@tangle-network/agent-runtime 0.89.0 → 0.90.1

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 (69) hide show
  1. package/README.md +51 -18
  2. package/dist/agent.d.ts +53 -2
  3. package/dist/agent.js +168 -6
  4. package/dist/agent.js.map +1 -1
  5. package/dist/analyst-loop.d.ts +1 -1
  6. package/dist/chunk-5QOB7H74.js +387 -0
  7. package/dist/chunk-5QOB7H74.js.map +1 -0
  8. package/dist/{chunk-4WXGK6GV.js → chunk-7ON74BQO.js} +2 -2
  9. package/dist/{chunk-7LO5GMAO.js → chunk-CMYMTRGA.js} +14 -5
  10. package/dist/chunk-CMYMTRGA.js.map +1 -0
  11. package/dist/{chunk-UD4BHQMI.js → chunk-IVGYLCFH.js} +2 -2
  12. package/dist/{chunk-UD4BHQMI.js.map → chunk-IVGYLCFH.js.map} +1 -1
  13. package/dist/{chunk-VNOOH22O.js → chunk-OOL3675H.js} +85 -4
  14. package/dist/chunk-OOL3675H.js.map +1 -0
  15. package/dist/{chunk-5AVV7KAH.js → chunk-QK4DV5PR.js} +2 -2
  16. package/dist/chunk-R2VAJGR3.js +214 -0
  17. package/dist/chunk-R2VAJGR3.js.map +1 -0
  18. package/dist/{chunk-N7EJV7N3.js → chunk-RYBVU4M3.js} +3592 -3562
  19. package/dist/chunk-RYBVU4M3.js.map +1 -0
  20. package/dist/{chunk-BQPFZE2C.js → chunk-WRUSWK4F.js} +5 -5
  21. package/dist/{chunk-PIPPLSOF.js → chunk-ZV4LXYCJ.js} +451 -24
  22. package/dist/chunk-ZV4LXYCJ.js.map +1 -0
  23. package/dist/{worktree-harness-CX_McRLp.d.ts → completion-gate-DkAnUmpb.d.ts} +45 -45
  24. package/dist/{coordination-CuDLO8wj.d.ts → coordination-rRj5hjJK.d.ts} +415 -227
  25. package/dist/environment-provider.d.ts +3 -3
  26. package/dist/index.d.ts +255 -16
  27. package/dist/index.js +463 -72
  28. package/dist/index.js.map +1 -1
  29. package/dist/intelligence.d.ts +228 -84
  30. package/dist/intelligence.js +259 -166
  31. package/dist/intelligence.js.map +1 -1
  32. package/dist/knowledge.d.ts +120 -0
  33. package/dist/knowledge.js +28 -0
  34. package/dist/knowledge.js.map +1 -0
  35. package/dist/lifecycle.d.ts +2 -2
  36. package/dist/lifecycle.js +2 -2
  37. package/dist/{local-harness-sI0S_XNA.d.ts → local-harness-dcD5WTTr.d.ts} +3 -0
  38. package/dist/{loop-runner-bin-hQDUL8Ld.d.ts → loop-runner-bin-DTbZVGfM.d.ts} +2 -2
  39. package/dist/loop-runner-bin.d.ts +6 -6
  40. package/dist/loop-runner-bin.js +7 -7
  41. package/dist/loops.d.ts +34 -369
  42. package/dist/loops.js +31 -5
  43. package/dist/mcp/bin.js +5 -5
  44. package/dist/mcp/index.d.ts +10 -11
  45. package/dist/mcp/index.js +10 -10
  46. package/dist/{mcp-serve-verifier-FL7-ZEb_.d.ts → mcp-serve-verifier-XsX8rkB9.d.ts} +12 -1
  47. package/dist/{openai-tools-BnrOmGjN.d.ts → openai-tools-C4ZfUD4L.d.ts} +1 -1
  48. package/dist/profiles.d.ts +196 -5
  49. package/dist/profiles.js +8 -0
  50. package/dist/profiles.js.map +1 -1
  51. package/dist/{router-client-r8y_VFVM.d.ts → router-client-DJImUDlm.d.ts} +1 -1
  52. package/dist/structural-rollout-MwlpgQ-6.d.ts +446 -0
  53. package/dist/supervise-DPmYPk0j.d.ts +131 -0
  54. package/dist/{types-ESeMOj94.d.ts → types-SyuwunY_.d.ts} +1 -1
  55. package/dist/{types-Driepl87.d.ts → types-eMNgWgFi.d.ts} +2 -2
  56. package/dist/{worktree-fanout-D6xR2CIA.d.ts → worktree-fanout-BDFQIO-Y.d.ts} +234 -234
  57. package/package.json +12 -4
