLI `status` line print a maintainability headline next- to defect-risk health.-- Every finding carries a `dimension` (`defect` / `maintainability` /- `performance`) naming the pillar it homes under, so findings can be filtered- per signal.--## Performance: static performance risk--The third signal, performance, flags shapes that waste work (code whose-structure does redundant I/O), rather than measured runtime. It is deliberately-high-precision and low-recall: a few real findings the rest of the toolchain can-trust beat a wall of maybes.--Standard linters do not find this class of problem. clippy, ruff's `PERF` rules,-ESLint, and golangci-lint read one function at a time, so a loop in one file-whose database call lives in another is invisible to them. On a 12,000-file-benchmark the standard linters found **0** of the cross-function I/O-in-loop-cases; repowise surfaced 557 findings across the run, about 90 of them spanning-function boundaries, and 98% fell in categories ruff has no rule for. The-findings are ordered by impact rather than raw count (NDCG 0.755 against 0.292-for severity-only). One caveat: the Rust dialect was new when the benchmark ran,-and clippy could not be run end-to-end on the corpus because of a Windows build-wall, so the Rust comparison there is catalogue-level, not a measured-head-to-head. The data is in the-[perf-detection benchmark](https://github.com/repowise-dev/repowise-bench/tree/master/perf-detection).--The detectors (all under one bounded `performance` category cap of 1.0, so the-pillar stays advisory) are:--- **`io_in_loop`**: a database call, network request, filesystem read, or- subprocess spawn that runs once per loop iteration: the classic N+1. This is- the moat. Two things make it more than a file-local lint:- - **Dependency classification.** The loop-nested call is resolved through a- shared I/O-boundary classifier (`io_kind ∈ {db, network, filesystem,- subprocess, lock}`) and only fires on a classified execution sink (an- actual round-trip like `.execute` / awaited HTTP / `subprocess.run`), not a- query-builder chain or a same-named pure helper.- - **Call-graph reachability.** The loop and the I/O call need not be in the- same function. A bounded-depth (≤3 hops) walk over the resolved `calls` graph- catches the interprocedural case (loop in `A`, sink in a helper `A` calls)- that no file-local linter can see. Cross-function findings carry their- resolved `caller -> ... -> sink` path for explainability.-- **`string_concat_in_loop`**: quadratic `+=` string building in a loop.-- **`blocking_sync_in_async`**: a synchronous blocking call inside an `async`- function, which stalls the whole event loop (mirrors ruff `ASYNC210/230/251`).-- **`resource_construction_in_loop`**: a heavy I/O client or connection- (`sqlite3.connect` / `httpx.Client` / `boto3.client` / `new PrismaClient` /- `new HttpClient` / `sql.Open`) constructed every iteration instead of hoisted:- connection churn, and socket exhaustion for `HttpClient`.-- **`lock_in_loop`**: a mutex acquired on every iteration (`lock.acquire` /- `mu.Lock` / `synchronized` / `lock(x){}`): a contention site. Activates the- `lock` I/O-boundary kind.-- **`serial_await_in_loop`**: an awaited I/O round-trip run one at a time in a- loop where a `gather` / `Promise.all` could fan it out. Advisory: a static- analyzer cannot prove the iterations are independent, so the finding suggests- rather than asserts.-- **`membership_test_against_list_in_loop`**: `x in big_list` (or- `big_list.includes(x)`) inside a loop is O(n·m); a set makes each lookup O(1).- Fires only when the right operand is provably a list, never a set or dict.-- **`nested_loop_with_io`**: an I/O sink in the inner body of a **nested** loop:- O(n·m) round-trips, the quadratic cousin of `io_in_loop`. The nesting itself- raises confidence the finding is real, so it surfaces alongside `io_in_loop`.-- **`blocking_io_under_lock`**: an I/O round-trip reached while a block-scoped- lock is held (a C# `lock(){}` / Java `synchronized(){}` block, directly or- through a call). Every other thread blocks for the full I/O wait, a throughput- killer. The cross-function case reuses the same bounded-reachability engine as- the N+1 moat, with a `lock -> io` entry set.--Two markers use centrality as a precision gate rather than only a sort key: a-shape that is noisy when flagged everywhere only fires in a hot function (one-with top-quintile call-graph in-degree, or in a churny/hotspot file), computed by-a reusable severity ranker over the same call graph the N+1 pass uses:--- **`hot_path_sync_io`**: a blocking subprocess or filesystem call in a hot,- request-reachable function, even outside a loop. It generalizes the pillar- beyond loops: its latency is paid on every call through the function. (DB and- network are excluded, since both are awaited in async code, and the un-awaited- calls a static pass sees are result materializers or chained awaits, not- blocking round-trips; subprocess and filesystem are synchronous by- construction.) Advisory: a latency signal ranked by centrality, not a defect.-- **`nested_loop_quadratic`**: a data-dependent loop nested inside another- (O(n²)) in a hot function. The centrality gate makes the list short and- reviewable (it cut a 13x volume of raw nested loops), but centrality answers- "is this function important", not "is n large", so it ships advisory and- informational only, never at a weight that moves the score.--A few markers are language-specific, contributed by that language's dialect (see-below) rather than the shared core:--- **`regex_compile_in_loop`** (Java, Go): a `Pattern.compile` /- `regexp.MustCompile` recompiled every iteration instead of hoisted. Skipped on- Python / .NET, which cache compiled patterns; on Go it fires only for a- string-literal pattern (a dynamic argument may legitimately vary per iteration- and cannot be hoisted).-- **`defer_in_loop`** (Go): a `defer` inside a loop holds the deferred handle- until the enclosing function returns, not the iteration: the classic Go file or- row-handle leak. A pure syntactic shape, very high precision.-- **`goroutine_in_unbounded_loop`** (Go): a `go ...()` spawned per element of a- `for k, v := range coll` loop, with no concurrency bound: a spawn explosion- (use a worker pool or bounded `errgroup`). Restricted to the two-variable- `range` form, which is only legal over a collection (a single-variable- `for i := range n` is a bounded count loop). Advisory.-- **`list_insert_zero_in_loop`** (Python): `lst.insert(0, x)` each iteration- shifts the whole list (O(n²)); use `collections.deque.appendleft`. Gated to a- literal `0` index and to a list not re-created each iteration.-- **`pd_concat_in_loop`** (Python): `pd.concat([acc, chunk])` inside a loop- copies the whole frame each pass (O(n²)); collect chunks and concat once.-- **`json_parse_in_loop`** (JS/TS): the `JSON.parse(JSON.stringify(x))`- deep-clone idiom in a loop (use `structuredClone`). Restricted to that idiom;- parsing a distinct payload each iteration is necessary work, not waste.- Advisory.-- **`array_spread_in_reduce`** (JS/TS): `arr.reduce((a, x) => [...a, x], [])`- rebuilds the accumulator every step (O(n²)); push-and-return instead. The- `.reduce` is itself the loop, so this fires regardless of an enclosing loop.-- **sync-over-async** (C#, via `blocking_sync_in_async`): `.Result` / `.Wait()`- / `.GetAwaiter().GetResult()` inside an `async` method blocks a thread-pool- thread. C# is the one non-Python language with real `async`/`await`.-- **`sql_cartesian_join`** (SQL): a comma-join (`FROM a, b`) with no join- predicate anywhere in the statement produces the full cross product (O(n·m)- rows). An explicit `CROSS JOIN` states intent and is not flagged; a comma-join- whose statement carries a `WHERE` is old-style join syntax and is not flagged- either. Detected on the sqlglot AST; a statement whose parse looks garbled- (dialect mismatch) emits nothing. Advisory.--Each performance finding's `details` carry the `boundary_kind` it crosses, a-`cross_function` flag, and the reachability `path` for the cross-function case.-Severity is ranked by centrality (an N+1 in a high-traffic, churny function-outranks one in a leaf), not by raw count.--**Languages.** The performance signal fires on Python, TypeScript/JavaScript,-Java, Go, and C#. Each language is a self-contained `PerfDialect` plugin-(`analysis/health/perf/dialects/`) that owns its callee-extraction grammar, its-execution-sink lexicon, the loop / string / async predicates, and its own marker-list, registered in `PERF_DIALECTS` like the rest of the per-language pipeline.-A language without a dialect emits no perf findings (never a wrong one). The-db/network/filesystem/subprocess lexicons and the per-language precision hazards-(Java `.find`/`.get`, GORM `Find`/`Save`, C# in-memory-vs-`IQueryable` LINQ) are-each handled inside that language's dialect. The verb sets are gated for-precision: distinctive sinks (EF `*Async`, Spring-Data `findBy*`, JDBC-`executeQuery`) fire on name alone, while ambiguous verbs require file-level-db-import evidence. `io_in_loop` is validated across languages on an 11-repo OSS-corpus: Go 96.7%, TypeScript 100%, Python 96.2% hand-labeled precision; the-`blocking_sync_in_async` C# `.Result`/Result-pattern collision and the Go-`*sql.Rows.Scan` cursor FP were caught and fixed by that validation.-`string_concat_in_loop` is validated at 100% (26/26) after a reset-per-iteration-guard (an accumulator re-initialized each iteration is bounded, not O(n²)).-Python's filesystem lexicon covers the round-trip verbs of `pathlib`-(`read_text` / `read_bytes` / `write_text` / `write_bytes`, matched on the method-name because only a `Path` spells them and the receiver is usually a plain-local), plus `os` and `shutil` verbs gated on the literal module root. Metadata-syscalls (`exists` / `is_file` / `stat` / `glob`) are deliberately excluded: a-stat per iteration is cheap and usually correct, so flagging it would be the-obvious FP class. The module-rooted half carries a ceiling worth knowing —-the root name resolves to the *leftmost* identifier of the callee chain, so-`os.path.join(a, b).verb()` also arrives as `root == "os"`; the verb set-therefore admits no name a `str` / `list` / `dict` answers to (`replace` is-omitted for exactly this reason, at no recall cost since `os.rename` covers it).--Because `hot_path_sync_io` fires *outside* any loop, it reads a narrower set-still (`hot_path_excluded_methods`): anything whose cost is bounded by one inode-— a single file read/write, one unlink/mkdir/rename, one directory listing — is-not a finding there, while an unbounded `os.walk`, `shutil.rmtree` or-`copytree` remains one. Keeping those two sets distinct is deliberate: a-per-language lexicon and a separately-calibrated marker should not widen-together just because they share a boundary kind.--Hand-labeled precision for the Python filesystem verbs is **87.6%** (163/186-non-test findings, every one labeled rather than sampled: 86.7% on this repo,-95% on Django). Where it loses precision is worth stating exactly, because it is-not the lexicon — same-function findings are near-perfect, and every false-positive came from a *cross-function* chain hitting one of three pre-existing-limits: a **memoized callee** (`get_or_build_*` index builders, `lru_cache`, a-`if path not in text_cache` guard) where the read runs once rather than per-iteration; a **first-hop name collision** (`set.update`, `sqlalchemy.update`-matched by name); and a **pre-read fast path** whose only caller always supplies-the text, leaving the disk branch unreachable. Memoization is already on the-documented static-blind list; what this change did was make far more of those-chains reachable by making their terminal sink visible. Narrowing that is the-top follow-up for this detector.--**Actionability caveat, pillar-wide.