anvil-engineering-workbench 1.0.1 → 1.0.4

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (53) hide show
  1. package/CHANGELOG.md +41 -0
  2. package/README.md +20 -11
  3. package/SECURITY.md +1 -1
  4. package/dist/cli.js +276 -183
  5. package/dist/cli.js.map +1 -1
  6. package/dist/core/agent-config-release.js +1 -1
  7. package/dist/core/agent-config-versioning.js +1 -1
  8. package/dist/core/agent-evaluation-intelligence.js +1 -1
  9. package/dist/core/agent-evolution.js +1 -1
  10. package/dist/core/agent-experiment-studio.js +2 -2
  11. package/dist/core/agent-optimization.js +1 -1
  12. package/dist/core/agent-security-audit.js +1 -1
  13. package/dist/core/architecture-map.js +181 -20
  14. package/dist/core/architecture-map.js.map +1 -1
  15. package/dist/core/benchmark-engine.js +1 -1
  16. package/dist/core/benchmark-matrix.js +1 -1
  17. package/dist/core/benchmark-regression.js +1 -1
  18. package/dist/core/change-intelligence.js +1 -1
  19. package/dist/core/code-intelligence.js +1 -1
  20. package/dist/core/code-review.js +1 -1
  21. package/dist/core/dashboard-server.js +48 -44
  22. package/dist/core/dashboard-server.js.map +1 -1
  23. package/dist/core/dependency-impact.js +1 -1
  24. package/dist/core/dependency-intelligence.js +1 -1
  25. package/dist/core/doctor.d.ts +6 -0
  26. package/dist/core/doctor.js +175 -19
  27. package/dist/core/doctor.js.map +1 -1
  28. package/dist/core/engineering-risk.js +1 -1
  29. package/dist/core/git-intelligence.js +1 -1
  30. package/dist/core/intelligence-timeline.js +1 -1
  31. package/dist/core/patch-validation.js +1 -1
  32. package/dist/core/project-cleaner.d.ts +78 -0
  33. package/dist/core/project-cleaner.js +847 -0
  34. package/dist/core/project-cleaner.js.map +1 -0
  35. package/dist/core/project-health.js +1 -1
  36. package/dist/core/project-inspector.js +2 -2
  37. package/dist/core/quality-audit.js +2 -2
  38. package/dist/core/regression-intelligence.js +1 -1
  39. package/dist/core/remediation-planner.js +4 -4
  40. package/dist/core/review-history.js +1 -1
  41. package/dist/core/sandbox-cleanup-audit.js +2 -2
  42. package/dist/core/sandbox-compliance.js +1 -1
  43. package/dist/core/sandbox-lifecycle.js +3 -3
  44. package/dist/core/sandbox-policy.js +1 -1
  45. package/dist/core/sandbox-runtime.js +1 -1
  46. package/dist/core/skill-registry.js +1 -1
  47. package/dist/core/verification.js +1 -1
  48. package/dist/index.d.ts +2 -1
  49. package/dist/index.js +2 -1
  50. package/dist/index.js.map +1 -1
  51. package/dist/mcp/server.js +38 -1
  52. package/dist/mcp/server.js.map +1 -1
  53. package/package.json +11 -3
@@ -70,13 +70,19 @@ const PAGE = String.raw `<!doctype html>
70
70
  <section class="grid" id="metrics"><div class="card"><div class="metric-label">Source files</div><div class="metric">—</div><div class="metric-sub">Scanning workspace</div></div><div class="card"><div class="metric-label">Directories</div><div class="metric">—</div><div class="metric-sub">Project structure</div></div><div class="card"><div class="metric-label">Dependencies</div><div class="metric">—</div><div class="metric-sub">Production + development</div></div><div class="card"><div class="metric-label">Saved runs</div><div class="metric">—</div><div class="metric-sub">Local execution history</div></div></section>
71
71
  <section class="section columns"><div class="card"><div class="section-head"><h2>Language composition</h2><span class="pill">PROJECT INTELLIGENCE</span></div><div id="languages" class="list"><div class="loading">Loading language data…</div></div></div><div class="card"><div class="section-head"><h2>Detected stack</h2><span class="pill">ENVIRONMENT</span></div><div id="stack" class="list"><div class="loading">Detecting frameworks…</div></div></div></section>
72
72
  <section class="section columns"><div class="card"><div class="section-head"><h2>Specialist agents</h2><span class="pill">AGENT REGISTRY</span></div><div id="agents"><div class="loading">Loading agent roles…</div></div></div><div class="card"><div class="section-head"><h2>Recent runs</h2><span class="pill">LOCAL HISTORY</span></div><div id="runs"><div class="loading">Loading saved runs…</div></div></div><div class="card"><div class="section-head"><h2>Project signals</h2><span class="pill">INSIGHTS</span></div><div id="signals" class="list"><div class="loading">Collecting project signals…</div></div></div></section>
73
- <section class="section card"><div class="section-head"><h2>Project Health Center</h2><div class="actions"><span class="pill">SNAPSHOT + TRENDS</span><button class="btn" id="refreshHealth">Refresh snapshot</button></div></div><div id="healthPanel" class="list"><div class="loading">Generating project health snapshot…</div></div></section>\n<section class="section columns"><div class="card"><div class="section-head"><h2>Regression intelligence</h2><div class="actions"><span class="pill">VERIFICATION TRENDS</span><button class="btn" id="refreshRegression">Refresh</button></div></div><div id="regressionPanel" class="list"><div class="loading">Comparing saved verification runs…</div></div></div><div class="card"><div class="section-head"><h2>Code review history</h2><div class="actions"><span class="pill">SAVED REVIEWS</span><button class="btn" id="refreshReviews">Refresh</button></div></div><div id="reviewPanel" class="list"><div class="loading">Loading saved review snapshots…</div></div></div></section>
74
- <section class="section columns"><div class="card"><div class="section-head"><h2>Engineering quality</h2><span class="pill">STATIC AUDIT</span></div><div id="qualityPanel" class="list"><div class="loading">Loading quality findings…</div></div></div><div class="card"><div class="section-head"><h2>Verification history</h2><span class="pill">REAL COMMAND RESULTS</span></div><div id="verificationPanel" class="list"><div class="loading">Loading verification history…</div></div></div></section>\n<section class="section card"><div class="section-head"><h2>Code search</h2><span class="pill">SYMBOL-AWARE SOURCE SEARCH</span></div><div class="actions"><input id="codeSearchQuery" type="search" placeholder="Search symbols or source text…" style="flex:1;min-width:0;background:#0b1018;color:var(--text);border:1px solid var(--line);border-radius:9px;padding:10px;font:inherit"><button class="btn" id="codeSearchButton">Search</button></div><div id="codeSearchMeta" class="run-meta">Searches source files only; results open in the workspace preview.</div><div id="codeSearchResults" class="list"><div class="empty">Enter at least two characters to search.</div></div></section>\n<section class="section card"><div class="section-head"><h2>Workspace explorer</h2><span class="pill">SAFE FILE PREVIEW</span></div><div class="columns"><div><div class="actions"><button class="btn" id="workspaceUp">↑ Up</button><span id="workspacePath" class="pill">/</span></div><div id="workspaceFiles" class="list" style="margin-top:10px"><div class="loading">Loading workspace files…</div></div></div><div><div id="workspacePreview" class="inspector"><div class="empty">Select a source file to preview it here.</div></div></div></div></section>
73
+ <section class="section card"><div class="section-head"><h2>Project Health Center</h2><div class="actions"><span class="pill">SNAPSHOT + TRENDS</span><button class="btn" id="refreshHealth">Refresh snapshot</button></div></div><div id="healthPanel" class="list"><div class="loading">Generating project health snapshot…</div></div></section>
74
+ <section class="section columns"><div class="card"><div class="section-head"><h2>Regression intelligence</h2><div class="actions"><span class="pill">VERIFICATION TRENDS</span><button class="btn" id="refreshRegression">Refresh</button></div></div><div id="regressionPanel" class="list"><div class="loading">Comparing saved verification runs…</div></div></div><div class="card"><div class="section-head"><h2>Code review history</h2><div class="actions"><span class="pill">SAVED REVIEWS</span><button class="btn" id="refreshReviews">Refresh</button></div></div><div id="reviewPanel" class="list"><div class="loading">Loading saved review snapshots…</div></div></div></section>
75
+ <section class="section columns"><div class="card"><div class="section-head"><h2>Engineering quality</h2><span class="pill">STATIC AUDIT</span></div><div id="qualityPanel" class="list"><div class="loading">Loading quality findings…</div></div></div><div class="card"><div class="section-head"><h2>Verification history</h2><span class="pill">REAL COMMAND RESULTS</span></div><div id="verificationPanel" class="list"><div class="loading">Loading verification history…</div></div></div></section>
76
+ <section class="section card"><div class="section-head"><h2>Code search</h2><span class="pill">SYMBOL-AWARE SOURCE SEARCH</span></div><div class="actions"><input id="codeSearchQuery" type="search" placeholder="Search symbols or source text…" style="flex:1;min-width:0;background:#0b1018;color:var(--text);border:1px solid var(--line);border-radius:9px;padding:10px;font:inherit"><button class="btn" id="codeSearchButton">Search</button></div><div id="codeSearchMeta" class="run-meta">Searches source files only; results open in the workspace preview.</div><div id="codeSearchResults" class="list"><div class="empty">Enter at least two characters to search.</div></div></section>
77