  58. package/skills/agent-runtime-adoption/SKILL.md +74 -24
  59. package/skills/build-with-agent-runtime/SKILL.md +23 -20
  60. package/skills/loop-writer/SKILL.md +6 -5
  61. package/skills/supervise/SKILL.md +2 -2
  62. package/dist/chunk-7LO5GMAO.js.map +0 -1
  63. package/dist/chunk-N7EJV7N3.js.map +0 -1
  64. package/dist/chunk-PIPPLSOF.js.map +0 -1
  65. package/dist/chunk-VNOOH22O.js.map +0 -1
  66. package/dist/otel-export-BKmNwiCb.d.ts +0 -180
  67. /package/dist/{chunk-4WXGK6GV.js.map → chunk-7ON74BQO.js.map} +0 -0
  68. /package/dist/{chunk-5AVV7KAH.js.map → chunk-QK4DV5PR.js.map} +0 -0
  69. /package/dist/{chunk-BQPFZE2C.js.map → chunk-WRUSWK4F.js.map} +0 -0
@@ -1 +1 @@
1
- {"version":3,"sources":["../src/otel-export.ts"],"sourcesContent":["/**\n * OTEL span exporter — streams LoopTraceEvents to an OTLP/HTTP collector.\n *\n * Reads OTEL_EXPORTER_OTLP_ENDPOINT + OTEL_EXPORTER_OTLP_HEADERS from env\n * when no explicit config is given. Keeps the runtime dep-free from\n * @opentelemetry/sdk-trace-base — minimal OTLP/JSON serializer.\n *\n * The exporter accepts both raw OtelSpan objects and LoopTraceEvents\n * (which get converted to OTLP spans automatically).\n */\n\nexport interface OtelExportConfig {\n /** OTLP endpoint. Reads OTEL_EXPORTER_OTLP_ENDPOINT env by default. */\n endpoint?: string\n /** OTLP headers. Reads OTEL_EXPORTER_OTLP_HEADERS env by default. */\n headers?: Record<string, string>\n /** Batch size before flush. Default 64. */\n batchSize?: number\n /** Flush interval ms. Default 5000. */\n flushIntervalMs?: number\n /** Resource attributes stamped on every export. */\n resourceAttributes?: Record<string, string | number | boolean>\n /** Service name. Default 'agent-runtime'. */\n serviceName?: string\n}\n\nexport interface OtelExporter {\n /** Export a span. */\n exportSpan(span: OtelSpan): void\n /** Force flush pending spans. */\n flush(): Promise<void>\n /** Shutdown cleanly. */\n shutdown(): Promise<void>\n}\n\nexport interface OtelSpan {\n traceId: string\n spanId: string\n parentSpanId?: string\n name: string\n kind?: number\n startTimeUnixNano: string\n endTimeUnixNano: string\n attributes?: OtelAttribute[]\n status?: { code: number; message?: string }\n}\n\nexport interface OtelAttribute {\n key: string\n value: { stringValue?: string; intValue?: string; doubleValue?: number; boolValue?: boolean }\n}\n\ninterface OtlpResourceSpans {\n resource: { attributes: OtelAttribute[] }\n scopeSpans: Array<{ scope: { name: string; version: string }; spans: OtelSpan[] }>\n}\n\ninterface OtlpExport {\n resourceSpans: OtlpResourceSpans[]\n}\n\nconst SCOPE = { name: '@tangle-network/agent-runtime', version: '0.83.0' }\n\n/**\n * Current (non-deprecated) OpenTelemetry GenAI semantic-convention keys.\n * Registry: https://opentelemetry.io/docs/specs/semconv/registry/attributes/gen-ai/\n * NB: `gen_ai.system` / `gen_ai.usage.prompt_tokens` / `completion_tokens` are\n * DEPRECATED — do not emit them. We use `provider.name` + `input/output_tokens`.