** `io_in_loop` reports a per-iteration I/O-round-trip; it does not claim the work is avoidable. Database and network-findings are usually batchable, but filesystem ones often are not — deleting N-files genuinely needs N unlinks. On Django, 16 of 19 true filesystem findings-were irreducible per-item work and 3 were batchable. The finding is still-correct and still tells you where the time goes; it is not, on its own, a defect.-`nested_loop_quadratic` now fires only on a same-collection shape (two nested-loops over the same collection = all-pairs O(n²)) instead of raw nesting depth;-that makes it precision-safe-by-construction but rare, so it stays advisory-only,-as do `blocking_io_under_lock`, `pd_concat_in_loop`, `json_parse_in_loop`, and-`goroutine_in_unbounded_loop` (high-precision by construction, low corpus recall).--**Soundness limits (honest, by design).** Performance is a static signal, so it-under-reports rather than over-reports (these cap recall, not precision): dynamic-dispatch, monkeypatching, and callbacks-as-values produce no `calls` edge and are-invisible; ORM lazy-load N+1 fires on attribute access (no visible call) and is-explicitly out of scope (this includes Hibernate lazy-load N+1, which fires on a-getter, and EF Core navigation-property lazy load), so we catch explicit-repository or query calls in loops, not attribute-triggered lazy loads; chains-longer than three hops from the loop are not followed; and an unmodelled library-is untyped (`None`), so its sinks don't fire. We call this performance RISK, never-measured performance, and never fold it into the defect score. The-commit-agreement precision study and its caveats are published in the-[perf-detection methodology](https://github.com/repowise-dev/repowise-bench/blob/master/perf-detection/METHODOLOGY.md).--**Verified findings.** `serial_await_in_loop` and `nested_loop_quadratic` are-advisory because a quick scan cannot tell whether a loop's iterations depend on-each other. When a deeper analysis can prove they do not, the finding is marked-verified: it asserts the fix (fan out the awaits, or a genuine quadratic scan)-instead of hedging, and carries `dataflow_verified: true`. The check is-conservative (anything it cannot prove stays advisory) and only touches-languages with a def/use dialect: Python, TypeScript, JavaScript, Go, Java, and-Rust. It changes the wording, not the score.--Performance surfaces exactly where maintainability does: a `performance_average`-on the overview summary and MCP `kpis`, a per-file `performance_score`, a-Performance KPI card and per-pillar finding filter on the dashboard, the-per-file Health tab and drawer, and a `Performance risk` line in CLAUDE.md and-the CLI `status` summary (each omitted/`null` on indexes built before the-detectors landed). The dimension names are mirrored in `@repowise-dev/types`-(`HEALTH_DIMENSIONS`) with a parity test on each side.--## The overall score: defect, not a blend--The single number repowise surfaces as the headline (the dashboard ring, the-band, the badge, the "does the score find the bugs?" stat) is, and stays, the-defect score. Maintainability and performance are presented as co-equal pillars,-not blended into the headline. This is a deliberate decision, for three reasons:--1. **Band calibration.** The Healthy/Warning/Alert cutoffs are calibrated to the- defect score (Alert is roughly 17x the defect rate). A blended headline would- invalidate those boundaries with no recalibration corpus behind the new number.-2. **Honesty of the validation stat.** "Does the score find the bugs?" is a claim- about the defect pillar; it must stay bound to the number it measures.-3. **Different precision profiles.** Maintainability is expert-set and performance- is high-precision/low-recall advisory. Neither is a calibrated bug predictor,- so neither should move the bug-calibrated headline.--A golden test (`tests/unit/health/test_scoring_dimensions`) locks the defect-score byte-for-byte against the pre-split single score, so no pillar can ever-regress it. Introducing a blended overall score would require a written rationale-and a recalibration plan; until then, overall = defect.--## Bands and distribution--On top of the 1-10 number, every score falls into one of three **bands**. These-are the single categorical scheme repowise surfaces (there is deliberately no-letter grade: a letter on top of the number would be a third overlapping scale-with arbitrary cliffs):+| Category | Cap |+|---|---|+| Organizational | −3.5 |+| Structural complexity | −2.5 |+| Test coverage | −2.0 |+| Test coverage gradient | −2.0 |+| Size & complexity | −1.5 |+| Duplication | −1.0 |+| Test quality | −0.5 |+| Error handling | −0.5 |-| Band | Score | Meaning |-|------|-------|---------|-| **Healthy** | `≥ 8.0` | Low-risk, maintainable. |-| **Warning** | `4.0 – 8.0` | Worth watching; rising complexity or process risk. |-| **Alert** | `< 4.0` | High-risk; concentrates defects. |--The cutoffs are not arbitrary. On our calibration corpus, Alert files carry-roughly 17x the per-file defect rate of Healthy files, so the band boundaries-are empirically defensible. They are defined once in core-(`analysis/health/grading.py`) and mirrored in `@repowise-dev/types` for the UI;-a parity test on each side locks the values.--The **health distribution** is the NLOC-weighted split of the repo across the-three bands: what share of your code (by volume, not file count) is Healthy,-Warning, or Alert. `repowise health` prints it as a one-line summary; the-dashboard renders it as a bar.--```text-Distribution (by code volume): 8% alert (12 files) · 21% warning (88 files) · 71% healthy (410 files)-```+When a category exceeds its cap, every finding in it is **scaled proportionally**+rather than truncated, so each finding's reported impact stays linear and+attributable — you can always explain exactly why a file scored what it scored.+The final score is clamped to `[1.0, 10.0]`.-## Badge+Three repo-level KPIs: **Hotspot Health** (NLOC-weighted average over files the+git layer classifies as hotspots), **Average Health** (NLOC-weighted over all+files), and **Worst Performer**.-`repowise health --badge` prints ready-to-paste Markdown for a README health-badge (a Shields-style color and `N.N/10` badge, no letter). A running Repowise-server (or the hosted app) also serves the badge directly:+### Bands-```text-GET /api/repos/{repo_id}/health/badge.svg # self-rendered flat SVG-GET /api/repos/{repo_id}/health/badge.json # Shields endpoint payload-```--Embed the dynamic form via Shields:--```markdown--```+| Band | Score | Meaning |+|---|---|---|+| **Healthy** | `≥ 8.0` | Low-risk, maintainable |+| **Warning** | `4.0 – 8.0` | Rising complexity or process risk |+| **Alert** | `< 4.0` | High-risk; concentrates defects |++The cutoffs are empirical, not arbitrary. On the 2,770-file measured corpus,+Alert files carry **16.9× the per-file defect rate of Healthy files**+(95% CI 8.6–29.0), and **2.18×** the defects-per-KLOC once file size is+normalized out (95% CI 1.00–3.58). Both numbers belong together: the raw ratio+is the headline, and the size-normalized one is the proof it is not simply an+artifact of large files being large.++The bands are defined once in core (`analysis/health/grading.py`) and mirrored+in `@repowise-dev/types`, with a parity test on each side.++## How the weights were calibrated++This is the part that separates "we wrote some heuristics" from "we measured+which heuristics predict bugs."++**The protocol.** Every file is scored at a commit that *precedes* the bug+window (T0), in a detached git worktree whose history stops there. Defect labels+are then collected from the following six months. Nothing after the scoring+commit can reach the score. An L2-regularized logistic regression with **NLOC as+an explicit control** fits each marker's defect lift *beyond file size*, with+leave-one-repo-out cross-validation for generalization. Only the learned+constants ship; the runtime stays fully deterministic.++**The trap that makes this hard.** Leakage here is two-sided, and both sides are+silent failures:++- Score at HEAD and the bug-fix commits themselves *manufacture* the churn,+ co-change and congestion that the process markers then "detect." The score+ looks excellent and means nothing.+- Score naively at T0 and every recency-windowed marker silently stops firing,+ because a worktree six months in the past sees an empty "last 90 days."++Getting both right requires anchoring the recency windows to the worktree's own+HEAD rather than wall-clock time. The evidence that this is load-bearing rather+than ceremonial: `developer_congestion` shipped at weight **1.5** under HEAD+scoring, and came back at coefficient **−0.08** under correct T0 scoring. It is+now floored to 0.5. A shipped constant was wrong by a factor of three, and only+the leakage-free protocol revealed it.++**What calibration rejected.** The markers listed as floored above are there+because the corpus said so. Candidate signals were also tested and formally+declared equivalent-to-null using bootstrap TOST equivalence testing rather than+a mere failure to reach significance: graph centrality (PageRank), code+naturalness, change bursts, and error-handling density all came back flat. Whole+experiments were run and refuted — size-relative scoring (lifts small files but+costs 0.07–0.12 AUC overall), function-level prediction (Popt at or below random+effort-ordering), and recalibrating on SZZ labels (*lower* out-of-fold AUC than+the shipped weights, so the shipped weights were kept).++Full methodology, per-marker coefficients and the rejected-candidate tables:+[repowise-bench/health-defect](https://github.com/repowise-dev/repowise-bench/tree/master/health-defect).## Does the score find the bugs?-The score is only worth anything if the files it flags are the files that-actually break. After an index, repowise checks that against the repo's own-history and prints a one-line callout:+### On your repo, at index time++After an index, Repowise checks the score against your repo's own history and+prints a one-line callout:```Does the score find the bugs? 16/20 lowest-health files had a bug fix in thelast 6 months, 3.3x the 24% baseline (80% vs 24%).```-It ranks every file by health score, takes the 20 lowest, and counts how many-were touched by a `fix:` commit in the trailing ~180-day window (the same-signal the `prior_defect` marker uses). That precision is contrasted with the-repo-wide base rate (the fraction of all files with a recent fix) to give the-lift. The same number appears on the web `health` and `overview` dashboards,-where it expands into a per-K table (worst 10/20/30), a concentration stat (what-share of recently-fixed files fall in the least-healthy 20%), and the exact-flagged files.--Agents can read the same stat over MCP, so a coding agent can confirm the score-is trustworthy on this repo before acting on it:--```python-# MCP: dashboard mode, the same precision@K / lift block-get_health(include=["accuracy"])-```+It ranks every file by score, takes the 20 lowest, and counts how many a `fix:`+commit touched in the trailing ~180 days, contrasted against the repo-wide base+rate. Agents can read the same stat over MCP with+`get_health(include=["accuracy"])`, so a coding agent can confirm the score is+trustworthy on *this* repo before acting on it.-It stays silent on repos without enough history to be honest (fewer than 25+It stays silent on repos with too little history to be honest (fewer than 25scored files, or fewer than 5 recently-fixed files). One caveat it discloses:-`prior_defect` is itself one (down-weighted) input to the score, so this is an-association on the indexed history, not a leakage-free forward prediction.--### Beyond your repo: cross-project validation--The per-repo callout is a quick local check. The score is also validated across-projects, so the number is not tuned to a single codebase. Every file is graded-at a commit that precedes the bug window, so no future information leaks in, then-checked against what actually broke:--- Across 21 repositories, 9 languages, and 2,826 files, the score reaches a mean- ROC AUC of **0.737** (95% CI 0.683 to 0.787). ROC AUC measures how often the- score ranks a known-buggy file worse than a clean one: 0.5 is a coin flip, 1.0- is perfect. It beats raw churn by 0.10 AUC and a prior-defects baseline by 0.117- AUC (DeLong p < 1e-9).