Code graph and memory for your coding agents, in one Rust program. For Claude Code, Codex, Cursor and any MCP client.
Rust agent-memoryai-agentsclaude-codecode-analysiscode-graphcode-intelligencecode-searchcodexcoding-agentscontext-engineeringcursordeveloper-toolsgraphragknowledge-graphlocal-firstmcpmcp-servermodel-context-protocolrusttree-sitter
<!doctype html>
<html lang="en">
<head>
<meta charset="utf-8">
<meta name="viewport" content="width=device-width, initial-scale=1">
<title>Graphicle demo</title>
<link rel="stylesheet" href="/style.css">
</head>
<body>
<header>
<h1>Graphicle <span class="dim">on Muse Spark</span></h1>
<p class="lede">Ask a repository far larger than a 1M-token window, or point at a pixel of a screenshot and get the code that draws it.</p>
</header>
<main>
<div class="bar">
<label>Repository
<select id="repo"></select>
</label>
<span id="commit" class="dim"></span>
<nav role="tablist">
<button role="tab" id="tab-ask" aria-selected="true">Ask</button>
<button role="tab" id="tab-pixel" aria-selected="false">Point at a pixel</button>
</nav>
</div>
<section id="ask" class="panel">
<form id="ask-form">
<textarea id="question" rows="3" maxlength="600" placeholder="How does the scheduler pick a node for a pod?"></textarea>
<div class="row">
<button type="submit" id="ask-button">Ask</button>
<span id="ask-status" class="dim" aria-live="polite"></span>
</div>
</form>
<div id="curated" class="chips" aria-label="Questions with recorded answers"></div>
<div id="ask-result" class="result" hidden>
<div id="answer" class="answer"></div>
<h3>Sources</h3>
<ul id="citations" class="links"></ul>
<h3>The pack</h3>
<dl id="pack" class="facts"></dl>
<h3>Cost</h3>
<p id="ask-cost" class="facts-line"></p>
</div>
</section>
<section id="pixel" class="panel" hidden>
<p class="dim">Pixel to code runs on the vscode repository: drop a screenshot of VS Code, then click an element.</p>
<label id="drop" class="drop" tabindex="0">
<input id="file" type="file" accept="image/png,image/jpeg,image/webp,image/gif">
<span>Drop a screenshot here, or choose a file (PNG, JPEG, WebP or GIF, at most 8 MB)</span>
</label>
<div id="stage" class="stage" hidden>
<img id="shot" alt="Your screenshot: click an element">
<div id="box" class="box" hidden></div>
<div id="dot" class="dot" hidden></div>
</div>
<p id="pixel-status" class="dim" aria-live="polite"></p>
<div id="pixel-result" class="result" hidden>
<p id="read" class="read"></p>
<ol id="candidates" class="candidates"></ol>
<p id="pixel-cost" class="facts-line"></p>
</div>
</section>
</main>
<footer>
<p>A read-only demo. Questions and screenshots are not stored: the spend log keeps the time, model, tokens and cost of each call. <span id="budget"></span></p>
</footer>
<script src="/app.js"></script>
</body>
</html>