\n */\nconst GEN_AI = {\n operation: 'gen_ai.operation.name',\n agentName: 'gen_ai.agent.name',\n conversationId: 'gen_ai.conversation.id',\n inputTokens: 'gen_ai.usage.input_tokens',\n outputTokens: 'gen_ai.usage.output_tokens',\n} as const\n\n/**\n * Create an OTEL exporter. Returns undefined when no endpoint is configured.\n */\nexport function createOtelExporter(config?: OtelExportConfig): OtelExporter | undefined {\n const resolvedEndpoint =\n config?.endpoint ??\n (typeof process !== 'undefined' ? process.env.OTEL_EXPORTER_OTLP_ENDPOINT : undefined)\n if (!resolvedEndpoint) return undefined\n const endpoint: string = resolvedEndpoint\n\n const headers = config?.headers ?? parseHeadersFromEnv()\n const batchSize = config?.batchSize ?? 64\n const flushIntervalMs = config?.flushIntervalMs ?? 5000\n const serviceName = config?.serviceName ?? 'agent-runtime'\n const resourceAttrs = config?.resourceAttributes ?? {}\n\n const pending: OtelSpan[] = []\n let timer: ReturnType<typeof setInterval> | undefined\n let stopped = false\n\n const exporter: OtelExporter = {\n exportSpan(span: OtelSpan): void {\n if (stopped) return\n pending.push(span)\n if (pending.length >= batchSize) {\n void doFlush()\n }\n },\n\n async flush(): Promise<void> {\n await doFlush()\n },\n\n async shutdown(): Promise<void> {\n stopped = true\n if (timer !== undefined) {\n clearInterval(timer)\n timer = undefined\n }\n await doFlush()\n },\n }\n\n timer = setInterval(() => {\n if (pending.length > 0) void doFlush()\n }, flushIntervalMs)\n if (typeof timer === 'object' && 'unref' in timer) {\n ;(timer as NodeJS.Timeout).unref()\n }\n\n async function doFlush(): Promise<void> {\n if (pending.length === 0) return\n const batch = pending.splice(0)\n const body: OtlpExport = {\n resourceSpans: [\n {\n resource: {\n attributes: toAttributes({\n 'service.name': serviceName,\n ...resourceAttrs,\n }),\n },\n scopeSpans: [{ scope: SCOPE, spans: batch }],\n },\n ],\n }\n const url = `${endpoint.replace(/\\/+$/, '')}/v1/traces`\n try {\n await fetch(url, {\n method: 'POST',\n headers: { 'content-type': 'application/json', ...headers },\n body: JSON.stringify(body),\n })\n } catch {\n // Best-effort — telemetry export must not crash the runtime.\n }\n }\n\n return exporter\n}\n\n/**\n * Convert a LoopTraceEvent into an OtelSpan for export.\n */\nexport function loopEventToOtelSpan(\n event: {\n kind: string\n runId: string\n timestamp: number\n payload: object\n },\n traceId: string,\n parentSpanId?: string,\n): OtelSpan {\n const spanId = generateSpanId()\n const attrs: Record<string, string | number | boolean> = {\n 'loop.event_kind': event.kind,\n 'loop.run_id': event.runId,\n }\n for (const [k, v] of Object.entries(event.payload)) {\n if (typeof v === 'string' || typeof v === 'number' || typeof v === 'boolean') {\n attrs[`loop.${k}`] = v\n }\n }\n const ts = msToNs(event.timestamp)\n return {\n traceId: padTraceId(traceId),\n spanId,\n parentSpanId: parentSpanId ? padSpanId(parentSpanId) : undefined,\n name: event.kind,\n kind: 1,\n startTimeUnixNano: ts,\n endTimeUnixNano: ts,\n attributes: toAttributes(attrs),\n status: { code: 1 },\n }\n}\n\n/**\n * Build a single flat OtelSpan whose attribute keys are emitted VERBATIM — no\n * `loop.