-- On the public PROMISE/jEdit defect dataset, which played no part in- calibration, the same markers score AUC **0.76 to 0.78**. That held-out- result is the main evidence the signal is not overfit.-- The limits ship with the wins. Among files of similar size the signal is weak- (within-size-band AUC near 0.49), so part of the headline is simply that larger- files carry more risk; and a prior-defects baseline still finds bugs slightly- more efficiently under a fixed review budget (it wins on Popt by 0.085).--Full methodology, confidence intervals, and the named head-to-head against-CodeScene are in the [benchmark hub](https://github.com/repowise-dev/repowise-bench).--## The markers--**brain_method**: A single function that is simultaneously long, deeply-nested, highly complex, and central to the dependency graph. The strongest-single signal of fragile code. Centrality is judged against the repo's own-dependency density (top-quintile of connected files, with an absolute-hub bar), so it fires on sparse-graph languages too, not just Python.--**low_cohesion**: A class whose methods split into groups that share no-fields and don't call each other (LCOM4 ≥ 2). Measured by the walker's-class-level model; a high value usually means several smaller,-single-responsibility classes are hiding inside one.--**god_class**: A large class (≥ 200 lines, ≥ 15 methods) that also-contains a brain method. Size alone isn't flagged: the brain-method-requirement keeps flat data holders and config tables from firing.--**nested_complexity**: Functions with control-flow nesting ≥ 4 levels.-Hard to read, hard to test, hard to refactor.--**bumpy_road**: Multiple branches stacked at the same depth, usually a-sign the function is doing several jobs that should be split.--**complex_method**: Cyclomatic complexity ≥ 9. Each branch is a path the-test suite has to cover.--**large_method**: Long functions that also carry at least some branching.-A long-but-perfectly-flat body (a big config/data literal, a wall of-sequential assignments) is a layout artefact rather than a complexity smell,-so it is excluded: the trigger is about length-with-substance, not raw line-count.--**primitive_obsession**: Many primitive parameters in one signature. A-dataclass or parameter object would name the inputs. Suppressed in very small-modules (under ~60 non-blank lines), where a wide signature is an idiomatic-config/builder/forwarder rather than a design smell.--**dry_violation**: Cross-file code clones, detected by a native Rabin-Karp-rolling hash over tree-sitter tokens (variable renames don't hide a clone).-Pairs are ranked by co-change so dormant duplicates rank lower than active-ones.--**untested_hotspot**: A hotspot file with low or zero coverage and many-dependents. The textbook "write tests before refactoring" case.--**coverage_gap**: Non-test files with meaningful uncovered surface.-Severity grades along coverage depth.--**coverage_gradient**: A continuous coverage deduction that scales with the-uncovered fraction (`4.0 × (1 − line_coverage_pct/100)`, capped), so files stay-penalised in proportion to how much code is untested rather than only when they-fall below a hard threshold. Fires across the whole 0-100% range for files with-known coverage; silent (no imputation) where coverage was never ingested.--**developer_congestion**: Too many active authors touching the same file.-Usually an ownership problem dressed up as a code problem.--**knowledge_loss**: The primary authors of the file are no longer active-on the project. Refactor while someone still remembers why. Gated on recent-activity: an abandoned-but-stable file is low risk (the survivor effect),-so this only fires while the code is still being changed.--**hidden_coupling**: Files that consistently change in the same commits-without an explicit import or dependency edge between them. Captures-behavioral coupling (shared protocols, parallel config, copy-pasted-constants) that static analysis cannot see. Tier-aware: empty on-ESSENTIAL-tier repos until co-change backfill runs.--**complex_conditional**: Branch / loop guards that combine three or more-boolean operators. Severity grows with the operator count (LOW at 3, MED-at 4, HIGH at 5, CRIT at 6+).--**function_hotspot**: Functions that are both structurally complex and-frequently modified. Per-function modification counts come from a-per-line blame index built once per file (FULL git tier) and shared-with `code_age_volatility`. Fires when a function's distinct-commit-count is at or above the repo-wide p80 AND the function meets a-structural floor (CCN ≥ 10 or max nesting ≥ 3). Tier-aware: returns no-findings on ESSENTIAL-tier repos until `backfill_blame()` runs.--**code_age_volatility**: Functions whose median line age is at least a-year old that are suddenly being modified. Strong defect predictor:-the editor is usually working in unfamiliar territory. Uses the same-per-line blame index: `median_age_days` from per-line author-timestamps, `recent_mod_count` from distinct shas inside the last 30-days. Severity escalates with both axes (CRIT when median age ≥ 2y AND-≥ 5 recent commits). Tier-aware: same ESSENTIAL no-op as-`function_hotspot`.--**ownership_risk**: Long-run ownership dispersion. Counts minor-contributors (authors who each own less than 5% of the file's commits) and-the dominant owner's share. Many drive-by authors with no clear owner is-the single strongest defect correlate in the literature (Bird et al.). Fires-on files with ≥ 5 commits where ≥ 3 contributors are minor or no owner holds-40%. Complements `developer_congestion`, which measures active (90-day)-contention rather than lifetime dispersion.--**churn_risk**: Relative churn: the fraction of a file's lines rewritten in-the last 90 days, normalized by file size. A file whose recent window rewrote-more lines than it contains is structurally unstable regardless of how big it-is. Because the trigger is a ratio to NLOC, it does not simply re-flag large-files. Fires when the file is actively churning (≥ 5 recent commits, top-quartile of repo churn) and relative churn ≥ 1.0.--**change_entropy**: How scattered a file's change history is, adapted from-Hassan's History Complexity Metric. Each commit is treated as a one-period-window whose entropy is `log2(files-touched)`, distributed across its files and-decayed over time. A file repeatedly caught up in wide, scattered commits-scores high; one changed in focused, single-purpose commits stays low even if-it changes often, so this is not a churn proxy. Fires when the file is-actively changing (≥ 3 recent commits) and sits in the top 20% of repo change-entropy. Tier-aware: silent on ESSENTIAL-tier repos (no co-change walk).--**co_change_scatter**: Breadth of coupling. Counts the distinct files a file-co-changes with above the indexer's recording threshold; a high count means-editing it tends to ripple across the codebase (shotgun surgery). This is the-breadth complement to `hidden_coupling`, which flags specific undeclared-coupled pairs. Fires on actively-changing files (≥ 3 recent commits) coupled to-≥ 8 distinct partners. Tier-aware: silent on ESSENTIAL-tier repos.--## Test quality--These two fire only on test files and live in a deliberately small category-(cap −0.5), so a noisy test never dominates its own health score.--**large_assertion_block**: A test that fires 15 or more assertions in one-uninterrupted run. Such a test usually checks several behaviours at once: when-it fails it points at a line, not a cause, and it's brittle to unrelated-changes. Splitting it into focused cases makes failures legible.--**duplicated_assertion_block**: The same run of assertions copy-pasted across-tests. Reuses the Rabin-Karp clone detector and keeps only the clone regions-that overlap an assertion block on a test file. A change to the asserted-behaviour then has to be edited in several places, and usually isn't, so the-copies drift.--**error_handling**: Swallowed-exception and unsafe-unwrap anti-patterns: an-empty or comment-only `catch`/`except` body, a Python catch-all `except:` /-`except Exception:`, Rust `.unwrap()` / `.expect()` / `panic!`-family macros,-and Go's empty `if err != nil {}` or blank-identifier discard of a call's-error. Detection is precision-first: only the unambiguous shapes fire, and an-unsupported language or parse failure yields no signal rather than a guess.-Each occurrence is a LOW finding anchored to its line, and the whole category-is capped at −0.5 per file: this is an advisory maintainability flag (every-linter is expected to surface `except: pass`), deliberately not a calibrated-defect predictor. On the 21-repo benchmark it is AUC-neutral, so it is-excluded from the weight calibration and bounded so it can never move a file's-score by more than half a point.--## Test coverage--Ingest coverage reports, then run health:+`prior_defect` is itself one down-weighted input to the score, so this is an+association on indexed history, **not** a leakage-free forward prediction. That+is what the next section is for.++### Across projects, leakage-free++Every file graded at a commit preceding the bug window, then checked against+what actually broke. **21 repositories, 9 languages, 2,826 files, 379+defect-bearing.**++| Result | Value |+|---|---|+| Cross-project mean ROC AUC | **0.737** (95% CI 0.683–0.787) |+| Held-out (PROMISE / jEdit 4.0, 4.1) | **0.761** and **0.776** |+| vs. recent churn | **+0.100 AUC** (DeLong p = 5e-10) |+| vs. prior-defect history | **+0.117 AUC** (DeLong p = 3e-15) |+| Per-repo range | 0.55 (axios) to **0.86** (zod) |++ROC AUC measures how often the score ranks a known-buggy file worse than a clean+one: 0.5 is a coin flip, 1.0 is perfect. The **jEdit result is the important+one** — it is a public academic defect dataset that played no part in+calibration, scored from a single commit snapshot with no git history available,+so only the structural markers could fire. That is the main evidence the signal+is not overfit to our corpus.++Confidence intervals come from a two-stage repo-cluster bootstrap (2,000 seeded+replicates, resampling repositories then files within them), because the unit of+generalization is the repository, not the file.++### The limits, stated plainly++These are published because a validation section that only contains wins is not+a validation section.++- **The score does not beat raw file size on discrimination.** LOC-only scores+ AUC 0.742 against our 0.737 (ΔAUC −0.001, p = 0.92). Where it wins decisively+ is effort-aware ranking: **ΔPopt +0.134 (95% CI +0.080 to +0.198)**. Same+ discrimination as counting lines, far better at ordering a fixed review+ budget — and unlike a line count, it tells you *why* a file is risky.+- **Within a size band, most of the signal disappears.** Holding file size+ fixed, within-band AUC is 0.525 / 0.572 / 0.593 / 0.718 across NLOC quartiles.+ The two smallest quartiles straddle 0.5. Discrimination survives cleanly only+ in the largest quartile.+- **That last finding is a real absence, not a measurement artifact.** A+ purpose-built positive control injected a synthetic size-orthogonal signal of+ the headline magnitude into the same corpus, band membership and clustering,+ and recovered it with power 0.998–1.000 in exactly the quartiles where the+ real signal vanishes. So the collapse is a property of the score, not of the+ sample size.+- **A prior-defects baseline still beats us on effort-aware ranking.**+ ΔPopt −0.085 (95% CI −0.141 to −0.035). "Re-inspect whatever broke before" is+ a hard heuristic to beat per line reviewed. Our edge is discrimination plus an+ attributable explanation, not raw bugs-found-per-LOC.