+ <section class="section card"><div class="section-head"><h2>Workspace explorer</h2><span class="pill">SAFE FILE PREVIEW</span></div><div class="columns"><div><div class="actions"><button class="btn" id="workspaceUp">↑ Up</button><span id="workspacePath" class="pill">/</span></div><div id="workspaceFiles" class="list" style="margin-top:10px"><div class="loading">Loading workspace files…</div></div></div><div><div id="workspacePreview" class="inspector"><div class="empty">Select a source file to preview it here.</div></div></div></div></section>
75
78
  <section class="section card" id="agentConfigStudio"><div class="section-head"><h2>Agent Configuration Studio</h2><span class="pill">WORKSPACE-OWNED MARKDOWN</span></div><p class="run-meta">Edit custom agent and skill instructions in .anvil/agents and .anvil/skills (or .forge/). Built-in agents stay read-only. Saving changes local workspace files only; it does not execute agents or automatically promote a configuration.</p><div class="columns"><div class="inspector"><label class="metric-label" for="configCategory">Configuration type</label><select id="configCategory" class="btn"><option value="agents">Custom agent</option><option value="skills">Skill</option></select><label class="metric-label" for="configFile">Existing configuration</label><select id="configFile" class="btn"><option value="">Create a new file…</option></select><label class="metric-label" for="configId">File ID</label><input id="configId" maxlength="64" placeholder="e.g. api-specialist" style="background:#0b1018;color:var(--text);border:1px solid var(--line);border-radius:9px;padding:10px;font:inherit"><label class="metric-label" for="configContent">Markdown instructions</label><textarea id="configContent" rows="12" maxlength="131072" placeholder="# Agent name&#10;&#10;Describe responsibilities, constraints and expected evidence…" style="width:100%;background:#0b1018;color:var(--text);border:1px solid var(--line);border-radius:9px;padding:10px;font:inherit"></textarea><div class="actions"><button class="btn" id="newConfigFile">New file</button><button class="btn" id="saveConfigFile">Save configuration</button><button class="btn" id="refreshConfigFiles">Refresh list</button><button class="btn" id="captureConfigSnapshot">Capture snapshot</button></div><div id="configMessage" class="run-meta" aria-live="polite">Changes are saved locally and remain reviewable in Git.</div></div><div><h3 style="margin-top:0">Configuration inventory</h3><div id="configInventory" class="list"><div class="loading">Loading custom agent and skill files…</div></div><h3>Provider readiness</h3><div id="providerReadiness" class="list"><div class="loading">Checking local provider configuration…</div></div><p class="run-meta">Shows credential presence and configured models only. It does not test provider connectivity or expose API keys.</p><h3>Benchmark-gated promotion</h3><div class="inspector"><label class="metric-label" for="releaseCandidate">Candidate snapshot ID</label><input id="releaseCandidate" placeholder="cfg-…" style="background:#0b1018;color:var(--text);border:1px solid var(--line);border-radius:9px;padding:10px;font:inherit"><label class="metric-label" for="releaseBaseline">Approved baseline snapshot ID</label><input id="releaseBaseline" placeholder="cfg-…" style="background:#0b1018;color:var(--text);border:1px solid var(--line);border-radius:9px;padding:10px;font:inherit"><label class="metric-label" for="releaseMatrix">Benchmark matrix ID</label><input id="releaseMatrix" placeholder="matrix-id" style="background:#0b1018;color:var(--text);border:1px solid var(--line);border-radius:9px;padding:10px;font:inherit"><button class="btn" id="prepareConfigRelease">Check regression gate & prepare release</button><div id="releaseMessage" class="run-meta" aria-live="polite">Only a release-ready benchmark with matching snapshot fingerprint can proceed.</div></div><div id="configReleases" class="list"><div class="loading">Loading release history…</div></div><h3>Version history</h3><div id="configVersions" class="list"><div class="loading">Loading snapshots…</div></div><div class="workflow-analysis"><b>Safety boundary</b><br>Only Markdown files with validated IDs under .anvil/agents (.forge/) or .anvil/skills (.forge/) can be saved. This editor cannot write arbitrary paths, run shell commands, or modify built-in agent definitions.</div></div></div></section>
76
- <section class="section card" id="workflowEditor"><div class="section-head"><h2>Visual workflow editor</h2><span class="pill">REUSABLE AGENT PIPELINES</span></div><p class="run-meta">Compose specialist tasks, connect dependencies and add human approval gates. Saving a workflow stores a local reusable definition; creating a draft does not execute agents.</p><div class="columns"><div class="inspector"><label class="metric-label" for="workflowName">Workflow name</label><input id="workflowName" maxlength="100" placeholder="e.g. Secure feature delivery" style="background:#0b1018;color:var(--text);border:1px solid var(--line);border-radius:9px;padding:10px;font:inherit"><label class="metric-label" for="workflowDescription">Description</label><input id="workflowDescription" maxlength="500" placeholder="What this pipeline is for" style="background:#0b1018;color:var(--text);border:1px solid var(--line);border-radius:9px;padding:10px;font:inherit"><div class="divider"></div><h3 style="margin:4px 0">Add specialist task</h3><label class="metric-label" for="workflowAgent">Agent</label><select id="workflowAgent" class="btn" style="text-align:left"><option value="planner">Planner</option><option value="architect">Architect</option><option value="code-reviewer">Code reviewer</option><option value="security-reviewer">Security reviewer</option><option value="test-engineer">Test engineer</option><option value="documentation-writer">Documentation writer</option><option value="performance-engineer">Performance engineer</option><option value="debugger">Debugger</option><option value="devops-engineer">DevOps engineer</option><option value="database-reviewer">Database reviewer</option><option value="accessibility-reviewer">Accessibility reviewer</option><option value="release-manager">Release manager</option></select><label class="metric-label" for="workflowTaskTitle">Task title</label><input id="workflowTaskTitle" maxlength="120" placeholder="Review API boundaries" style="background:#0b1018;color:var(--text);border:1px solid var(--line);border-radius:9px;padding:10px;font:inherit"><label class="metric-label" for="workflowTaskPrompt">Task instructions</label><textarea id="workflowTaskPrompt" maxlength="4000" rows="3" placeholder="Describe the specialist's job and expected evidence…" style="background:#0b1018;color:var(--text);border:1px solid var(--line);border-radius:9px;padding:10px;font:inherit"></textarea><label class="metric-label" for="workflowDependency">Depends on</label><select id="workflowDependency" class="btn" multiple size="4" style="text-align:left;min-height:100px"><option value="">No dependency (start independently)</option></select><div class="run-meta">Use Ctrl/Cmd to select multiple prerequisites for branching and join points.</div><label style="display:flex;gap:8px;align-items:center"><input id="workflowRequiresApproval" type="checkbox"> Require human approval before this task</label><div class="actions"><button class="btn" id="addWorkflowTask">Add task</button><button class="btn" id="clearWorkflowTasks">Clear tasks</button></div><div class="actions"><button class="btn" id="templateFeature">Load feature-delivery template</button><button class="btn" id="templateSecurity">Load security-audit template</button></div></div><div class="inspector"><div class="section-head"><h3 style="margin:0">Pipeline canvas</h3><span class="pill" id="workflowTaskCount">0 TASKS</span></div><div id="workflowCanvas" class="activity"><div class="empty">Add a specialist task to start composing your pipeline.</div></div><div class="actions"><button class="btn" id="applyWorkflowLinks">Apply selected links</button></div><div class="actions"><button class="btn" id="saveWorkflow">Save workflow</button><button class="btn" id="createWorkflowRun">Create draft run</button></div><div id="workflowMessage" class="run-meta" aria-live="polite"></div><div class="section-head" style="margin-top:18px"><h3 style="margin:0">Saved workflows</h3><button class="btn" id="refreshWorkflows">Refresh</button></div><div id="savedWorkflows" class="activity"><div class="loading">Loading reusable workflows…</div></div></div></div></section>\n<section class="section card"><div class="section-head"><div><h2>Evaluation history</h2><div class="run-meta">Saved deterministic rubric results · use <code>anvil evaluate RUNID</code> to score a run</div></div><span id="evaluationUpdated" class="pill">LOCAL REPORTS</span></div><div id="evaluationHistory" class="activity"><div class="loading">Loading evaluation reports…</div></div></section>\n<section class="section card"><div class="section-head"><div><h2>Benchmark Lab</h2><div class="run-meta">Compare recorded runs by provider and model · use <code>anvil benchmark RUN_ID</code> after evaluation</div></div><span id="benchmarkUpdated" class="pill">NO TRIALS</span></div><div id="benchmarkHistory" class="activity"><div class="loading">Loading benchmark history…</div></div></section>
79
