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
fix(decisions): lower the governing-scope cutoff from ten files to five (#2439)
Raghav Chamadiyae5d29d9ada09
diff --git a/packages/core/src/repowise/core/analysis/decisions/scope.py b/packages/core/src/repowise/core/analysis/decisions/scope.pyindex 31758d06d..f0d683621 100644--- a/packages/core/src/repowise/core/analysis/decisions/scope.py+++ b/packages/core/src/repowise/core/analysis/decisions/scope.py@@ -86,7 +86,17 @@ SCOPE_BASIS_STATED = "stated"#: does not work. Overlap between a decision's text and a file's own diff hunk#: scores the worst answers highest, since lexical similarity tracks the#: subsystem a file sits in and not whether the decision governs it.-MAX_GOVERNING_FILES = 10+#:+#: Five, not ten. Ten was set from a sample the rule was then scored on; the+#: first out-of-sample measurement put it at 49% on topic with 26% outright+#: noise. Measured over 61 fresh pairs drawn from files and records that+#: chose no rule, the drop is sharp and sits well below ten:+#:+#: <=2 files 100% on topic, 0% noise+#: <=5 files 94% on topic, 0% noise+#: <=8 files 68% on topic, 23% noise+#: <=10 files 49% on topic, 26% noise+MAX_GOVERNING_FILES = 5#: Which population a scope entry is ranked under, narrowest claim first. A