++## Head to head against CodeScene++CodeScene is the closest commercial product and the only other vendor in this+category with a published empirical defect study. Both tools scored the **same+2,770 files at the same leakage-free commit against the same labels**, through+the same estimator, with paired tests resampling the same repositories for both.++| Paired test | Repowise | CodeScene | p |+|---|---:|---:|---|+| Recall @ 20%-of-lines review budget | **0.173** | 0.074 | **0.003** |+| Effort-aware ranking (Popt) | **0.607** | 0.462 | **0.003** |+| Defect density, size-normalized | **2.18×** | 0.56× | **0.003** |+| Discrimination (ROC AUC) | 0.731 | 0.705 | 0.054 — *not significant* |+| Defect density, raw | 16.9× | 14.2× | 0.65 — *not significant* |+| Precision @ 20%-of-lines budget | 0.580 | **0.636** | 0.64 — *a tie* |+| Beyond file size (partial ρ) | −0.148 | −0.137 | *both beat size — a tie* |++Ranking by Repowise health surfaces **2.3× the defects under a fixed review+budget**. Three of seven axes are significant wins; the rest are honest ties or+losses:++- **AUC is not a win.** p = 0.054 is above 0.05. The correct statement is "at+ least as good, consistent small edge," not "significantly better."+- **CodeScene's precision lead is a real design choice.** It flags **27** Alert+ files where we flag **132**. That is a deliberately more conservative+ operating point: a short list a team will actually work through, traded+ against recall. If you want a handful of files to fix this quarter rather than+ the ranking that catches the most defects, that operating point is better, and+ it is not a calibration flaw.+- **We could not replicate their business-impact result.** CodeScene's "Code+ Red" study reports a Pearson correlation of −0.58 between Code Health and+ issue-resolution time on proprietary Jira data. On open GitHub data we+ measured **−0.09 (95% CI −0.19 to +0.14)**, and six queue-independent effort+ signals (commit span, review rounds, changes-requested, and others) all came+ back flat with every interval spanning zero. The likely reason is structural —+ GitHub merge time measures maintainer review-queue availability, not change+ difficulty — but the honest summary is that **the business-impact axis remains+ CodeScene's, unreplicated on open data.**++Reports:+[BENCHMARK_REPORT.md](https://github.com/repowise-dev/repowise-bench/blob/master/health-defect/BENCHMARK_REPORT.md) ·+[COMPARISON_REPORT.md](https://github.com/repowise-dev/repowise-bench/blob/master/health-defect/COMPARISON_REPORT.md)-```bash-pytest --cov --cov-report=lcov:coverage.lcov-repowise coverage add coverage.lcov-repowise health--# Cobertura, Clover, or multiple sources also work:-repowise coverage add backend/coverage.xml --format cobertura-repowise coverage add frontend/lcov.info-repowise health-```--Formats are auto-detected: **LCOV**, **Cobertura** XML, **Clover** XML, and a-**normalized JSON** (`repowise-coverage-v1`) keyed by repo-relative path; the-last lets you feed coverage from any runner once it's mapped to one shape:--```json-{ "format": "repowise-coverage-v1",- "files": { "src/foo.py": { "line_coverage_pct": 87.5,- "total_coverable_lines": 40 } } }-```+## Three health signals: defect risk, maintainability, and performance-Coverage data feeds into `untested_hotspot`, `coverage_gap`, and-`coverage_gradient` (a continuous deduction proportional to the uncovered-fraction), and shows up on the `/repos/<id>/health/coverage` dashboard.+The three signals are computed from the same marker stream by one shared scoring+kernel against independent weight, category and cap tables. They are co-equal+views, **never blended into a single number**.++**Defect risk** is the calibrated headline: the number on the dashboard ring,+the band, the badge, and every accuracy claim above.++**Maintainability** exists because not every smell predicts bugs. The markers+the defect calibration floored are real problems about how hard code is to read+and change; floored inside a defect-framed score they do two unhelpful things at+once — nudge a calibrated number with noise, and get no credit for the problem+they actually describe. Maintainability weights are expert-set, tuned only+against their own category budget, never against the defect corpus.++**Performance risk** flags code whose *structure* wastes work — redundant I/O,+quadratic accumulation, contention — rather than measured runtime. It is+deliberately high-precision and low-recall.++The headline stays exactly the defect score, byte-for-byte, locked by a golden+test against the pre-split implementation. Blending would require a written+rationale and a recalibration corpus, because the band cutoffs, the 16.9×+separation and every AUC figure above are claims about the defect pillar+specifically.++Every finding carries a `dimension` naming its pillar, and all three surface+identically: `summary.*_average` on the REST overview, `kpis.*` on MCP+`get_health`, per-file scores on every metric row, and a line each in `CLAUDE.md`+and `repowise status`.++## Performance risk++The pillar's core is **`io_in_loop`** — a database call, network request,+filesystem read or subprocess spawn executed once per loop iteration, the+classic N+1 — resolved through a shared I/O-boundary classifier and traced+across function boundaries as described above. Around it sit detectors for+quadratic string building, blocking calls inside `async` functions, connection+churn, lock contention, serial awaits that could fan out, membership tests+against lists, and language-specific shapes.++Two markers use call-graph centrality as a *precision gate* rather than a sort+key, firing only in hot functions (top-quintile in-degree, or in a churny+hotspot file), which keeps a noisy shape reviewable.++**Standard linters do not find this class of problem.** On a 12,600-file+benchmark, clippy, ruff, ESLint and golangci-lint together found **0** of the+cross-function I/O-in-loop cases; Repowise surfaced 557 findings of which ~87+span function boundaries, and 95%+ fell in categories ruff has no rule for.+Findings are ordered by impact rather than raw count (**NDCG 0.755** against+0.292 for severity-only ranking).++**Languages.** Performance analysis ships as a self-contained dialect plugin per+language — 11 dialects covering 16 language tags, including Python,+TypeScript/JavaScript, Java, Go, C#, Rust, Kotlin, Scala, Ruby, C++ and Dart. A+language without a dialect emits no findings, never a wrong one.++**Verified findings.** Two advisory markers (`serial_await_in_loop`,+`nested_loop_quadratic`) can be *promoted* when a def/use dataflow pass proves+the loop iterations are independent. A promoted finding asserts its fix instead+of hedging and carries `dataflow_verified: true`. The check is conservative —+anything it cannot prove stays advisory — and it changes the wording, not the+score.++**Soundness limits, by design.** Performance is a static signal, so it+under-reports rather than over-reports. Dynamic dispatch, monkeypatching and+callbacks-as-values produce no call edge and are invisible; ORM lazy-load N+1+fires on attribute access with no visible call and is explicitly out of scope;+chains beyond three hops are not followed; an unmodelled library has no+classified sinks. We call this performance **risk**, never measured performance,+and it never folds into the defect score.++One honest caveat about actionability: `io_in_loop` reports a per-iteration+round-trip, it does not claim the work is avoidable. Database and network+findings are usually batchable; filesystem ones often are not, since deleting N+files genuinely needs N unlinks. The finding still tells you where the time+goes.++Methodology and raw data:+[perf-detection](https://github.com/repowise-dev/repowise-bench/tree/master/perf-detection).## Refactoring targets@@ -664,281 +415,119 @@ fraction), and shows up on the `/repos/<id>/health/coverage` dashboard.repowise health --refactoring-targets```-A health score tells you a file is in trouble; a refactoring target names the-specific fix. Repowise emits one structured `RefactoringSuggestion` per-opportunity, computed deterministically during the health pass from data it has-already produced: the call graph, the class cohesion model, the clone pairs, and-git co-change. No re-parse, no LLM, inside the same <30s budget. Five detectors-ship today:--| Type | What it names | Built from |-|------|---------------|------------|-| **Extract Class** | The cohesion groups an incohesive / god class should split into: the exact methods + fields per group. | LCOM4 union-find components; god-class shape confirmed by Lanza-Marinescu (WMC = Σ McCabe, TCC). |-| **Extract Helper** | A clone's exact occurrences and where the shared helper should live. | Rabin-Karp clone pairs (line ranges + token count + co-change); extraction site = community centroid of the files. Transitive clones are clustered into one suggestion, not pairwise nags. |-| **Move Method** | A feature-envy method and the class it actually belongs to. | Jaccard distance of the method's entity set (fields/methods it touches) to each class over the call graph; fires only when a foreign class is clearly nearer than its own. |-| **Break Cycle** | The minimal set of import edges to invert to break a dependency cycle. | A strongly-connected component, then a greedy minimum feedback arc set over the real import edges. |-| **Split File** | The cohesive files an oversized module should decompose into: which symbols move where, plus the import edits in every dependent. | Community detection (Leiden/Louvain) over a weighted intra-file symbol graph, gated on partition **modularity**. Language-agnostic (reads `defines`/`calls` edges); the file-level analog of Extract Class. |--Each suggestion is structured data, not a string: a `plan` (the split groups,-the move target, the cut edges), the `evidence` that justifies it (LCOM4=3, the-clone ranges, the cycle size), the `impact_delta` (the health deduction it-recovers), an `effort_bucket` (`S`/`M`/`L`/`XL`), and a `blast_radius` (the-callers and co-changing files that must move with it). Human-readable text is-rendered from the structure at the edges (CLI / MCP / web); the structure is the-source of truth.--Ranking is graph-aware. Suggestions sort by `impact_delta × call-graph-centrality × blast_radius`, so a plan on a central hub file outranks the same-plan on a leaf, not the churn-only sort other tools use. The default surface-honors a `min_confidence` gate (`low` / `medium` / `high`, default `medium`).--For agentic workflows, the same data is one MCP call away:+A score tells you a file is in trouble; a refactoring target names the fix.+Repowise emits structured suggestions computed deterministically during the+health pass from data it already has — the call graph, the cohesion model, the+clone pairs, git co-change. No re-parse, no LLM, inside the same budget. Six+detectors ship:++| Type | What it names |+|---|---|+| **Extract Class** | The cohesion groups an incohesive class should split into — exact methods and fields per group |+| **Extract Method** | A behavior-preserving extraction, with the IN/OUT signature inferred from reaching-definitions and liveness analysis |+| **Extract Helper** | A clone's exact occurrences and where the shared helper should live |+| **Move Method** | A feature-envy method and the class it actually belongs to |+| **Break Cycle** | The minimal set of import edges to invert to break a dependency cycle |+| **Split File** | The cohesive files an oversized module should decompose into, plus the import edits in every dependent |++Each suggestion is structured data, not a string: a `plan`, the `evidence` that+justifies it, the `impact_delta` it recovers, an `effort_bucket`, and a+`blast_radius` of callers and co-changing files that must move with it. Ranking+is graph-aware — `impact × centrality × blast_radius`, so a plan on a central hub+outranks the same plan on a leaf.