+ <section class="section card" id="workflowEditor"><div class="section-head"><h2>Visual workflow editor</h2><span class="pill">REUSABLE AGENT PIPELINES</span></div><p class="run-meta">Compose specialist tasks, connect dependencies and add human approval gates. Saving a workflow stores a local reusable definition; creating a draft does not execute agents.</p><div class="columns"><div class="inspector"><label class="metric-label" for="workflowName">Workflow name</label><input id="workflowName" maxlength="100" placeholder="e.g. Secure feature delivery" style="background:#0b1018;color:var(--text);border:1px solid var(--line);border-radius:9px;padding:10px;font:inherit"><label class="metric-label" for="workflowDescription">Description</label><input id="workflowDescription" maxlength="500" placeholder="What this pipeline is for" style="background:#0b1018;color:var(--text);border:1px solid var(--line);border-radius:9px;padding:10px;font:inherit"><div class="divider"></div><h3 style="margin:4px 0">Add specialist task</h3><label class="metric-label" for="workflowAgent">Agent</label><select id="workflowAgent" class="btn" style="text-align:left"><option value="planner">Planner</option><option value="architect">Architect</option><option value="code-reviewer">Code reviewer</option><option value="security-reviewer">Security reviewer</option><option value="test-engineer">Test engineer</option><option value="documentation-writer">Documentation writer</option><option value="performance-engineer">Performance engineer</option><option value="debugger">Debugger</option><option value="devops-engineer">DevOps engineer</option><option value="database-reviewer">Database reviewer</option><option value="accessibility-reviewer">Accessibility reviewer</option><option value="release-manager">Release manager</option></select><label class="metric-label" for="workflowTaskTitle">Task title</label><input id="workflowTaskTitle" maxlength="120" placeholder="Review API boundaries" style="background:#0b1018;color:var(--text);border:1px solid var(--line);border-radius:9px;padding:10px;font:inherit"><label class="metric-label" for="workflowTaskPrompt">Task instructions</label><textarea id="workflowTaskPrompt" maxlength="4000" rows="3" placeholder="Describe the specialist's job and expected evidence…" style="background:#0b1018;color:var(--text);border:1px solid var(--line);border-radius:9px;padding:10px;font:inherit"></textarea><label class="metric-label" for="workflowDependency">Depends on</label><select id="workflowDependency" class="btn" multiple size="4" style="text-align:left;min-height:100px"><option value="">No dependency (start independently)</option></select><div class="run-meta">Use Ctrl/Cmd to select multiple prerequisites for branching and join points.</div><label style="display:flex;gap:8px;align-items:center"><input id="workflowRequiresApproval" type="checkbox"> Require human approval before this task</label><div class="actions"><button class="btn" id="addWorkflowTask">Add task</button><button class="btn" id="clearWorkflowTasks">Clear tasks</button></div><div class="actions"><button class="btn" id="templateFeature">Load feature-delivery template</button><button class="btn" id="templateSecurity">Load security-audit template</button></div></div><div class="inspector"><div class="section-head"><h3 style="margin:0">Pipeline canvas</h3><span class="pill" id="workflowTaskCount">0 TASKS</span></div><div id="workflowCanvas" class="activity"><div class="empty">Add a specialist task to start composing your pipeline.</div></div><div class="actions"><button class="btn" id="applyWorkflowLinks">Apply selected links</button></div><div class="actions"><button class="btn" id="saveWorkflow">Save workflow</button><button class="btn" id="createWorkflowRun">Create draft run</button></div><div id="workflowMessage" class="run-meta" aria-live="polite"></div><div class="section-head" style="margin-top:18px"><h3 style="margin:0">Saved workflows</h3><button class="btn" id="refreshWorkflows">Refresh</button></div><div id="savedWorkflows" class="activity"><div class="loading">Loading reusable workflows…</div></div></div></div></section>
80
+ <section class="section card"><div class="section-head"><div><h2>Evaluation history</h2><div class="run-meta">Saved deterministic rubric results · use <code>anvil evaluate RUNID</code> to score a run</div></div><span id="evaluationUpdated" class="pill">LOCAL REPORTS</span></div><div id="evaluationHistory" class="activity"><div class="loading">Loading evaluation reports…</div></div></section>
81
+ <section class="section card"><div class="section-head"><div><h2>Benchmark Lab</h2><div class="run-meta">Compare recorded runs by provider and model · use <code>anvil benchmark RUN_ID</code> after evaluation</div></div><span id="benchmarkUpdated" class="pill">NO TRIALS</span></div><div id="benchmarkHistory" class="activity"><div class="loading">Loading benchmark history…</div></div></section>
77
82
  <section class="section card"><div class="section-head"><div><h2>Regression intelligence</h2><div class="run-meta">Compare the latest matrix with a previous run using the same goal and evaluation suite</div></div><span id="regressionStatus" class="pill">CHECKING HISTORY</span></div><div id="benchmarkRegression" class="activity"><div class="loading">Analyzing benchmark history…</div></div><div class="actions"><button class="btn" id="refreshRegression">Refresh regression analysis</button></div></section>
78
83
  <section class="section card"><div class="section-head"><div><h2>Agent Performance</h2><div class="run-meta">Latest matrix · deterministic evaluation scorecards by agent and model</div></div><span id="agentPerformanceUpdated" class="pill">NO SCORECARDS</span></div><div id="agentPerformance" class="activity"><div class="loading">Loading agent scorecards…</div></div></section>
79
- <section class="section card"><div class="section-head"><div><h2>Skill Experiments</h2><div class="run-meta">Controlled same-model comparison · baseline with no skills versus treatment with selected skills</div></div><span id="skillExperimentsUpdated" class="pill">NO EXPERIMENTS</span></div><div id="skillExperiments" class="activity"><div class="loading">Loading controlled skill experiments…</div></div></section>\n<section class="section card"><div class="section-head"><div><h2>Experiment History & Leaderboard</h2><div class="run-meta">Historical candidate rankings · grouped by evaluation suite · estimates are not billing data</div></div><div class="actions"><span id="experimentHistoryUpdated" class="pill">LOADING HISTORY</span><button class="btn" id="refreshExperimentHistory">Refresh</button></div></div><div class="columns"><div><label class="metric-label" for="experimentSuiteFilter">Evaluation suite</label><input id="experimentSuiteFilter" placeholder="All suites" style="width:100%;background:#0b1018;color:var(--text);border:1px solid var(--line);border-radius:9px;padding:9px;font:inherit"></div><div><label class="metric-label" for="experimentCandidateFilter">Candidate</label><input id="experimentCandidateFilter" placeholder="All candidates" style="width:100%;background:#0b1018;color:var(--text);border:1px solid var(--line);border-radius:9px;padding:9px;font:inherit"></div></div><div id="experimentLeaderboard" class="activity" style="margin-top:14px"><div class="loading">Aggregating saved benchmark matrices…</div></div><div class="section" style="margin-top:18px"><div class="section-head"><div><h3>Visual candidate comparison</h3><div class="run-meta">Score and pass rate are normalized to 100 · duration and token values are estimates</div></div><span class="pill">TOP 8</span></div><div id="experimentComparison" class="activity"><div class="loading">Building comparison from saved benchmark history…</div></div></div><div id="experimentRegressions" class="activity" style="margin-top:12px"></div><div class="actions" style="margin-top:12px"><button class="btn" id="exportExperimentHistory">Export JSON</button></div><div id="experimentHistoryCaveats" class="run-meta" style="margin-top:10px"></div></section>
84
+ <section class="section card"><div class="section-head"><div><h2>Skill Experiments</h2><div class="run-meta">Controlled same-model comparison · baseline with no skills versus treatment with selected skills</div></div><span id="skillExperimentsUpdated" class="pill">NO EXPERIMENTS</span></div><div id="skillExperiments" class="activity"><div class="loading">Loading controlled skill experiments…</div></div></section>
85
+ <section class="section card"><div class="section-head"><div><h2>Experiment History & Leaderboard</h2><div class="run-meta">Historical candidate rankings · grouped by evaluation suite · estimates are not billing data</div></div><div class="actions"><span id="experimentHistoryUpdated" class="pill">LOADING HISTORY</span><button class="btn" id="refreshExperimentHistory">Refresh</button></div></div><div class="columns"><div><label class="metric-label" for="experimentSuiteFilter">Evaluation suite</label><input id="experimentSuiteFilter" placeholder="All suites" style="width:100%;background:#0b1018;color:var(--text);border:1px solid var(--line);border-radius:9px;padding:9px;font:inherit"></div><div><label class="metric-label" for="experimentCandidateFilter">Candidate</label><input id="experimentCandidateFilter" placeholder="All candidates" style="width:100%;background:#0b1018;color:var(--text);border:1px solid var(--line);border-radius:9px;padding:9px;font:inherit"></div></div><div id="experimentLeaderboard" class="activity" style="margin-top:14px"><div class="loading">Aggregating saved benchmark matrices…</div></div><div class="section" style="margin-top:18px"><div class="section-head"><div><h3>Visual candidate comparison</h3><div class="run-meta">Score and pass rate are normalized to 100 · duration and token values are estimates</div></div><span class="pill">TOP 8</span></div><div id="experimentComparison" class="activity"><div class="loading">Building comparison from saved benchmark history…</div></div></div><div id="experimentRegressions" class="activity" style="margin-top:12px"></div><div class="actions" style="margin-top:12px"><button class="btn" id="exportExperimentHistory">Export JSON</button></div><div id="experimentHistoryCaveats" class="run-meta" style="margin-top:10px"></div></section>