` namespace. Use for non-loop spans (e.g. the intelligence per-run\n * span) whose keys ARE the downstream contract (`gen_ai.request.model`,\n * `tangle.sessionId`) and are exact-matched by readers. `loopEventToOtelSpan`\n * namespaces payload keys because loop payload fields are free-form; do not\n * reuse it for spans with contract keys.\n */\nexport function flatOtelSpan(\n name: string,\n attributes: Record<string, string | number | boolean>,\n traceId: string,\n timestampMs: number,\n parentSpanId?: string,\n): OtelSpan {\n const ts = msToNs(timestampMs)\n return {\n traceId: padTraceId(traceId),\n spanId: generateSpanId(),\n parentSpanId: parentSpanId ? padSpanId(parentSpanId) : undefined,\n name,\n kind: 1,\n startTimeUnixNano: ts,\n endTimeUnixNano: ts,\n attributes: toAttributes(attributes),\n status: { code: 1 },\n }\n}\n\n/**\n * Sink-neutral node in a reconstructed loop span tree. The root node's\n * `parentSpanId` is `undefined` — sinks decide how to parent it (the OTEL\n * mapper attaches the inherited delegation span; the delegation journal\n * leaves it as the tree root).\n */\nexport interface LoopSpanNode {\n spanId: string\n parentSpanId?: string\n /** `'loop'` | `'loop.round'` | `'loop.iteration'`. */\n name: string\n /** Topology level: loop root, plan round, or iteration branch. */\n kind: 'loop' | 'round' | 'branch'\n startMs: number\n endMs: number\n attrs: Record<string, string | number | boolean>\n /** True when the iteration carried an error — maps to OTEL status code 2. */\n error: boolean\n}\n\n/**\n * Build a nested, real-duration OTLP span tree for ONE loop run from its full\n * ordered `LoopTraceEvent` stream. Unlike `loopEventToOtelSpan` (one flat,\n * zero-duration span per event), this reconstructs the topology hierarchy a\n * GenAI trace viewer renders natively:\n *\n * loop (invoke_workflow)\n * └─ loop.round[k] (invoke_workflow) ← tangle.loop.move.{kind,width,rationale}\n * ├─ loop.iteration[i] (invoke_agent) ← gen_ai.agent.name + usage + verdict + placement\n * └─ …\n *\n * Attributes follow the current GenAI semconv (`gen_ai.*`) where they apply and\n * a namespaced `tangle.loop.*` / `tangle.cost.usd` extension for topology /\n * verdict / placement / cost (not yet standardized). Pure: feed it a buffered\n * per-runId event array (e.g. flushed on `loop.ended`) and export the result.\n */\nexport function buildLoopOtelSpans(\n events: ReadonlyArray<{ kind: string; runId: string; timestamp: number; payload: object }>,\n traceId: string,\n rootParentSpanId?: string,\n): OtelSpan[] {\n const tid = padTraceId(traceId)\n return buildLoopSpanNodes(events).map((node) => ({\n traceId: tid,\n spanId: node.spanId,\n parentSpanId: node.parentSpanId\n ? padSpanId(node.parentSpanId)\n : rootParentSpanId\n ? padSpanId(rootParentSpanId)\n : undefined,\n name: node.name,\n kind: 1,\n startTimeUnixNano: msToNs(node.startMs),\n endTimeUnixNano: msToNs(node.endMs),\n attributes: toAttributes(node.attrs),\n status: { code: node.error ? 2 : 1 },\n }))\n}\n\n/**\n * Sink-neutral core behind {@link buildLoopOtelSpans}: reconstruct the\n * loop → round → branch span tree from one run's ordered `LoopTraceEvent`\n * stream. Consumed by the OTEL mapper above and by the MCP delegation\n * journal's compact trace tee — one topology reconstruction, two sinks.