```pythonget_health(include=["refactoring"]) # ranked structured plansget_health(targets=["src/api/server.py"]) # one file in detail-get_health(targets=["module:src.api"]) # everything in a module```-The web **Refactoring** tab renders each plan as a card (split groups as a tree,-move arrow, clone occurrences with line links, file-split groups with their-residual core and import-rewrite list) with a **copy-to-agent** prompt and an-opt-in **Generate code** action that expands a plan into generated code plus a-unified diff. Code generation runs only on an explicit request, never in the-indexing hot path. It is enabled by default and can be turned off with-`refactoring.llm.enabled: false`. Full reference:-**[docs/layers/REFACTORING.md](REFACTORING.md)**.+Full reference: **[REFACTORING.md](REFACTORING.md)**.-## Trends+## Trends and coverage-Every health run writes a `HealthSnapshot` row (rolling 50 entries per repo).-Each snapshot stores the repo KPIs **and** a compact `{path: score}` map, so-the history doubles as a per-file record.--Two repo-level alerts run over the history:--- **Declining Health**: current `hotspot_health` is ≥ 0.5 below the- snapshot 5 runs ago.-- **Predicted Decline**: the three most recent snapshots are each- strictly below the one before.--Inspect from the CLI:+Every health run writes a snapshot (rolling 50 per repo) storing both repo KPIs+and a per-file score map, which powers two repo-level alerts — **Declining+Health** and **Predicted Decline** — and a per-file score trajectory with a+delta and declining flag. Trends stay silent on thin history rather than showing+a misleading single point.```bashrepowise health --trend```-Or from MCP:--```python-get_health(include=["trend"])-```--### Per-file score over time--The same snapshots power a per-file trajectory: a file's score plotted-across runs (CodeScene's signature view). It surfaces on the file's Health-tab and in the health drawer as a sparkline, with a delta vs. the previous-run and a Declining flag (the per-file version of the alerts above:-≥ 0.5 below the run 5 snapshots back, or three consecutive drops).--A trend is silent on thin history: it needs at least two snapshots that-both carry the file, otherwise the UI shows "no score history yet" rather-than a misleading single dot. Gaps (a file absent from some snapshots) are-skipped, not zero-filled.--Fetch it directly:--```bash-# REST: one file's series + current delta + declining flag-GET /api/repos/{repo_id}/health/files/trend?file_path=path/to/file.py-```--```python-# MCP: targeted mode attaches a per-file `trends` block-get_health(targets=["path/to/file.py"])-```--## File signals--Every file carries process, people, and topology signals we already compute-during indexing. They answer "should I worry about this file?" with context-the score alone can't, and they surface together on the file's Health tab and-in the health drawer, grouped, captioned, and silent ("no signal") when the-underlying data is absent rather than imputed.--| Group | Signal | Means |-|-------|--------|-------|-| Process | Prior defects | Bug-fix commits touching this file in the last ~6 months. `0` is a real, reassuring signal. |-| Process | Change scatter | `change_entropy_pct` (0-100): how spread out its edits are across commits. High = chaotic change. |-| Process | 90-day churn | Commits and lines added/deleted in the trailing 90 days. |-| Process | Age | How long the file has existed in git history. |-| People | Primary owner | The all-time top committer and their commit share. |-| People | Recent owner | The top committer in the last 90 days. A different name from the primary owner flags a knowledge handoff. |-| Topology | Dependents | How many files depend on this one (graph in-degree). |-| Topology | Dependencies | How many files this one depends on (graph out-degree). |--These are pure surfacing: no new measurement, no scoring. Fetch them directly:--```bash-# REST: embedded in the file-detail aggregate and the drawer breakdown-GET /api/repos/{repo_id}/files/{path} # data.health.signals-GET /api/repos/{repo_id}/health/files/breakdown?file_path=path/to/file.py-```--```python-# MCP: attached to the get_context health block (null fields dropped)-get_context(targets=["path/to/file.py"], include=["health"])--# MCP: also on get_health targeted mode, one `signals` object per metric-get_health(targets=["path/to/file.py"], include=["signals"])-```--## Hotspot anatomy--Two views dissect where risk concentrates, both plotted from data already on-disk (churn from the git indexer, complexity from the walker, blame at function-granularity).--### Churn × complexity--One dot per recently-changed file: the x-axis is its 90-day commit count-(churn), the y-axis is its max cyclomatic complexity, dot size is NLOC, and dot-color is the health band. Dashed guides sit at the repo's median churn and-median complexity, so the tinted top-right corner reads "busier and more-complex than a typical file here", the refactor zone, where volatility and-tangle collide and defects concentrate. It lives on the **Hotspots & churn**-dashboard tab, toggleable with the churn × bus-factor view.+Coverage reports light up the test-coverage markers. **LCOV**, **Cobertura**,+**Clover** and a normalized JSON format are auto-detected:```bash-# REST: repo-level point list (one point per churned file)-GET /api/repos/{repo_id}/health/churn-complexity-```--```python-# MCP: the same point list, dashboard mode-get_health(include=["churn_complexity"])+pytest --cov --cov-report=lcov:coverage.lcov+repowise coverage add coverage.lcov+repowise health```-Files with no recent churn are omitted (they have nothing to say on the churn-axis); complexity is never used to filter, so a high-churn, low-complexity file-still shows in the bottom-right ("changes constantly but stays simple").--### Functions by churn--The file's Health tab lists its functions ranked by modification count, with-the 90-day recent-mod count, median age, and blame owner per function, the-same `git_function_blame` rollup the symbol page uses. It promotes per-function-ownership and volatility out of the buried marker cards into a first-class-table, so "which function in this file is the actual hotspot" is one glance away.-## ConfigurationPer-file overrides live in `.repowise/health-rules.json`:-```json-{- "disabled_biomarkers": ["primitive_obsession"],- "rules": [- {- "path": "tests/**/*.py",- "disabled_biomarkers": ["large_method", "complex_method"]- },- {- "path": "src/legacy/**",- "disabled_biomarkers": ["dry_violation"]- }- ]-}-```--`path` holds an fnmatch-style glob over the repo-relative POSIX path-(`path_glob` and `glob` are accepted aliases).--### Severity overrides and profiles--A team can soften a signal it treats as advisory without disabling it-outright by remapping its severity (typically a demotion). Overrides apply-repo-wide and per-path; an explicit per-path entry wins over the repo-wide one.-```json{"profile": "small-team",+ "disabled_biomarkers": ["primitive_obsession"],"severity_overrides": { "complex_method": "low" },"rules": [- { "path": "src/generated/**", "severity_overrides": { "large_method": "low" } }+ { "path": "tests/**/*.py", "disabled_biomarkers": ["large_method"] }]}```-Accepted severity values are `low`, `medium`, `high`, `critical`. The named-`small-team` profile expands to a preset demotion of the process/people and-noisier structural signals a 1-3 person repo can't support; an explicit-`severity_overrides` key always wins over the preset.+`path` is a gitignore-semantics glob over the repo-relative path. The+`small-team` profile demotes the process and people signals a 1-3 person repo+cannot support; an explicit `severity_overrides` key always wins over it.-Only the severity label is tunable. The per-marker weight multipliers and-the category caps are the calibrated constants the benchmark numbers rest on-and are deliberately not overridable, so a team's local policy never changes-what the published accuracy claims mean. Markers that carry a continuous-deduction (`coverage_gradient`) are unaffected by severity remaps.+**Only the severity label is tunable.** The per-marker weight multipliers and+the category caps are the calibrated constants the published accuracy numbers+rest on, and they are deliberately not overridable — so a team's local policy can+never silently change what those numbers mean.## Incremental updates-`repowise update` only re-scores the changed files. Findings and metrics for-unchanged files stay put, no nightly full re-index needed.--## Status one-liner--`repowise status` includes a single-line health summary (the maintainability and-performance pillars append once the index has populated them):--```-Health: 7.4 (avg) · 6.2 (hotspots) · 2.1 (worst: payments/processor.ts) · 7.0 (maintainability) · 9.1 (performance)-```+`repowise update` re-scores only changed files. Findings and metrics for+unchanged files stay put; no nightly full re-index.## Comparison-How the code-health signal compares to the tools it is most often placed next-to. The honest dividing line: each of these has a rules or linter engine and a-definitional rating; repowise predicts which files harbor the next bug and-validates that forward in time against a labeled bug corpus. Each tool also has-real strengths repowise does not claim to beat, noted under the table.+The honest dividing line: each tool below has a rules engine and a definitional+rating. Repowise predicts which files harbor the next bug and validates that+forward in time against a labeled corpus.-| Capability | Repowise | CodeScene | SonarQube | Code Climate / Qlty¹ | Codacy |+| Capability | Repowise | CodeScene | SonarQube | Qlty¹ | Codacy ||---|---|---|---|---|---|-| Per-file health / maintainability score | ✅ 1-10, 25 markers | ✅ 1-10 Code Health | ⚠️ A-E ratings from rule counts | ✅ A-F grade | ✅ A-F grade |-| Score uses git / behavioral signals (churn, ownership, co-change, hotspots) | ✅ | ✅ its core | ❌ static rules only | ⚠️ churn-vs-complexity, no co-change | ❌ |-| Cross-file / call-graph analysis | ✅ interprocedural call graph | ⚠️ git temporal coupling, not a static call graph | ⚠️ cross-file taint (security only) | ❌ file-local | ❌ file-local |-| Defect-validated score vs a real bug corpus (published numbers) | ✅ mean ROC AUC 0.737, held-out 0.76-0.78 | ⚠️ "Code Red" study (defect density / cycle time), no per-file AUC | ❌ | ❌ | ❌ |-| Static performance risk (N+1 / IO-in-loop across the call graph) | ✅ | ❌ | ❌ | ❌ | ❌ |-| Test-coverage ingestion | ✅ LCOV/Cobertura/Clover/JSON | ✅ JaCoCo/Cobertura/LCov/Clover | ⚠️ imports reports | ✅ | ✅ |-| Untested-hotspot detection (coverage × risk) | ✅ | ✅ | ❌ | ⚠️ coverage beside hotspots, no composite | ❌ |-| Concrete cross-file refactoring plans (Extract Class / Move Method / Split File) | ✅ + opt-in LLM code-gen | ⚠️ AI fixes 5 in-function smells, no cross-file plan | ❌ flags + AI CodeFix | ❌ flags only | ❌ flags + per-issue AI autofix |-| Health trend tracking + declining-health alerts | ✅ | ✅ | ✅ quality gates, new-code focus | ✅ | ✅ quality gates |-| AI-agent / MCP integration | ✅ MCP tools | ✅ Code Health MCP server | ✅ official MCP server | ❌ | ✅ official MCP + Guardrails |-| Security scanning (adjacent axis) | ⚠️ separate security layer | ⚠️ secondary | ✅ strong (cross-file taint) | ❌ | ✅ full SAST/SCA/secrets/IaC |-| License | ✅ AGPL-3.0, free internal use | ⚠️ free OSS; paid per active author | ⚠️ Community free; paid by LOC | ⚠️ free OSS; paid teams | ⚠️ free OSS; paid per dev |--¹ Code Climate Quality was spun out as Qlty Software in November 2024; "Code-Climate" now sells the Velocity engineering-analytics product. This row tracks-the Quality / Qlty capabilities. SonarQube "Security Hotspots" are-static-analysis security flags, not git-churn hotspots.--What each tool does that repowise does not: **SonarQube** has the broadest-security scanning (cross-file taint analysis), the widest language coverage, and-the most widely adopted merge-gate model. **CodeScene** is the most mature-behavioral-analysis product (knowledge maps, off-boarding simulation, 28+-languages) and has the only published empirical defect study in this group (the-"Code Red" correlation study, Tornhill and Borg, TechDebt 2022). **Code Climate /-Qlty** defined the churn-vs-complexity quadrant and pairs with Velocity for DORA-analytics. **Codacy** has the widest security suite of the four and polished PR-automation.