80
86
  <section class="section card"><div class="section-head"><div><h2>Adaptive Agent Recommendations</h2><div class="run-meta">Repeated rubric failures translated into reviewable improvements · no automatic prompt or skill edits</div></div><span id="agentOptimizationStatus" class="pill">ANALYZING</span></div><div id="agentOptimization" class="activity"><div class="loading">Analyzing agent failure patterns…</div></div></section>
81
87
  <section class="section card"><div class="section-head"><div><h2>Agent Evaluation Intelligence</h2><div class="run-meta">Historical scorecards, score trends, repeated failure signals and evidence-backed recommendations</div></div><span id="agentIntelligenceStatus" class="pill">SCORECARDS</span></div><div id="agentIntelligence" class="activity"><div class="loading">Loading agent evaluation intelligence…</div></div></section>
82
88
  <section class="section card"><div class="section-head"><div><h2>Agent observability</h2><div class="run-meta">Estimated usage and provider outcomes from saved execution traces · last 30 runs</div></div><div class="actions"><span id="observabilityUpdated" class="pill">LOCAL TRACE DATA</span><button class="btn" id="refreshObservability">Refresh</button></div></div><div id="observabilityMetrics" class="grid"><div class="loading">Loading execution telemetry…</div></div><div class="section columns" style="margin-top:14px"><div><h3>Per-agent utilization</h3><div id="agentObservability" class="activity"><div class="loading">Aggregating agent traces…</div></div></div><div><h3>Provider health</h3><div id="providerObservability" class="activity"><div class="loading">Aggregating provider traces…</div></div></div></div><div id="observabilityWarnings" class="run-meta" style="margin-top:12px"></div></section>
@@ -91,7 +97,7 @@ const PAGE = String.raw `<!doctype html>
91
97
  <footer class="footer"><span>ANVIL v1.0.0 · ENGINEERING INTELLIGENCE</span><span>Read-only dashboard · Data stays on this machine</span></footer>
92
98
  </main>
93
99
  <script>
94
- const el=id=>document.getElementById(id);
100
+ const el=id=>document.getElementById(id);const on=(id,event,handler)=>{const node=el(id);if(node)node.addEventListener(event,handler);};
95
101
  const esc=value=>String(value??"").replace(/[&<>"']/g,c=>({"&":"&amp;","<":"&lt;",">":"&gt;",'"':"&quot;","'":"&#39;"}[c]));
96
102
  async function api(path){const response=await fetch(path,{headers:{"accept":"application/json"}});if(!response.ok)throw new Error("Request failed ("+response.status+")");return response.json()}
97
103
  function connectSSE(){const es=new EventSource('/api/events');const pill=el("liveStatus");const setLive=at=>{pill.className="status live-pulse";pill.style.color="#4ade80";pill.style.borderColor="#2d6a1f";pill.textContent="● LIVE · "+new Date(at).toLocaleTimeString();};es.addEventListener("connected",e=>{try{const d=JSON.parse(e.data);setLive(d.at);}catch{}});es.addEventListener("run-updated",e=>{try{const d=JSON.parse(e.data);setLive(d.at||new Date().toISOString());loadRuns();loadActivity();if(selectedRunId)inspectRun(selectedRunId);}catch{}});es.addEventListener("heartbeat",e=>{try{const d=JSON.parse(e.data);setLive(d.at);}catch{}});es.onerror=()=>{pill.className="status";pill.style.color="";pill.style.borderColor="";pill.textContent="● LOCAL WORKSPACE";setTimeout(connectSSE,5000);es.close();};}
@@ -150,7 +156,7 @@ async function decideConfigRelease(id,decision){
150
156
  const reviewer=window.prompt("Reviewer label for "+decision+":","local-user");if(!reviewer)return;
151
157
  try{const response=await fetch("/api/config-releases/"+encodeURIComponent(id)+"/"+decision,{method:"POST",headers:{"content-type":"application/json"},body:JSON.stringify({reviewer})});const data=await response.json();if(!response.ok)throw new Error(data.error||"Release decision failed.");el("releaseMessage").textContent="Release "+data.id+" is "+data.status+". Approval updates the active approved configuration pointer; it does not deploy code.";await loadConfigReleases();await loadConfigVersions();}catch(error){el("releaseMessage").textContent=error.message;}
152
158
  }
153
- el("prepareConfigRelease").addEventListener("click",async()=>{
159
+ on("prepareConfigRelease","click",async()=>{
154
160
  const candidateVersionId=el("releaseCandidate").value.trim(),baselineVersionId=el("releaseBaseline").value.trim(),matrixId=el("releaseMatrix").value.trim();
155
161
  try{const response=await fetch("/api/config-releases",{method:"POST",headers:{"content-type":"application/json"},body:JSON.stringify({candidateVersionId,baselineVersionId,matrixId})});const data=await response.json();if(!response.ok)throw new Error(data.error||"Could not prepare release.");el("releaseMessage").textContent="Prepared "+data.id+" · benchmark gate "+data.benchmarkStatus+" · human approval required.";await loadConfigReleases();}catch(error){el("releaseMessage").textContent=error.message;}
156
162
  });
@@ -158,7 +164,7 @@ async function loadConfigVersions(){
158
164
  try{const versions=await api("/api/config-snapshots");el("configVersions").innerHTML=versions.length?versions.map(item=>'<div class="rowitem"><div><b>'+esc(item.id)+' · '+esc(item.status)+'</b><div class="run-meta">'+esc(item.createdAt)+' · '+esc(item.files.length)+' files · '+esc(item.note||"No note")+'</div></div><span class="pill">'+(item.status==="approved"?"APPROVED":"CANDIDATE — BENCHMARK REQUIRED")+'</span></div>').join(""):'<div class="empty">No snapshots yet. Capture one after editing agent or skill instructions.</div>';
159
165
  }catch(error){el("configVersions").innerHTML='<div class="error">'+esc(error.message)+'</div>';}
160
166
  }
161
- el("captureConfigSnapshot").addEventListener("click",async()=>{
167
+ on("captureConfigSnapshot","click",async()=>{
162
168
  const note=window.prompt("Short note describing this configuration snapshot:","Agent configuration update");if(note===null)return;
163
169
  try{const response=await fetch("/api/config-snapshots",{method:"POST",headers:{"content-type":"application/json"},body:JSON.stringify({note})});const data=await response.json();if(!response.ok)throw new Error(data.error||"Snapshot failed.");el("configMessage").textContent="Captured "+data.id+" with "+data.files.length+" configuration files.";await loadConfigVersions();}catch(error){el("configMessage").textContent=error.message;}
164
170
  });
@@ -176,11 +182,11 @@ async function openConfigFile(key){
176
182
  const [category,id]=key.split("/");
177
183
  try{const item=await api("/api/config-files?category="+encodeURIComponent(category)+"&id="+encodeURIComponent(id));el("configCategory").value=category;el("configId").value=id;el("configFile").value=id;el("configContent").value=item.content;el("configMessage").textContent="Loaded "+key+".md";}catch(error){el("configMessage").textContent=error.message;}
178
184
  }
179
- el("configCategory").addEventListener("change",()=>{el("configId").value="";el("configContent").value="";loadConfigFiles();});
180
- el("configFile").addEventListener("change",()=>{const id=el("configFile").value;if(id)openConfigFile(el("configCategory").value+"/"+id);else{el("configId").value="";el("configContent").value="";}});
181
- el("newConfigFile").addEventListener("click",()=>{el("configFile").value="";el("configId").value="";el("configContent").value="";el("configId").focus();});
182
- el("refreshConfigFiles").addEventListener("click",loadConfigFiles);
183
- el("saveConfigFile").addEventListener("click",async()=>{
185
+ on("configCategory","change",()=>{el("configId").value="";el("configContent").value="";loadConfigFiles();});
186
+ on("configFile","change",()=>{const id=el("configFile").value;if(id)openConfigFile(el("configCategory").value+"/"+id);else{el("configId").value="";el("configContent").value="";}});
187
+ on("newConfigFile","click",()=>{el("configFile").value="";el("configId").value="";el("configContent").value="";el("configId").focus();});
188
+ on("refreshConfigFiles","click",loadConfigFiles);
189
+ on("saveConfigFile","click",async()=>{
184
190
  const category=el("configCategory").value,id=el("configId").value.trim(),content=el("configContent").value;
185
191
  try{const response=await fetch("/api/config-files",{method:"POST",headers:{"content-type":"application/json"},body:JSON.stringify({category,id,content})});const data=await response.json();if(!response.ok)throw new Error(data.error||"Could not save configuration.");el("configMessage").textContent="Saved .forge/"+category+"/"+id+".md · "+data.bytes+" bytes. Run a fresh evaluation before approving a snapshot.";await loadConfigFiles();el("configFile").value=id;}catch(error){el("configMessage").textContent=error.message;}
186
192
  });
@@ -218,7 +224,7 @@ async function inspectRun(id){
218
224
  document.querySelectorAll(".run").forEach(item=>item.classList.toggle("selected",item.getAttribute("data-run-id")===id));
219
225
  }catch(error){el("runInspector").innerHTML='<div class="error">'+esc(error.message)+'</div>';}
220
226
  }
221