+| Per-file health score | ✅ 1-10 | ✅ 1-10 | ⚠️ A-E from rule counts | ✅ A-F | ✅ A-F |+| Score uses git / behavioral signals | ✅ | ✅ its core | ❌ static rules only | ⚠️ churn vs complexity | ❌ |+| Cross-file / call-graph analysis | ✅ interprocedural | ⚠️ git temporal coupling | ⚠️ taint, security only | ❌ file-local | ❌ file-local |+| Defect-validated against a bug corpus | ✅ AUC 0.737, held-out 0.76-0.78 | ⚠️ "Code Red" study, no per-file AUC | ❌ | ❌ | ❌ |+| Static performance risk across the call graph | ✅ | ❌ | ❌ | ❌ | ❌ |+| Test-coverage ingestion | ✅ | ✅ | ⚠️ imports reports | ✅ | ✅ |+| Cross-file refactoring plans | ✅ + opt-in codegen | ⚠️ 5 in-function smells | ❌ | ❌ | ❌ |+| Trend tracking + declining alerts | ✅ | ✅ | ✅ quality gates | ✅ | ✅ |+| MCP / agent integration | ✅ | ✅ | ✅ | ❌ | ✅ |+| Security scanning | ⚠️ separate layer | ⚠️ secondary | ✅ strong | ❌ | ✅ full suite |+| License | ✅ AGPL-3.0 | ⚠️ proprietary, on-prem Docker | ⚠️ Community free, paid by LOC | ⚠️ free OSS, paid teams | ⚠️ free OSS, paid per dev |++¹ Code Climate Quality was spun out as Qlty Software in November 2024.++**What each does that we do not.** **SonarQube** has the broadest security+scanning, the widest language coverage, and the most adopted merge-gate model.+**CodeScene** is the most mature behavioral-analysis product — knowledge maps,+off-boarding simulation, 28+ languages — and holds the only published business-+impact study in this group, which we could not replicate on open data.+**Qlty** defined the churn-vs-complexity quadrant. **Codacy** has the widest+security suite of the four and polished PR automation.## See also-- [`packages/core/src/repowise/core/analysis/health/README.md`][hr]:- developer overview of the layer.-- Sub-package READMEs: `complexity/`, `coverage/`, `duplication/`,- `biomarkers/`.--[hr]: ../packages/core/src/repowise/core/analysis/health/README.md+- [`docs/architecture/code-health.md`](../architecture/code-health.md): internals,+ the full marker roster, and the complete weight tables.+- [`docs/BENCHMARKS.md`](../BENCHMARKS.md): every published number with its+ sample size and test.+- [REFACTORING.md](REFACTORING.md) · [TEST_INTELLIGENCE.md](TEST_INTELLIGENCE.md) ·+ [BUG_HISTORY.md](BUG_HISTORY.md)diff --git a/docs/layers/INTELLIGENCE_LAYERS.md b/docs/layers/INTELLIGENCE_LAYERS.mdindex 4668e8e70..134adcd02 100644--- a/docs/layers/INTELLIGENCE_LAYERS.md+++ b/docs/layers/INTELLIGENCE_LAYERS.md@@ -1,16 +1,19 @@-# The Five Intelligence Layers+# The Intelligence Layers-repowise indexes your codebase **once**, builds five intelligence layers, then-keeps them in sync on every commit. This document is the deep dive; the README-gives the one-paragraph version of each layer and links here for the detail.+repowise indexes your codebase **once**, builds five foundational layers, then+keeps them in sync on every commit. Four further layers are derived from those+five. This document is the deep dive; the README gives the one-paragraph version+of each and links here for the detail.<div align="center">-<img src="../../.github/assets/intelligence-layers.svg" alt="repowise's five intelligence layers: one index (repowise init) fans into Graph, Git, Docs, Decisions, and Code Health, each surfaced through its signature MCP tool and delivered through 10 task-shaped tools, the CLI, the local dashboard, auto-generated CLAUDE.md/AGENTS.md, and the PR bot" width="100%" />+<img src="../../.github/assets/intelligence-layers.svg" alt="repowise's intelligence layers: one index (repowise init) fans into Graph, Git, Docs, Decisions, and Code Health, each surfaced through its signature MCP tool and delivered through task-shaped tools, the CLI, the local dashboard, auto-generated CLAUDE.md/AGENTS.md, and the PR bot" width="100%" /></div>-The layers compound. The graph locates what git flags,-code health scores it, decisions explain why it is shaped that way, and the docs-make all of it searchable in natural language.+The layers compound. The graph locates what git flags, code health scores it,+decisions explain why it is shaped that way, and the docs make all of it+searchable in natural language.++**Foundational**1. [Graph Intelligence](#graph-intelligence)2. [Git Intelligence](#git-intelligence)@@ -18,100 +21,115 @@ make all of it searchable in natural language.4. [Decision Intelligence](#decision-intelligence)5. [Code Health Intelligence](#code-health-intelligence)-Two cross-cutting capabilities sit on top of the layers:+**Derived** — built on the five, each with its own reference page++6. [Change Risk](#change-risk) · [Test Intelligence](#test-intelligence) ·+ [Bug History](#bug-history) · [Security](#security) ·+ [Dead Code](#dead-code-detection)++**Cross-cutting**- [Proactive context enrichment: hooks](#proactive-context-enrichment-hooks)-- [Auto-sync: five ways to stay current](#auto-sync-five-ways-to-stay-current)+- [Auto-sync](#auto-sync-five-ways-to-stay-current)- [Auto-generated CLAUDE.md](#auto-generated-claudemd)+Everything below is computed **without model calls**. An LLM is an optional+upgrade for prose quality in the docs layer, never a requirement for the index.+---## Graph Intelligence-tree-sitter parses every file across 18 languages into a **two-tier dependency-graph**: file nodes and symbol nodes (functions, classes, methods). A 3-tier-call resolver with confidence scoring handles import aliases, barrel-re-exports, and namespace imports. Heritage extraction covers `extends`,-`implements`, trait impls, derive macros, mixins, and extension conformance.+tree-sitter parses your source into a **two-tier dependency graph**: file nodes+and symbol nodes (functions, classes, methods). 18 languages parse to a full+AST; see [`LANGUAGE_SUPPORT.md`](LANGUAGE_SUPPORT.md) for per-language tiers.++A **3-tier call resolver with confidence scoring** handles import aliases,+barrel re-exports and namespace imports: same-file resolution at 0.95,+import-scoped at 0.90, globally-unique at 0.50, with language-specific pre-tiers+for Go packages, JVM same-package and C++ same-target. Heritage extraction+covers `extends`, `implements` and trait impls.-- **Leiden community detection** finds logical modules even when your directory- structure doesn't reflect them.+- **Leiden community detection** (Louvain fallback) finds logical modules even+ when your directory structure doesn't reflect them.- **PageRank, betweenness centrality, SCC analysis, and execution-flow tracing**- from entry points identify your most central, most coupled, and most- traversed code.-- **Framework-aware edges** connect routes to handlers for Django, FastAPI,- Flask, ASP.NET, Spring Boot, Express/NestJS, Gin/Echo/Chi, Axum/Actix, Rails,- Laravel, and more.+ from entry points identify your most central, most coupled, and most traversed+ code.+- **Framework-aware edges** connect routes to handlers across 22 framework+ detectors, including Django, FastAPI, Flask, ASP.NET, Spring, Micronaut,+ Quarkus, Jakarta, Express/NestJS, Next.js App Router, Remix, tRPC, Hono,+ Gin/Echo/Chi, Axum/Actix/Rocket, Rails, Laravel, TYPO3, Flutter and the+ pytest/gtest test runners.-See [`docs/layers/LANGUAGE_SUPPORT.md`](LANGUAGE_SUPPORT.md) for per-language coverage-and [`docs/reference/COMPUTED_GLOSSARY.md`](../reference/COMPUTED_GLOSSARY.md) for every derived metric.+Every derived metric is defined in+[`COMPUTED_GLOSSARY.md`](../reference/COMPUTED_GLOSSARY.md).---## Git Intelligence-repowise mines your git history (per-file, configurable depth) to produce-signals no static analysis can find.+repowise mines your git history to produce signals no static analysis can find.-**Hotspots**: files in the top 25% of *both* churn and complexity. These are-where bugs live. Flagged in the dashboard, in `CLAUDE.md`, and surfaced by-`get_risk()` before your agent touches them.+**Hotspots**: files in the top quartile of *decayed* churn (exponential+half-life, so last month outweighs last year) that also clear minimum-activity+floors — at least 3 commits in 90 days and a meaningful temporal score. Churn+alone would flag every file a bulk refactor touched; the floors keep the list+about sustained activity. Surfaced by `get_risk()` before your agent edits.-**Ownership**: `git blame` aggregated into ownership percentages per author.-Know who to ping. Know where knowledge silos exist.+**Ownership**: `git blame` aggregated into per-author percentages. Know who to+ping, and where knowledge silos are.**Co-change pairs**: files that change together in the same commit *without* an-import link. Hidden coupling that AST parsing cannot detect. `get_context()`-surfaces co-change partners alongside direct dependencies.+import link. Hidden coupling AST parsing cannot detect. `get_context()` surfaces+co-change partners alongside direct dependencies.-**Bus factor**: files owned >80% by a single author. Shown in the ownership-view, surfaced in `CLAUDE.md` as knowledge risk.+**Bus factor**: how many authors it takes to cover 80% of a file's commits. A+bus factor of 1 is the classic single-owner risk, surfaced in `CLAUDE.md`.-**Significant commits**: the last 10 meaningful commit messages per file-(filtered: no merges, no dependency bumps, no lint) feed generation prompts, so-the wiki explains *why* code is structured the way it is.+**Significant commits**: up to 50 meaningful commit messages per file — merges,+dependency bumps, lint-only and bot commits filtered out — with the commit body+retained when it carries decision intent. These feed generation prompts, so the+wiki can explain *why* code is structured the way it is.-**Contributor profiles**: every author with commits gets a profile page:-modules they own, top files, co-authors, commit category mix-(feat / fix / refactor / docs / test / chore / perf), silo modules they're-solely on, bus-factor risk files, and dead-code burden. Surfaced via-`/repos/<id>/owners` and linked from every owner reference.+**Contributor profiles**: every author gets a page — modules they own, top+files, co-authors, commit category mix, silo modules, bus-factor risk files, and+dead-code burden.-**Module health**: a composite 0–100 score per top-level module derived from-silo penalty, hotspot density, dead-code percentage, average churn, doc-coverage, and median bus factor. Surfaced on the Risk page and the per-module-detail page, with cross-links to owners, hotspots, and governing decisions.+**Module health**: a 0–100 composite per top-level module from silo penalty,+hotspot density, dead-code percentage, average churn, doc coverage and median+bus factor.-**Reviewer suggestions**: paste a PR file list into Blast Radius and get a-ranked list of likely reviewers, scored by direct authorship (×1.0), co-change-partners (×0.5), and recency (×0.4), capped at the 5 strongest co-change signals-per file.+**Reviewer suggestions**: paste a PR file list into Blast Radius for a ranked+reviewer list, scored by direct authorship (×1.0), co-change partners (×0.5) and+recency (×0.4), capped at the 5 strongest co-change signals per file.---## Documentation Intelligence-An LLM-generated wiki for every module and file, rebuilt **incrementally** on-every commit.+A wiki for every module and file, rebuilt **incrementally** on every commit.+Deterministic templates render it with no model calls; supplying an LLM key+upgrades the prose rather than enabling the layer.- **Coverage tracking**: what's documented and what isn't.-- **Freshness scoring** per page: confidence scores show how current each page- is relative to the underlying code.-- **Semantic search via RAG**: hybrid retrieval (full-text + vector merged via- Reciprocal Rank Fusion) with PageRank bias and 1-hop graph expansion.+- **Freshness scoring** per page, relative to the underlying code.+- **Semantic search via RAG**: hybrid retrieval merging full-text and vector+ results through Reciprocal Rank Fusion, with PageRank bias and a 1–2 hop graph+ expansion that walks the imports and projected-calls graph for flow-shaped+ questions.-A typical single-commit update touches 3–10 pages and completes in under 30-seconds; only the pages your change actually touched are regenerated.+A typical single-commit update regenerates only the handful of pages your change+actually touched.---## Decision Intelligence-Architectural decisions mined from **eight sources**: ADR files (Nygard/MADR),-CHANGELOG entries, PR and squash-commit bodies, inline markers, git archaeology,-README/docs, centrality-bounded code comments, and the LLM doc-generation pass-itself. These are linked to the graph nodes they govern and tracked for-staleness as code evolves.+Architectural decisions mined at index time from **five sources**: ADR files+(Nygard/MADR), PR and squash-commit bodies, inline markers, git archaeology, and+centrality-bounded code comments. Two further capture paths exist for humans and+agents: `repowise decision add`, and mining your Claude Code or Codex session+transcripts.```python# WHY: JWT chosen over sessions — API must be stateless for k8s horizontal scaling@@ -119,155 +137,188 @@ staleness as code evolves.# TRADEOFF: Accepted eventual consistency in preferences for write throughput```-Every decision is **evidence-backed**: each rationale traces to a verbatim-source span (ADR quote, commit body, code comment), and an anti-hallucination-substring gate stamps each as **exact / fuzzy / unverified**: corroborating-sources raise confidence rather than overwrite each other.--Decisions form a **graph**: typed edges (`supersedes` / `refines` /-`relates_to` / `conflicts_with`) let `get_why()` answer *"why is auth structured-this way?"* with a lineage chain (sessions -> JWT -> OAuth2). **No edges are-written today**: the only detector that produced them scoped a conflict by-similarity, which does not scope, so it is off and the edges it wrote have been-removed (see `docs/layers/DECISIONS.md`). Lineage stays empty until a-structural detector replaces it. Retirement itself still works — the-diff-driven pass on `repowise update` marks a decision a new commit reversed,-and `repowise decision deprecate --superseded-by` records the successor on the-record.--These structured records surface everywhere your agent already looks:-`get_why()` for the full archaeology, governing decisions in `get_context()`, a-`governance_risk` flag in `get_risk()` PR review, a Key Decisions section in-`get_overview()`, and `ungoverned_hotspot` / `stale_governance` /-`contradictory_decision` findings in the code-health layer.+Every decision is **evidence-backed**: each rationale traces to a verbatim source+span, and an anti-hallucination substring gate stamps it **exact**, **fuzzy** or+**unverified**. Corroborating sources raise confidence rather than overwrite each+other.++**On lineage, precisely.** Decisions have a typed-edge schema+(`supersedes` / `refines` / `relates_to` / `conflicts_with`), but **no edges are+written today**. The only detector that produced them scoped conflicts by+similarity, which does not scope, so it is disabled and the edges it wrote were+removed. Lineage stays empty until a structural detector replaces it. What does+work: the diff-driven pass on `repowise update` marks decisions a new commit+reversed, and `repowise decision deprecate --superseded-by` records the successor+on the record itself. See [`DECISIONS.md`](DECISIONS.md).++These records surface everywhere your agent already looks: `get_why()` for the+archaeology, governing decisions in `get_context()`, a `governance_risk` flag in+`get_risk()` PR review, a Key Decisions section in `get_overview()`, and the+`ungoverned_hotspot` / `stale_governance` / `contradictory_decision` findings in+the code-health layer.```bash-repowise decision add # guided interactive capture (~90 seconds)-repowise decision confirm # review auto-proposed decisions from git history+repowise decision add # guided interactive capture+repowise decision confirm # review auto-proposed decisionsrepowise decision health # stale, conflicting, ungoverned hotspots```+The "why" usually walks out the door: when a teammate leaves, or when you reopen+your own repo six months later. This keeps it in the codebase.++---++## Code Health Intelligence++repowise scores **every file 1–10** on three co-equal signals — defect risk,+maintainability, and performance risk — from a roster of **49 deterministic+detectors**, of which only **26 are permitted to move the defect number**. Pure+static analysis over tree-sitter and git data, budgeted (and CI-tested) to+finish in **under 30 seconds on a 3,000-file repo**.++The defect weights are **calibrated offline against a real bug corpus, not+hand-tuned**: every file scored at a commit preceding the bug window (no+leakage), an L2-logistic fit with NLOC as an explicit control so a marker only+earns weight for defect lift *beyond* file size. Only the learned constants+ship; the runtime stays fully deterministic.++Validated leakage-free across **21 repositories, 9 languages, 2,826 files** at+mean ROC AUC **0.737**, and **0.76–0.78** on the public PROMISE/jEdit dataset+that played no part in calibration.++It does not stop at scoring — the layer closes the loop into concrete,+graph-aware refactoring plans an agent can execute: Extract Class, Extract+Method, Extract Helper, Move Method, Break Cycle and Split File, each with its+plan, recovered impact and blast radius.++```bash+repowise health # KPIs + lowest-scoring files+repowise health --refactoring-targets # ranked by impact / effort+repowise health --trend # snapshots + declining-health alerts```-repowise decision health- 2 stale decisions- → "JWT over sessions" — auth/service.ts rewritten 3 months ago, decision may be outdated- → "EventBus in-process only" — 8 of 14 governed files changed since recorded+Full guide, the calibration story and the head-to-head against CodeScene:+[`CODE_HEALTH.md`](CODE_HEALTH.md).- 1 conflict- → payments/: two decisions with overlapping scope and contradictory rationale+---++## Change Risk- 1 ungoverned hotspot- → payments/processor.ts — 47 commits/month, no architectural decisions recorded+A calibrated 0–10 defect-risk score for **a whole commit or `base..head`+range**, computed from the diff's shape against the live checkout — no index+lookup, no model call. Distinct from `get_risk()`, which scores indexed files by+path. Lead with `risk_percentile`, which ranks the change against sampled recent+commits in the same repo.++```bash+repowise risk HEAD+repowise risk main..feature-branch```-The "why" usually walks out the door: when a teammate leaves, or when you-reopen your own repo six months later. Decision intelligence keeps it in the-codebase.+Reference: [`CHANGE_RISK.md`](CHANGE_RISK.md).----## Code Health Intelligence+## Test Intelligence-repowise computes a **1–10 health score for every file** from **25 deterministic-markers**: McCabe complexity, deep nesting, brain methods, class cohesion-(LCOM4), god classes, native Rabin–Karp clone detection, untested hotspots,-function-level churn, code-age volatility, ownership dispersion, change entropy,-co-change scatter, prior-defect history, test-quality smells, and more.+Which tests actually exercise the code you changed, which changed files have no+guarding test at all, and per-file coverage merged across every test that+touches it. Coverage ingests from LCOV, Cobertura, Clover or normalized JSON and+feeds the code-health coverage markers.-**Zero LLM calls. Zero cloud requirement. Zero new runtime dependencies**:-pure Python over tree-sitter and git data, designed to finish in under 30-seconds on a 3,000-file repo.+The practical payoff is `tests_to_run` in `get_risk()` PR mode: a+coverage-backed list rather than a guess, so an agent runs the tests that can+actually catch its change.-The marker **weights are calibrated offline against a real defect corpus, not-hand-tuned**: each file is scored at the pre-window commit (T0, no leakage) and-an L2-logistic regression (with NLOC as an explicit control) fits each-marker's defect lift *beyond* file size. Only the learned constants ship; the-runtime stays fully deterministic.+Reference: [`TEST_INTELLIGENCE.md`](TEST_INTELLIGENCE.md).-The same marker stream produces three orthogonal signals: defect risk-(the calibrated headline number), maintainability, and performance risk. These-are co-equal views, never blended into one number. It does not stop at scoring:-the layer closes the loop into concrete, graph-aware refactoring plans-(see below) an agent can execute.+----```bash-repowise health # KPIs + lowest-scoring files-repowise coverage add cov.lcov # ingest coverage, light up untested-hotspot-repowise health --refactoring-targets # ranked by impact / effort-repowise health --trend # last 10 snapshots + declining/predicted-decline alerts-repowise status # one-line summary in the status report-```+## Bug History-- **Coverage ingestion**: LCOV, Cobertura, Clover, or normalized JSON light up- the test-coverage markers (`untested_hotspot`, `coverage_gap`,- `coverage_gradient`).-- **Trend tracking**: a rolling 50-row snapshot history powers `Declining- Health` and `Predicted Decline` alerts.-- **Refactoring plans**: deterministic, structured, **graph-aware**: Extract- Class (LCOM4 cohesion split), Extract Helper (clone dedup), Move Method (feature- envy), Break Cycle (minimum feedback arc set), and Split File (modularity-gated- module decomposition with the import-rewrite blast radius), each carrying its- concrete plan, recovered impact, and blast radius. Ranked by `impact ×- centrality × blast radius`, on the dashboard **Refactoring** tab, via `repowise- health --refactoring-targets`, and via `get_health(include=["refactoring"])`. An- opt-in LLM pass expands any plan into generated code + a diff. See- [`docs/layers/REFACTORING.md`](REFACTORING.md).-- **Per-file overrides** via `.repowise/health-rules.json`.--Validated against real defect history; see-[`docs/layers/CODE_HEALTH.md`](CODE_HEALTH.md) for the full user guide, the per-marker-reference, and the calibration story, and-[repowise-bench](https://github.com/repowise-dev/repowise-bench) for the-reproducible defect-prediction and head-to-head benchmarks.+Bug-fix commits attributed back to the files and functions they repaired, giving+each file a fix count, a last-fixed age, and a "bug magnet" flag. This is why the+generated `CLAUDE.md` orders its *files that need care* by **bug-fix history+first, then churn** — a file that keeps breaking is a better warning than a file+that merely changes often.++Reference: [`BUG_HISTORY.md`](BUG_HISTORY.md).----## Proactive context enrichment: hooks+## Security-Most MCP tools are passive: the agent has to know to call them. repowise hooks-are active. They inject graph context into every search automatically, so-agents are smarter even when they don't explicitly ask for help. Hooks are-installed automatically during `repowise init`.+A static security scan over the working tree **and the full history**, so a+finding carries the commit that introduced it and its author, not just a line+number. Findings are idempotent across re-scans and surface through+`repowise security`, the REST API and the dashboard.-### PreToolUse: every search gets graph context+---++## Dead code detection-When your AI agent runs `Grep` or `Glob`, repowise intercepts the call and-enriches it with the top 3 related files, found via multi-signal search (symbol-name match, file-path match, full-text search on wiki content), ranked by-relevance then PageRank. No LLM calls. No network. Pure local SQLite queries.+Pure graph traversal and SQL. No model calls.