- async function loadRuns(){try{const data=await api("/api/runs");el("runs").innerHTML=data.length?data.slice(0,7).map(run=>'<div class="run" data-run-id="'+esc(run.id)+'" tabindex="0" role="button"><div class="run-title"><span class="run-goal">'+esc(run.goal)+'</span><span class="pill">'+esc(run.status)+'</span></div><div class="run-meta">'+esc(run.id)+' · '+esc(new Date(run.createdAt).toLocaleString())+' · '+run.tasks.length+' tasks</div></div>').join(""):'<div class="empty">No saved runs yet. Create one with <code>forge plan --goal "..."</code>.</div>';el("runs").querySelectorAll("[data-run-id]").forEach(item=>{const select=()=>inspectRun(item.getAttribute("data-run-id"));item.addEventListener("click",select);item.addEventListener("keydown",event=>{if(event.key==="Enter"||event.key===" "){event.preventDefault();select();}});});}catch(error){el("runs").innerHTML='<div class="error">'+esc(error.message)+'</div>';}}
227
+ async function loadRuns(){try{const data=await api("/api/runs");el("runs").innerHTML=data.length?data.slice(0,7).map(run=>'<div class="run" data-run-id="'+esc(run.id)+'" tabindex="0" role="button"><div class="run-title"><span class="run-goal">'+esc(run.goal)+'</span><span class="pill">'+esc(run.status)+'</span></div><div class="run-meta">'+esc(run.id)+' · '+esc(new Date(run.createdAt).toLocaleString())+' · '+run.tasks.length+' tasks</div></div>').join(""):'<div class="empty">No saved runs yet. Create one with <code>anvil plan --goal "..."</code>.</div>';el("runs").querySelectorAll("[data-run-id]").forEach(item=>{const select=()=>inspectRun(item.getAttribute("data-run-id"));item.addEventListener("click",select);item.addEventListener("keydown",event=>{if(event.key==="Enter"||event.key===" "){event.preventDefault();select();}});});}catch(error){el("runs").innerHTML='<div class="error">'+esc(error.message)+'</div>';}}
222
228
  async function loadActivity(){
223
229
  try{const runs=await api("/api/runs");el("activityUpdated").textContent="Updated "+new Date().toLocaleTimeString();
224
230
  if(!runs.length){el("activity").innerHTML='<div class="empty">No agent activity yet. Create a plan or run to populate this view.</div>';el("taskStatus").innerHTML='<div class="empty">No tasks recorded.</div>';return;}
@@ -227,7 +233,7 @@ el("activity").innerHTML=events.length?events.slice(-12).reverse().map(event=>'<
227
233
  el("taskStatus").innerHTML=latest.tasks.length?latest.tasks.map(task=>'<div class="task-card"><b>'+esc(task.title||task.id)+'</b><small>'+esc(task.agent||"Agent")+'</small><span class="pill">'+esc(task.status)+'</span></div>').join(""):'<div class="empty">No tasks recorded.</div>';
228
234
  }catch(error){el("activity").innerHTML='<div class="error">'+esc(error.message)+'</div>';}}
229
235
  async function loadDependencyIntelligence(){el("dependencyIntelList").innerHTML='<div class="loading">Reading local dependency manifests and lockfile…</div>';try{const data=await api("/api/dependency-intelligence");const s=data.summary;el("dependencyIntelSummary").innerHTML=metric("Direct dependencies",s.direct,"Production "+s.production+" · development "+s.development)+metric("Transitive packages",s.transitive,"From parsed npm lockfile")+metric("Vulnerabilities",data.sources.vulnerabilityReport===null?"N/A":s.vulnerabilities,"Imported report only")+metric("Outdated",data.sources.outdatedReport===null?"N/A":s.outdated,"Imported report only");el("dependencyIntelList").innerHTML=data.dependencies.length?data.dependencies.slice(0,30).map(item=>'<div class="activity-item"><div class="activity-main"><b>'+esc(item.name)+'</b><small>'+(item.direct?"Direct · ":"Transitive · ")+esc(item.kind)+' · '+esc(item.locked||item.requested)+' · license '+esc(item.license||"unknown")+'</small><small>'+esc(item.vulnerabilities.map(v=>v.severity.toUpperCase()+": "+v.title).join(" · ")||"No imported vulnerability findings")+(item.outdated?" · latest "+esc(item.outdated.latest):"")+'</small></div><span class="pill">'+esc(item.licenseClass.toUpperCase())+'</span></div>').join(""):'<div class="empty">No npm dependencies detected.</div>';el("dependencyIntelNotes").textContent=data.limitations.join(" · ");}catch(error){el("dependencyIntelList").innerHTML='<div class="error">'+esc(error.message)+'</div>';}}
230
- async function loadIntelligenceTimeline(){el("timelineHistory").innerHTML='<div class="loading">Loading snapshot history…</div>';try{const data=await api("/api/intelligence-timeline");const comparison=data.comparison;const snapshots=data.snapshots||[];if(!comparison){el("timelineSummary").innerHTML=metric("Snapshots",0,"No history yet")+metric("Trend","—","Capture with forge snapshot .")+metric("Latest risk","—","Not recorded")+metric("Latest commit","—","Not recorded");el("timelineHistory").innerHTML='<div class="empty">No snapshots yet. Run <code>forge snapshot .</code> to create the first baseline.</div>';el("timelineNotes").textContent="Snapshots stay local in .forge/intelligence and are not committed or uploaded.";return;}const current=comparison.current;const delta=comparison.delta;el("timelineSummary").innerHTML=metric("Snapshots",snapshots.length,"Saved locally")+metric("Trend",comparison.trend.toUpperCase(),delta?"Compared with previous":"Baseline")+metric("Risk index",current.metrics.score+"/100","Structural risk index")+metric("High hotspots",current.metrics.high,delta?"Change "+(delta.high>0?"+":"")+delta.high:"Current snapshot");el("timelineHistory").innerHTML=snapshots.slice(-12).reverse().map(item=>'<div class="activity-item"><div class="activity-main"><b>'+esc(item.generatedAt)+'</b><small>Risk '+item.metrics.score+'/100 · high '+item.metrics.high+' · '+item.metrics.hotspots+' hotspots · commit '+esc(item.commit||"unavailable")+'</small></div><span class="pill">'+esc(item.id)+'</span></div>').join("");el("timelineNotes").textContent="Trend: "+comparison.trend+". New top hotspots: "+(comparison.newHotspots.join(", ")||"none")+". No-longer-top hotspots: "+(comparison.resolvedHotspots.join(", ")||"none")+". Snapshots are local only."; }catch(error){el("timelineHistory").innerHTML='<div class="error">'+esc(error.message)+'</div>';}}
236
+ async function loadIntelligenceTimeline(){el("timelineHistory").innerHTML='<div class="loading">Loading snapshot history…</div>';try{const data=await api("/api/intelligence-timeline");const comparison=data.comparison;const snapshots=data.snapshots||[];if(!comparison){el("timelineSummary").innerHTML=metric("Snapshots",0,"No history yet")+metric("Trend","—","Capture with anvil snapshot .")+metric("Latest risk","—","Not recorded")+metric("Latest commit","—","Not recorded");el("timelineHistory").innerHTML='<div class="empty">No snapshots yet. Run <code>forge snapshot .</code> to create the first baseline.</div>';el("timelineNotes").textContent="Snapshots stay local in .forge/intelligence and are not committed or uploaded.";return;}const current=comparison.current;const delta=comparison.delta;el("timelineSummary").innerHTML=metric("Snapshots",snapshots.length,"Saved locally")+metric("Trend",comparison.trend.toUpperCase(),delta?"Compared with previous":"Baseline")+metric("Risk index",current.metrics.score+"/100","Structural risk index")+metric("High hotspots",current.metrics.high,delta?"Change "+(delta.high>0?"+":"")+delta.high:"Current snapshot");el("timelineHistory").innerHTML=snapshots.slice(-12).reverse().map(item=>'<div class="activity-item"><div class="activity-main"><b>'+esc(item.generatedAt)+'</b><small>Risk '+item.metrics.score+'/100 · high '+item.metrics.high+' · '+item.metrics.hotspots+' hotspots · commit '+esc(item.commit||"unavailable")+'</small></div><span class="pill">'+esc(item.id)+'</span></div>').join("");el("timelineNotes").textContent="Trend: "+comparison.trend+". New top hotspots: "+(comparison.newHotspots.join(", ")||"none")+". No-longer-top hotspots: "+(comparison.resolvedHotspots.join(", ")||"none")+". Snapshots are local only."; }catch(error){el("timelineHistory").innerHTML='<div class="error">'+esc(error.message)+'</div>';}}
231
237
  async function loadChangeRisk(){el("changeRiskList").innerHTML='<div class="loading">Analyzing working-tree changes…</div>';try{const data=await api("/api/change-intelligence");const s=data.summary;el("changeRiskSummary").innerHTML=metric("Changed files",s.changedFiles,"Compared with "+data.base)+metric("Overall risk",data.overallRisk.score+"/100",data.overallRisk.level.toUpperCase())+metric("Affected modules",s.affectedModules,"Static downstream impact")+metric("Test candidates",s.testCandidates,"Existing and suggested paths");el("changeRiskList").innerHTML=data.changes.length?data.changes.map(item=>'<div class="activity-item"><div class="activity-main"><b>'+esc(item.file)+'</b><small>'+esc(item.reasons.join(" "))+'</small><small>'+item.downstreamCount+' downstream modules · '+esc(item.recommendedTests.map(test=>test.file+(test.exists?"":" (suggested)")).join(", ")||"No test candidates")+'</small></div><span class="pill">'+esc(item.level.toUpperCase())+' · '+item.score+'</span></div>').join(""):'<div class="empty">No working-tree changes detected against the selected base.</div>';el("changeRiskNotes").textContent=(data.overallRisk.reasons||[]).concat(data.warnings||[]).join(" · ");}catch(error){el("changeRiskList").innerHTML='<div class="error">'+esc(error.message)+'</div>';}}
232
238
  async function loadEngineeringRisk(){el("riskHotspots").innerHTML='<div class="loading">Calculating structural hotspots…</div>';try{const data=await api("/api/engineering-risk");const s=data.summary;el("riskSummary").innerHTML=metric("Hotspots",s.hotspots,"Prioritized modules")+metric("High risk",s.high,"Review these first")+metric("Elevated",s.elevated,"Worth investigating")+metric("No matching tests",s.withoutMatchingTests,"Filename heuristic");el("riskHotspots").innerHTML=data.hotspots.length?data.hotspots.map(item=>'<div class="activity-item"><div class="activity-main"><b>'+esc(item.file)+'</b><small>'+esc(item.signals.join(" "))+'</small><small>Fan-in '+item.fanIn+' · fan-out '+item.fanOut+' · '+item.lines+' lines · test '+(item.hasMatchingTest?"detected":"not detected")+'</small></div><span class="pill">'+esc(item.level.toUpperCase())+' · '+item.score+'</span></div>').join(""):'<div class="empty">No structural hotspots detected by current heuristics.</div>';el("riskMethodology").textContent=data.methodology;el("riskWarnings").textContent=(data.warnings||[]).join(" · ");}catch(error){el("riskHotspots").innerHTML='<div class="error">'+esc(error.message)+'</div>';}}
233
239
  let dependencyGraphData=null;
@@ -283,12 +289,12 @@ async function loadRegression(){
283
289
  const top='<div class="grid">'+cards.map(([k,v])=>'<div class="card"><div class="metric-label">'+esc(k)+'</div><div class="metric">'+esc(v)+'</div></div>').join("")+'</div>';
284
290
  const checks=data.checks.slice(0,8).map(item=>'<div class="rowitem"><div><b>'+esc(item.script)+'</b><div class="run-meta">'+(item.previous?esc(item.previous.status)+' → '+esc(item.latest.status):'First recorded baseline')+(item.newFailureSignature?' · New failure signature':'')+'</div></div><span class="pill">'+esc(item.state.replaceAll("_"," "))+'</span></div>').join("");
285
291
  const changed=data.changedFiles.slice(0,6).map(item=>'<div class="rowitem"><b>'+esc(item.file)+'</b><span class="pill">'+esc(item.status)+'</span></div>').join("");
286
- el("regressionPanel").innerHTML=top+(checks||'<div class="empty">No verification history yet. Run forge verify SCRIPT to establish a baseline.</div>')+'<div class="run-meta"><b>Git changes</b></div>'+(changed||'<div class="empty">No changed files detected, or Git context is unavailable.</div>')+'<div class="run-meta">Signals compare saved command outcomes; they do not prove causation. Suggested tests are not executed.</div>';
292
+ el("regressionPanel").innerHTML=top+(checks||'<div class="empty">No verification history yet. Run anvil verify SCRIPT to establish a baseline.</div>')+'<div class="run-meta"><b>Git changes</b></div>'+(changed||'<div class="empty">No changed files detected, or Git context is unavailable.</div>')+'<div class="run-meta">Signals compare saved command outcomes; they do not prove causation. Suggested tests are not executed.</div>';
287
293
  }catch(error){el("regressionPanel").innerHTML='<div class="error">'+esc(error.message)+'</div>';}
288
294
  }
289
295
  async function loadReviews(){
290
296
  try{const data=await api("/api/reviews");
291
- el("reviewPanel").innerHTML=data.length?data.slice(0,8).map(item=>'<div class="rowitem"><div><b>'+esc(item.id)+'</b><div class="run-meta">'+esc(new Date(item.createdAt).toLocaleString())+' · '+esc(item.filesReviewed)+' files · '+esc(item.staticFindings)+' static findings · '+esc(item.aiSuggestions)+' AI suggestions</div>'+(item.topFindings.length?'<div class="run-meta">'+item.topFindings.map(esc).join(" · ")+'</div>':"")+'</div><span class="pill">'+esc(item.scope)+'</span></div>').join(""):'<div class="empty">No saved reviews yet. Run forge review . to create one.</div>';
297
+ el("reviewPanel").innerHTML=data.length?data.slice(0,8).map(item=>'<div class="rowitem"><div><b>'+esc(item.id)+'</b><div class="run-meta">'+esc(new Date(item.createdAt).toLocaleString())+' · '+esc(item.filesReviewed)+' files · '+esc(item.staticFindings)+' static findings · '+esc(item.aiSuggestions)+' AI suggestions</div>'+(item.topFindings.length?'<div class="run-meta">'+item.topFindings.map(esc).join(" · ")+'</div>':"")+'</div><span class="pill">'+esc(item.scope)+'</span></div>').join(""):'<div class="empty">No saved reviews yet. Run anvil review . to create one.</div>';
292
298
  }catch(error){el("reviewPanel").innerHTML='<div class="error">'+esc(error.message)+'</div>';}
293
299
  }
294
300
  async function loadQuality(){
@@ -317,7 +323,7 @@ async function previewWorkspace(filePath){
317
323
  async function runCodeSearch(){const query=el("codeSearchQuery").value.trim();if(query.length<2){el("codeSearchMeta").textContent="Enter at least two characters.";return;}el("codeSearchResults").innerHTML='<div class="loading">Searching source files…</div>';try{const data=await api("/api/search?q="+encodeURIComponent(query));el("codeSearchMeta").textContent=data.total+" matches · "+data.filesScanned+" files scanned"+(data.truncated?" · showing first "+data.results.length:"");el("codeSearchResults").innerHTML=data.results.length?data.results.map(item=>'<button class="btn" style="text-align:left;white-space:normal" data-search-path="'+esc(item.path)+'"><b>'+esc(item.symbol||item.path)+'</b> <span class="pill">'+esc(item.kind)+'</span><div class="run-meta">'+esc(item.path)+':'+item.line+':'+item.column+'</div><code>'+esc(item.snippet)+'</code></button>').join(""):'<div class="empty">No matches found.</div>';el("codeSearchResults").querySelectorAll("[data-search-path]").forEach(button=>button.addEventListener("click",()=>previewWorkspace(button.dataset.searchPath)));}catch(error){el("codeSearchResults").innerHTML='<div class="error">'+esc(error.message)+'</div>';}}
318
324
  async function loadVerifications(){
319
325
  try{const data=await api("/api/verifications");
320
- if(!data.length){el("verificationPanel").innerHTML='<div class="empty">No verification runs yet. Use forge verify SCRIPT.</div>';return;}
326
+ if(!data.length){el("verificationPanel").innerHTML='<div class="empty">No verification runs yet. Use anvil verify SCRIPT.</div>';return;}
321
327
  el("verificationPanel").innerHTML=data.slice(0,8).map(item=>'<div class="rowitem"><div><b>'+esc(item.script)+'</b><div class="run-meta">'+esc(new Date(item.startedAt).toLocaleString())+' · '+esc(item.durationMs)+' ms</div></div><span class="pill">'+esc(item.status)+'</span></div>').join("");
322
328
  }catch(error){el("verificationPanel").innerHTML='<div class="error">'+esc(error.message)+'</div>';}
323
329
  }
@@ -347,14 +353,11 @@ function workflowSlug(value){return value.toLowerCase().trim().replace(/[^a-z0-9
347
353
  function workflowEdges(){
348
354
  return workflowTasks.flatMap(task=>(task.dependsOn||[]).map(dep=>({from:dep,to:task.id})));
349
355
  }
350
- let workflowTasks=[];
351
356
  let workflowPositions={};
352
357
  let workflowScale=1;
353
358
  let workflowPan={x:0,y:0};
354
359
  let workflowDrag=null;
355
360
  let workflowSelected="";
356
- function workflowSlug(value){return value.toLowerCase().trim().replace(/[^a-z0-9]+/g,"-").replace(/^-|-$/g,"").slice(0,48)||"custom-workflow";}
357
- function workflowEdges(){return workflowTasks.flatMap(task=>(task.dependsOn||[]).map(dep=>({from:dep,to:task.id})));}
358
361
  function ensureWorkflowPositions(){
359
362
  const levels=new Map();workflowTasks.forEach(task=>{const level=(task.dependsOn||[]).reduce((max,dep)=>Math.max(max,(levels.get(dep)||0)+1),0);levels.set(task.id,level);});
360
363
  const rows=new Map();workflowTasks.forEach(task=>{const level=levels.get(task.id)||0,row=rows.get(level)||0;rows.set(level,row+1);if(!workflowPositions[task.id])workflowPositions[task.id]={x:70+level*270,y:65+row*145};});
@@ -372,12 +375,12 @@ function renderWorkflow(){
372
375
  svg.addEventListener("pointerdown",event=>{if(event.target.closest("[data-graph-node]"))return;workflowDrag={kind:"pan",x:event.clientX,y:event.clientY,px:workflowPan.x,py:workflowPan.y};svg.setPointerCapture(event.pointerId);});
373
376
  svg.addEventListener("pointermove",event=>{if(!workflowDrag)return;if(workflowDrag.kind==="node"){const q=point(event);workflowPositions[workflowDrag.id]={x:Math.max(0,Math.min(3900,q.x+workflowDrag.dx)),y:Math.max(0,Math.min(2900,q.y+workflowDrag.dy))};const n=world.querySelector('[data-graph-node="'+workflowDrag.id+'"]');if(n)n.setAttribute("transform","translate("+workflowPositions[workflowDrag.id].x+" "+workflowPositions[workflowDrag.id].y+")");}else{workflowPan={x:workflowDrag.px+event.clientX-workflowDrag.x,y:workflowDrag.py+event.clientY-workflowDrag.y};world.setAttribute("transform","translate("+workflowPan.x+" "+workflowPan.y+") scale("+workflowScale+")");}});
374
377
  svg.addEventListener("pointerup",()=>{if(workflowDrag&&workflowDrag.kind==="node")renderWorkflow();workflowDrag=null;});svg.addEventListener("pointercancel",()=>{workflowDrag=null;});
375
- const analysis=analyzeWorkflow();el("workflowAnalysis").innerHTML=analysis.cyclic?'<span class="error">Dependency cycle detected. Fix the highlighted workflow before saving.</span>':'<b>Critical path estimate: '+analysis.critical+' min</b> · '+workflowTasks.length+' tasks · '+workflowEdges().length+' links · planning estimate, not measured execution time.';
376
- el("wfSuggest").addEventListener("click",suggestWorkflowDependencies);
377
- function zoom(next){workflowScale=Math.max(.4,Math.min(2,next));world.setAttribute("transform","translate("+workflowPan.x+" "+workflowPan.y+") scale("+workflowScale+")");el("wfZoomLabel").textContent=Math.round(workflowScale*100)+"%";}
378
- svg.addEventListener("wheel",event=>{event.preventDefault();zoom(workflowScale*(event.deltaY<0?1.1:.9));},{passive:false});el("wfZoomIn").addEventListener("click",()=>zoom(workflowScale+.1));el("wfZoomOut").addEventListener("click",()=>zoom(workflowScale-.1));el("wfFit").addEventListener("click",()=>{workflowPan={x:0,y:0};zoom(1);});