```-[repowise] 3 related file(s) found:+repowise dead-code- src/core/ingestion/graph.py- Symbols: class:GraphBuilder, method:__init__, method:build- Imported by: src/core/ingestion/__init__.py- Uses: src/core/analysis/communities.py, src/core/analysis/execution_flows.py+ 23 findings · 4 safe to delete++ ✓ utils/legacy_parser.ts file 1.00 safe to delete+ ✓ auth/session.ts file 0.92 safe to delete+ ✓ helpers/formatDate export 0.71 safe to delete+ ✗ analytics/v1/tracker.ts file 0.41 recent activity — review first```-### PostToolUse: auto-detect stale wiki+Conservative by design. `safe_to_delete` requires confidence ≥ 0.70 and excludes+17 dynamically-loaded naming patterns (`*Plugin`, `*Handler`, `*Middleware`,+`register_*`, `on_*`, `*_route`, `*_callback` and more). Dynamic-import+detection and a per-language framework-convention registry further cut false+positives. repowise surfaces candidates; engineers decide.-After a successful `git commit`, repowise checks whether the wiki is out of date-and notifies the agent:+Reference: [`DEAD_CODE.md`](DEAD_CODE.md).-```-[repowise] Wiki is stale — last indexed at commit a1b2c3d4, HEAD is now f9a0499b.-Run `repowise update` to refresh documentation and graph context.-```+---++## Proactive context enrichment: hooks-> **Related capability:** [Distill](../agent/DISTILL.md) reuses these layers' index+Most MCP tools are passive — the agent has to know to call them. repowise hooks+are active: they act on what the agent is already doing. No LLM calls, no+network, pure local SQLite queries. Installed automatically during+`repowise init`.++There is deliberately **no unconditional pre-search injection**. An earlier+version enriched every `Grep`/`Glob` before it ran and was removed: it added+noise on the majority of searches where the agent had already found what it+wanted. What ships now fires on evidence that the agent needs help:++- **Zero-result rescue** — a grep found nothing, so wiki full-text, fuzzy symbol+ and decision matches surface the closest real hit.+- **Flood digest and triage** — a search returning far too much is replaced with+ a compact per-file digest of match counts and anchor lines, plus the top files+ by PageRank.+- **Wrong-path rescue** — a failed `Read`/`Edit`/`Write` path is resolved against+ the index to the file you almost certainly meant.+- **Read skeleton** — an unbounded read of a large indexed file is served as its+ skeleton with 1-indexed ranges, once per file per session.+- **Stale-read notice** — a file was edited after this session read it earlier.+- **Decision injection** — editing a file with a governing decision surfaces it+ in one line.+- **Session start** — index freshness, the trust protocol, and standing+ decisions.+- **Post-commit staleness** — after a successful commit, merge, rebase,+ cherry-pick or pull, a notice if the wiki has drifted from HEAD.++Claude Code and Codex are both supported.++> **Related capability:** [Distill](../agent/DISTILL.md) reuses the index> (symbol bounds, centrality, hotspots) to compress noisy command output and-> large file reads before the agent sees them, a capability built *on* the-> five layers, not a sixth layer.+> large reads before the agent sees them — built *on* the layers, not a layer.---## Auto-sync: five ways to stay current-repowise keeps your intelligence layers in sync with your code. Pick the method-that fits your workflow:-| Method | Command | Best for ||--------|---------|----------|| **Post-commit hook** | `repowise hook install` | Set-and-forget local development |@@ -277,71 +328,42 @@ that fits your workflow:| **Polling fallback** | Automatic with `repowise serve` | Safety net for missed webhooks |```bash-repowise hook install # install post-commit hook (current repo)-repowise hook install --workspace # install for all workspace repos-repowise hook status # check if hooks are installed-repowise watch # or use the file watcher (single repo)-repowise watch --workspace # all workspace repos+repowise hook install # post-commit hook (current repo)+repowise hook install --workspace # all workspace repos+repowise watch # or use the file watcher```-A typical single-commit update touches 3–10 pages and completes in under 30-seconds. Full guide: [`docs/scale/AUTO_SYNC.md`](../scale/AUTO_SYNC.md).+Updates are incremental: only the pages your change actually touched are+regenerated. Full guide: [`AUTO_SYNC.md`](../scale/AUTO_SYNC.md).---## Auto-generated CLAUDE.mdAfter every `repowise init` and `repowise update`, repowise regenerates your-`CLAUDE.md` from actual codebase intelligence, not a template. No LLM calls.-Under 5 seconds.+`CLAUDE.md` from actual codebase intelligence, not a template. No LLM calls. An+`AGENTS.md` generator shares the same pipeline.```bashrepowise generate-claude-md```-The generated section includes: architecture summary, module map, hotspot-warnings, ownership map, hidden coupling pairs, active architectural decisions,-and dead-code candidates. A user-owned section at the top is never touched.+The generated section includes: index freshness and a `_meta` explainer, how to+work in this repo, a **trust protocol** stating when a served result may be used+without re-reading, the MCP tool table, an architecture summary, key modules,+entry points, **files that need care** (ordered by bug-fix history, then churn),+code health across all three signals, standing architectural decisions, and your+build/test commands. A user-owned section at the top is never touched.```markdown-<!-- REPOWISE:START — managed automatically, do not edit -->+<!-- REPOWISE:START — Do not edit below this line. Auto-generated by Repowise. -->## ArchitectureMonorepo with 4 packages. Entry points: api/server.ts, cli/index.ts.-## Hotspots — handle with care-- payments/processor.ts — 47 commits/month, high complexity, primary owner: @sarah-- shared/events/EventBus.ts — 23 dependents, co-changes with all service listeners+## Files that need care (bug-fix history first, then churn)+- payments/processor.ts — 19 bug fixes, last fix 2 days ago (bug magnet); 23 commits/90d-## Active architectural decisions+## Standing decisions (ask get_why before diverging)- JWT over sessions (auth/service.ts) — stateless required for k8s horizontal scaling-- CircuitBreaker on all external calls — after payment provider outages in Q3 2024--## Hidden coupling (no import link, but change together)-- auth.ts ↔ middleware/session.ts — co-changed 31 times<!-- REPOWISE:END -->```-------## Dead code detection--Pure graph traversal and SQL. No LLM calls. Completes in under 10 seconds for-any repo size.--```-repowise dead-code-- 23 findings · 4 safe to delete-- ✓ utils/legacy_parser.ts file 1.00 safe to delete- ✓ auth/session.ts file 0.92 safe to delete- ✓ helpers/formatDate export 0.71 safe to delete- ✗ analytics/v1/tracker.ts file 0.41 recent activity — review first-```--Conservative by design. `safe_to_delete` requires confidence ≥ 0.70 and excludes-dynamically-loaded patterns (`*Plugin`, `*Handler`, `*Adapter`, `*Middleware`).-Dynamic-import detection (`importlib.import_module()`, `__import__()`) and-framework awareness (Flask/FastAPI/Django/Rails/Laravel/TYPO3 routes and-convention files) further reduce false positives. repowise surfaces candidates.-Engineers decide.diff --git a/docs/reference/CLI_REFERENCE.md b/docs/reference/CLI_REFERENCE.mdindex 08f67c4d5..01225cebf 100644--- a/docs/reference/CLI_REFERENCE.md+++ b/docs/reference/CLI_REFERENCE.md@@ -584,7 +584,7 @@ repowise impacted-tests main..HEAD --format list | xargs pytest### `repowise health [PATH]`-Compute per-file code-health scores from 25 deterministic markers (McCabe complexity, nesting, brain methods, LCOM4 cohesion, god classes, native clone detection, untested hotspots, coverage gradient, function/ownership/churn/change-entropy organizational risk, test-quality smells, and more). Zero LLM calls by default, pure Python over tree-sitter + git data. See [`docs/layers/CODE_HEALTH.md`](../layers/CODE_HEALTH.md) for the user guide and [`docs/architecture/code-health.md`](../architecture/code-health.md) for the internals.+Compute per-file code-health scores from 49 deterministic detectors (McCabe complexity, nesting, brain methods, LCOM4 cohesion, god classes, native clone detection, untested hotspots, coverage gradient, function/ownership/churn/change-entropy organizational risk, test-quality smells, and more). Zero LLM calls by default, pure Python over tree-sitter + git data. See [`docs/layers/CODE_HEALTH.md`](../layers/CODE_HEALTH.md) for the user guide and [`docs/architecture/code-health.md`](../architecture/code-health.md) for the internals.**Options:**diff --git a/plugins/claude-code/README.md b/plugins/claude-code/README.mdindex 931e77023..c91d5e492 100644--- a/plugins/claude-code/README.md+++ b/plugins/claude-code/README.md@@ -39,7 +39,7 @@ MCP tools and skills activate automatically.Graph (tree-sitter dependency graph, 16 languages) · Git (hotspots, ownership,co-change, bus factor) · Docs (LLM-generated wiki + semantic search) · Decisions-(architectural rationale mined from eight sources) · Code Health (1–10+(architectural rationale mined from five sources) · Code Health (1–10defect-validated score from deterministic markers).### Slash commandsdiff --git a/plugins/claude-code/skills/architectural-decisions/SKILL.md b/plugins/claude-code/skills/architectural-decisions/SKILL.mdindex 19b8d4028..1b40b33d6 100644--- a/plugins/claude-code/skills/architectural-decisions/SKILL.md+++ b/plugins/claude-code/skills/architectural-decisions/SKILL.md@@ -22,11 +22,11 @@ Repowise captures architectural decisions — the *why* behind how code is builttarget-anchored search; decisions touching the targets get boosted.4. `get_why()` — the decision-health dashboard.-Decisions are mined from eight sources (ADR files, CHANGELOG, PR/commit bodies,-inline markers, git archaeology, README/docs, code comments, and the doc pass)-and linked by `supersedes` / `refines` / `relates_to` / `conflicts_with` edges,-so a single answer can return a whole lineage chain. When no decision exists for-a path, `get_why` falls back to git archaeology so the call is never empty.+Decisions are mined from five sources (ADR files, PR and squash-commit bodies,+inline markers, git archaeology, and centrality-bounded code comments). Each+rationale traces to a verbatim source span, stamped exact / fuzzy / unverified.+When no decision exists for a path, `get_why` falls back to git archaeology so+the call is never empty.## When the user asks "why is X built this way?"diff --git a/plugins/codex/skills/architectural-decisions/SKILL.md b/plugins/codex/skills/architectural-decisions/SKILL.mdindex 48bf79b4c..bf3c4a2a6 100644--- a/plugins/codex/skills/architectural-decisions/SKILL.md+++ b/plugins/codex/skills/architectural-decisions/SKILL.md@@ -16,11 +16,11 @@ Repowise captures architectural decisions: the rationale behind how code is builtarget-anchored search; decisions touching the targets get boosted.4. `get_why()` — the decision-health dashboard.-Decisions are mined from eight sources (ADR files, CHANGELOG, PR/commit bodies,-inline markers, git archaeology, README/docs, code comments, and the doc pass)-and linked by `supersedes` / `refines` / `relates_to` / `conflicts_with` edges,-so a single answer can return a whole lineage chain. When no decision exists for-a path, `get_why` falls back to git archaeology so the call is never empty.+Decisions are mined from five sources (ADR files, PR and squash-commit bodies,+inline markers, git archaeology, and centrality-bounded code comments). Each+rationale traces to a verbatim source span, stamped exact / fuzzy / unverified.+When no decision exists for a path, `get_why` falls back to git archaeology so+the call is never empty.## When The User Asks Why Something Existsdiff --git a/scripts/gen_readme_hero.py b/scripts/gen_readme_hero.pyindex 49baee0c2..9329b1e22 100644--- a/scripts/gen_readme_hero.py+++ b/scripts/gen_readme_hero.py@@ -90,7 +90,7 @@ add(f'<text x="{PAD}" y="{80}" class="sub">Index once, in seconds, no LLM requirTOP = 104# ============ A: Code Health (large) ======================================ax, ay, aw, ah = PAD, TOP, 748, 352-card(ax, ay, aw, ah, "CODE HEALTH", "25 markers · zero LLM · <30s")+card(ax, ay, aw, ah, "CODE HEALTH", "49 detectors · zero LLM · <30s")# KPI chipskpis = [("Defect risk", "7.5", "/10", WARNING), ("Maintainability", "8.6", "/10", SUCCESS), ("Perf risks", "268", "", PLUM)]