379
- el("applyWorkflowLinks").addEventListener("click",()=>{const task=workflowTasks.find(item=>item.id===workflowSelected);if(!task){el("workflowMessage").textContent="Select a node first.";return;}task.dependsOn=[...dependency.selectedOptions].map(option=>option.value).filter(id=>id&&id!==task.id);renderWorkflow();el("workflowMessage").textContent="Links updated. Save to persist.";});
380
- el("removeSelectedWorkflowNode").addEventListener("click",()=>{if(!workflowSelected)return;const id=workflowSelected;workflowTasks=workflowTasks.filter(task=>task.id!==id).map(task=>({...task,dependsOn:(task.dependsOn||[]).filter(dep=>dep!==id)}));delete workflowPositions[id];workflowSelected="";renderWorkflow();});
378
+ const analysis=analyzeWorkflow();const wfAnalysisNode=el("workflowAnalysis");if(wfAnalysisNode)wfAnalysisNode.innerHTML=analysis.cyclic?'<span class="error">Dependency cycle detected. Fix the highlighted workflow before saving.</span>':'<b>Critical path estimate: '+analysis.critical+' min</b> · '+workflowTasks.length+' tasks · '+workflowEdges().length+' links · planning estimate, not measured execution time.';
379
+ on("wfSuggest","click",suggestWorkflowDependencies);
380
+ function zoom(next){workflowScale=Math.max(.4,Math.min(2,next));world.setAttribute("transform","translate("+workflowPan.x+" "+workflowPan.y+") scale("+workflowScale+")");const z=el("wfZoomLabel");if(z)z.textContent=Math.round(workflowScale*100)+"%";}
381
+ svg.addEventListener("wheel",event=>{event.preventDefault();zoom(workflowScale*(event.deltaY<0?1.1:.9));},{passive:false});on("wfZoomIn","click",()=>zoom(workflowScale+.1));on("wfZoomOut","click",()=>zoom(workflowScale-.1));on("wfFit","click",()=>{workflowPan={x:0,y:0};zoom(1);});
382
+ on("applyWorkflowLinks","click",()=>{const task=workflowTasks.find(item=>item.id===workflowSelected);if(!task){el("workflowMessage").textContent="Select a node first.";return;}task.dependsOn=[...dependency.selectedOptions].map(option=>option.value).filter(id=>id&&id!==task.id);renderWorkflow();el("workflowMessage").textContent="Links updated. Save to persist.";});
383
+ on("removeSelectedWorkflowNode","click",()=>{if(!workflowSelected)return;const id=workflowSelected;workflowTasks=workflowTasks.filter(task=>task.id!==id).map(task=>({...task,dependsOn:(task.dependsOn||[]).filter(dep=>dep!==id)}));delete workflowPositions[id];workflowSelected="";renderWorkflow();});
381
384
  }
382
385
  async function refreshWorkflows(){
383
386
  try{const workflows=await api("/api/workflows");el("savedWorkflows").innerHTML=workflows.length?workflows.map(workflow=>'<div class="run"><div class="run-title"><b>'+esc(workflow.name)+'</b><span class="pill">'+esc(workflow.tasks.length)+' TASKS</span></div><div class="run-meta">'+esc(workflow.description||"Reusable agent workflow")+' · '+esc(workflow.id)+'</div><div class="actions"><button class="btn" data-load-workflow="'+esc(workflow.id)+'">Load into editor</button><button class="btn" data-create-workflow-run="'+esc(workflow.id)+'">Create draft run</button></div></div>').join(""):'<div class="empty">No saved workflows yet. Compose one and save it here.</div>';
@@ -389,7 +392,7 @@ async function saveCurrentWorkflow(){
389
392
  const name=el("workflowName").value.trim();if(!name||!workflowTasks.length){el("workflowMessage").textContent="Enter a workflow name and add at least one task.";return null;}
390
393
  const analysis=analyzeWorkflow();if(analysis.cyclic){el("workflowMessage").textContent="Cannot save: dependency cycle detected. Remove a circular link first.";return null;}
391
394
  const definition={id:workflowSlug(name),name,description:el("workflowDescription").value.trim(),tasks:workflowTasks,positions:workflowPositions};
392
- try{const saved=await fetch("/api/workflows",{method:"POST",headers:{"content-type":"application/json"},body:JSON.stringify(definition)});const data=await saved.json();if(!saved.ok)throw new Error(data.error||"Could not save workflow.");el("workflowMessage").textContent="Saved locally as .forge/workflows/"+data.id+".json";await refreshWorkflows();return data.id;}
395
+ try{const saved=await fetch("/api/workflows",{method:"POST",headers:{"content-type":"application/json"},body:JSON.stringify(definition)});const data=await saved.json();if(!saved.ok)throw new Error(data.error||"Could not save workflow.");el("workflowMessage").textContent="Saved locally as .anvil/workflows/ (or .forge/) "+data.id+".json";await refreshWorkflows();return data.id;}
393
396
  catch(error){el("workflowMessage").textContent=error.message;return null;}
394
397
  }
395
398
  async function createWorkflowDraft(id){
@@ -422,21 +425,22 @@ function loadWorkflowTemplate(kind){
422
425
  el("workflowDescription").value=kind==="security"?"Parallel security and data review followed by gated verification.":"Architecture-led feature planning with parallel specialist checks and a final readiness review.";
423
426
  renderWorkflow();el("workflowMessage").textContent="Template loaded. Customize the tasks and save it as your own workflow.";
424
427
  }
425
- el("templateFeature").addEventListener("click",()=>loadWorkflowTemplate("feature"));
426
- el("templateSecurity").addEventListener("click",()=>loadWorkflowTemplate("security"));
427
- \nel("addWorkflowTask").addEventListener("click",()=>{
428
+ on("templateFeature","click",()=>loadWorkflowTemplate("feature"));
429
+ on("templateSecurity","click",()=>loadWorkflowTemplate("security"));
430
+ on("addWorkflowTask","click",()=>{
428
431
  const title=el("workflowTaskTitle").value.trim(),prompt=el("workflowTaskPrompt").value.trim();if(!title||!prompt){el("workflowMessage").textContent="A task title and instructions are required.";return;}if(workflowTasks.length>=30){el("workflowMessage").textContent="A workflow can contain at most 30 tasks.";return;}
429
432
  const id=workflowSlug(title).slice(0,30)+"-"+(workflowTasks.length+1);const dependencies=[...el("workflowDependency").selectedOptions].map(option=>option.value).filter(Boolean);
430
433
  workflowTasks.push({id,agent:el("workflowAgent").value,title,prompt,dependsOn:dependencies,requiresApproval:el("workflowRequiresApproval").checked});workflowPositions[id]={x:70+(workflowTasks.length-1)*40,y:65+(workflowTasks.length-1)*115};
431
434
  el("workflowTaskTitle").value="";el("workflowTaskPrompt").value="";el("workflowRequiresApproval").checked=false;renderWorkflow();el("workflowMessage").textContent="Task added to the pipeline.";
432
435
  });
433
- el("clearWorkflowTasks").addEventListener("click",()=>{workflowTasks=[];renderWorkflow();});
434
- el("applyWorkflowLinks").addEventListener("click",()=>{const id=el("applyWorkflowLinks").dataset.taskId;const task=workflowTasks.find(item=>item.id===id);if(!task)return;task.dependsOn=[...el("workflowDependency").selectedOptions].map(option=>option.value).filter(value=>value&&value!==id);renderWorkflow();el("workflowMessage").textContent="Dependency links updated. Save to persist them.";});
435
- el("saveWorkflow").addEventListener("click",saveCurrentWorkflow);
436
- el("createWorkflowRun").addEventListener("click",async()=>{const id=await saveCurrentWorkflow();if(id)await createWorkflowDraft(id);});
437
- el("refreshWorkflows").addEventListener("click",refreshWorkflows);renderWorkflow();refreshWorkflows();
438
- el("codeSearchButton").addEventListener("click",runCodeSearch);el("codeSearchQuery").addEventListener("keydown",event=>{if(event.key==="Enter")runCodeSearch();});\nel("workspaceUp").addEventListener("click",()=>loadWorkspace(workspaceDirectory.includes("/")?workspaceDirectory.slice(0,workspaceDirectory.lastIndexOf("/")):""));
439
- el("refreshDependencyIntel").addEventListener("click",loadDependencyIntelligence);loadDependencyIntelligence();el("refreshTimeline").addEventListener("click",loadIntelligenceTimeline);loadIntelligenceTimeline();el("refreshChangeRisk").addEventListener("click",loadChangeRisk);loadChangeRisk();el("refreshRisk").addEventListener("click",loadEngineeringRisk);el("refreshDependencyGraph").addEventListener("click",loadDependencyGraph);loadEngineeringRisk();el("dependencyFilter").addEventListener("input",()=>{if(dependencyGraphData)renderDependencyGraph(dependencyGraphData);});el("dependencyMode").addEventListener("change",()=>{if(dependencyGraphData)renderDependencyGraph(dependencyGraphData);});el("refreshMap").addEventListener("click",loadMap);loadDependencyGraph();loadWorkspace();el("refreshHealth").addEventListener("click",loadHealth);el("refreshRegression").addEventListener("click",loadRegression);el("refreshReviews").addEventListener("click",loadReviews);el("refreshActivity").addEventListener("click",()=>{loadActivity();loadRuns();});loadOverview();loadAgents();loadRuns();loadActivity();loadObservability();loadEvaluationHistory();loadBenchmarkHistory();loadBenchmarkRegression();loadAgentPerformance();loadSkillExperiments();loadExperimentHistory();loadAgentOptimization();loadAgentIntelligence();el("refreshExperimentHistory").addEventListener("click",loadExperimentHistory);el("experimentSuiteFilter").addEventListener("change",loadExperimentHistory);el("experimentCandidateFilter").addEventListener("change",loadExperimentHistory);el("exportExperimentHistory").addEventListener("click",()=>{if(experimentHistoryData){const blob=new Blob([JSON.stringify(experimentHistoryData,null,2)],{type:"application/json"});const url=URL.createObjectURL(blob);const link=document.createElement("a");link.href=url;link.download="forge-experiment-history.json";link.click();URL.revokeObjectURL(url);}});el("refreshRegression").addEventListener("click",loadBenchmarkRegression);loadMap();loadQuality();loadVerifications();loadHealth();loadRegression();loadReviews();loadConfigFiles();loadConfigVersions();loadConfigReleases();loadProviderReadiness();setInterval(()=>{loadActivity();loadRuns();loadVerifications();if(selectedRunId)inspectRun(selectedRunId);},5000);connectSSE();
436
+ on("clearWorkflowTasks","click",()=>{workflowTasks=[];renderWorkflow();});
437
+ on("applyWorkflowLinks","click",()=>{const id=el("applyWorkflowLinks")?.dataset?.taskId;const task=workflowTasks.find(item=>item.id===id);if(!task)return;task.dependsOn=[...el("workflowDependency").selectedOptions].map(option=>option.value).filter(value=>value&&value!==id);renderWorkflow();el("workflowMessage").textContent="Dependency links updated. Save to persist them.";});
438
+ on("saveWorkflow","click",saveCurrentWorkflow);
439
+ on("createWorkflowRun","click",async()=>{const id=await saveCurrentWorkflow();if(id)await createWorkflowDraft(id);});
440
+ on("refreshWorkflows","click",refreshWorkflows);renderWorkflow();refreshWorkflows();
441
+ on("codeSearchButton","click",runCodeSearch);on("codeSearchQuery","keydown",event=>{if(event.key==="Enter")runCodeSearch();});
442
+ on("workspaceUp","click",()=>loadWorkspace(workspaceDirectory.includes("/")?workspaceDirectory.slice(0,workspaceDirectory.lastIndexOf("/")):""));
443
+ on("refreshDependencyIntel","click",loadDependencyIntelligence);loadDependencyIntelligence();on("refreshTimeline","click",loadIntelligenceTimeline);loadIntelligenceTimeline();on("refreshChangeRisk","click",loadChangeRisk);loadChangeRisk();on("refreshRisk","click",loadEngineeringRisk);on("refreshDependencyGraph","click",loadDependencyGraph);loadEngineeringRisk();on("dependencyFilter","input",()=>{if(dependencyGraphData)renderDependencyGraph(dependencyGraphData);});on("dependencyMode","change",()=>{if(dependencyGraphData)renderDependencyGraph(dependencyGraphData);});on("refreshMap","click",loadMap);loadDependencyGraph();loadWorkspace();on("refreshHealth","click",loadHealth);on("refreshRegression","click",loadRegression);on("refreshReviews","click",loadReviews);on("refreshActivity","click",()=>{loadActivity();loadRuns();});loadOverview();loadAgents();loadRuns();loadActivity();loadObservability();loadEvaluationHistory();loadBenchmarkHistory();loadBenchmarkRegression();loadAgentPerformance();loadSkillExperiments();loadExperimentHistory();loadAgentOptimization();loadAgentIntelligence();on("refreshExperimentHistory","click",loadExperimentHistory);on("experimentSuiteFilter","change",loadExperimentHistory);on("experimentCandidateFilter","change",loadExperimentHistory);on("exportExperimentHistory","click",()=>{if(experimentHistoryData){const blob=new Blob([JSON.stringify(experimentHistoryData,null,2)],{type:"application/json"});const url=URL.createObjectURL(blob);const link=document.createElement("a");link.href=url;link.download="anvil-experiment-history.json";link.click();URL.revokeObjectURL(url);}});on("refreshRegression","click",loadBenchmarkRegression);loadMap();loadQuality();loadVerifications();loadHealth();loadRegression();loadReviews();loadConfigFiles();loadConfigVersions();loadConfigReleases();loadProviderReadiness();setInterval(()=>{loadActivity();loadRuns();loadVerifications();if(selectedRunId)inspectRun(selectedRunId);},5000);connectSSE();
440
444
  </script></body></html>`;
441
445
  export async function startDashboard(options) {
442
446
  const host = "127.0.0.1";
@@ -658,13 +662,13 @@ export async function startDashboard(options) {
658
662
  return;
659
663
  }
660
664
  if (requestUrl.pathname === "/api/providers") {
661
- const selected = (process.env.FORGE_AI_PROVIDER ?? "openai").toLowerCase();
665
+ const selected = (process.env.ANVIL_AI_PROVIDER ?? process.env.FORGE_AI_PROVIDER ?? "openai").toLowerCase();
662
666
  const rows = [
663
- { id: "openai", configured: Boolean(process.env.FORGE_API_KEY || process.env.OPENAI_API_KEY), model: process.env.FORGE_AI_MODEL || process.env.FORGE_MODEL || "gpt-4.1-mini", detail: "API credential presence only; connectivity is not tested." },
664
- { id: "anthropic", configured: Boolean(process.env.ANTHROPIC_API_KEY), model: process.env.FORGE_AI_MODEL || "claude-sonnet-4-5", detail: "API credential presence only; connectivity is not tested." },
665
- { id: "google", configured: Boolean(process.env.GOOGLE_API_KEY), model: process.env.FORGE_AI_MODEL || "gemini-2.5-flash", detail: "API credential presence only; connectivity is not tested." },
666
- { id: "ollama", configured: true, model: process.env.FORGE_AI_MODEL || "llama3.2", detail: "Endpoint origin: " + safeProviderOrigin(process.env.FORGE_BASE_URL || process.env.OLLAMA_HOST || "http://localhost:11434") + "; connectivity is not tested." },
667
- { id: "claude-code", configured: false, model: process.env.FORGE_AI_MODEL || "sonnet", detail: "Local CLI installation and account authentication are not checked by the dashboard." }
667
+ { id: "openai", configured: Boolean(process.env.ANVIL_API_KEY || process.env.FORGE_API_KEY || process.env.OPENAI_API_KEY), model: process.env.ANVIL_AI_MODEL || process.env.FORGE_AI_MODEL || process.env.ANVIL_MODEL || process.env.FORGE_MODEL || "gpt-4.1-mini", detail: "API credential presence only; connectivity is not tested." },
668
+ { id: "anthropic", configured: Boolean(process.env.ANTHROPIC_API_KEY), model: process.env.ANVIL_AI_MODEL || process.env.FORGE_AI_MODEL || "claude-sonnet-4-5", detail: "API credential presence only; connectivity is not tested." },
669
+ { id: "google", configured: Boolean(process.env.GOOGLE_API_KEY), model: process.env.ANVIL_AI_MODEL || process.env.FORGE_AI_MODEL || "gemini-2.5-flash", detail: "API credential presence only; connectivity is not tested." },
670
+ { id: "ollama", configured: true, model: process.env.ANVIL_AI_MODEL || process.env.FORGE_AI_MODEL || "llama3.2", detail: "Endpoint origin: " + safeProviderOrigin(process.env.ANVIL_BASE_URL || process.env.FORGE_BASE_URL || process.env.OLLAMA_HOST || "http://localhost:11434") + "; connectivity is not tested." },
671
+ { id: "claude-code", configured: false, model: process.env.ANVIL_AI_MODEL || process.env.FORGE_AI_MODEL || "sonnet", detail: "Local CLI installation and account authentication are not checked by the dashboard." }
668
672
  ].map(item => ({ ...item, selected: selected === item.id, status: item.id === "claude-code" ? "not-checked" : item.configured ? "configured" : "needs-credentials" }));
669
673
  response.writeHead(200, { "content-type": "application/json; charset=utf-8" }).end(JSON.stringify(rows));
670
674
  return;
@@ -906,7 +910,7 @@ export async function startDashboard(options) {
906
910
  catch {
907
911
  return {};
908
912
  } };
909
- response.writeHead(200, { "content-type": "application/json; charset=utf-8" }).end(JSON.stringify({ generatedAt: new Date().toISOString(), runsScanned: runs.slice(0, 30).length, totals: { requests: [...agents.values()].reduce((n, a) => n + a.requests, 0), inputTokensEstimate: [...agents.values()].reduce((n, a) => n + a.inputTokensEstimate, 0), outputTokensEstimate: [...agents.values()].reduce((n, a) => n + a.outputTokensEstimate, 0), durationMs: [...agents.values()].reduce((n, a) => n + a.durationMs, 0), failures: [...agents.values()].reduce((n, a) => n + a.failures, 0), fallbacks: [...agents.values()].reduce((n, a) => n + a.fallbacks, 0) }, agents: [...agents.values()].sort((a, b) => b.durationMs - a.durationMs), providers: [...providers.values()].sort((a, b) => b.requests - a.requests), budgets: { perAgentTokens: parseBudgetMap(process.env.FORGE_MAX_ESTIMATED_TOKENS_PER_AGENT), perAgentDurationMs: parseBudgetMap(process.env.FORGE_MAX_TASK_DURATION_MS_BY_AGENT), runTokens: Number(process.env.FORGE_MAX_ESTIMATED_TOKENS_PER_RUN) || null, taskDurationMs: Number(process.env.FORGE_MAX_TASK_DURATION_MS) || null }, recent }));
913
+ response.writeHead(200, { "content-type": "application/json; charset=utf-8" }).end(JSON.stringify({ generatedAt: new Date().toISOString(), runsScanned: runs.slice(0, 30).length, totals: { requests: [...agents.values()].reduce((n, a) => n + a.requests, 0), inputTokensEstimate: [...agents.values()].reduce((n, a) => n + a.inputTokensEstimate, 0), outputTokensEstimate: [...agents.values()].reduce((n, a) => n + a.outputTokensEstimate, 0), durationMs: [...agents.values()].reduce((n, a) => n + a.durationMs, 0), failures: [...agents.values()].reduce((n, a) => n + a.failures, 0), fallbacks: [...agents.values()].reduce((n, a) => n + a.fallbacks, 0) }, agents: [...agents.values()].sort((a, b) => b.durationMs - a.durationMs), providers: [...providers.values()].sort((a, b) => b.requests - a.requests), budgets: { perAgentTokens: parseBudgetMap(process.env.ANVIL_MAX_ESTIMATED_TOKENS_PER_AGENT || process.env.FORGE_MAX_ESTIMATED_TOKENS_PER_AGENT), perAgentDurationMs: parseBudgetMap(process.env.ANVIL_MAX_TASK_DURATION_MS_BY_AGENT || process.env.FORGE_MAX_TASK_DURATION_MS_BY_AGENT), runTokens: Number(process.env.ANVIL_MAX_ESTIMATED_TOKENS_PER_RUN || process.env.FORGE_MAX_ESTIMATED_TOKENS_PER_RUN) || null, taskDurationMs: Number(process.env.ANVIL_MAX_TASK_DURATION_MS || process.env.FORGE_MAX_TASK_DURATION_MS) || null }, recent }));
910
914
  return;
911
915
  }
912
916
  const traceRoute = requestUrl.pathname.match(/^\/api\/runs\/([^/]+)\/trace$/);