\n * Tolerates partial streams (a run that never reached `loop.ended` closes\n * at the last observed event's timestamp).\n */\nexport function buildLoopSpanNodes(\n events: ReadonlyArray<{ kind: string; runId: string; timestamp: number; payload: object }>,\n): LoopSpanNode[] {\n if (events.length === 0) return []\n const out: LoopSpanNode[] = []\n const num = (v: unknown): number | undefined =>\n typeof v === 'number' && Number.isFinite(v) ? v : undefined\n const str = (v: unknown): string | undefined =>\n typeof v === 'string' && v.length > 0 ? v : undefined\n const rec = (v: unknown): Record<string, unknown> =>\n v && typeof v === 'object' ? (v as Record<string, unknown>) : {}\n\n const started = events.find((e) => e.kind === 'loop.started')\n const ended = events.find((e) => e.kind === 'loop.ended')\n const runId = events[0]?.runId ?? ''\n const rootStart = started?.timestamp ?? events[0]!.timestamp\n const rootEnd = ended?.timestamp ?? events[events.length - 1]!.timestamp\n const rootId = generateSpanId()\n\n const make = (\n spanId: string,\n parentSpanId: string | undefined,\n name: string,\n kind: LoopSpanNode['kind'],\n startMs: number,\n endMs: number,\n attrs: Record<string, string | number | boolean>,\n error = false,\n ): LoopSpanNode => ({\n spanId,\n parentSpanId,\n name,\n kind,\n startMs,\n endMs,\n attrs,\n error,\n })\n\n // root\n const sp = rec(started?.payload)\n const rootAttrs: Record<string, string | number | boolean> = {\n [GEN_AI.operation]: 'invoke_workflow',\n [GEN_AI.conversationId]: runId,\n // Explicit run identity under the tangle namespace so a consuming system of\n // record (Tangle Intelligence's run spine) maps this span tree to one run\n // deterministically, instead of inferring it from the conversation id.\n 'tangle.run.id': runId,\n 'tangle.loop.driver': str(sp.driver) ?? 'driver',\n }\n if (Array.isArray(sp.agentRunNames) && sp.agentRunNames.length > 0) {\n rootAttrs['tangle.loop.agents'] = sp.agentRunNames.map(String).join(',')\n // Subject grain: the primary agent/profile this run drove. Lets the spine\n // group runs under the thing being worked on rather than one service-wide\n // bucket. `agentRunNames[0]` is the lead agent on a fan-out.\n rootAttrs['tangle.subject.key'] = String(sp.agentRunNames[0])\n }\n if (ended) {\n const ep = rec(ended.payload)\n const win = num(ep.winnerIterationIndex)\n if (win !== undefined) rootAttrs['tangle.loop.winner.iteration_index'] = win\n const cost = num(ep.totalCostUsd)\n if (cost !== undefined) rootAttrs['tangle.cost.usd'] = cost\n const dur = num(ep.durationMs)\n if (dur !== undefined) rootAttrs['tangle.loop.duration_ms'] = dur\n const iters = num(ep.iterations)\n if (iters !== undefined) rootAttrs['tangle.loop.iterations'] = iters\n }\n out.push(make(rootId, undefined, 'loop', 'loop', rootStart, rootEnd, rootAttrs))\n\n // rounds + iterations\n const iterStartTs = new Map<number, number>()\n const placementByIdx = new Map<number, Record<string, string>>()\n let currentRoundId: string | undefined\n let pendingRound:\n | { id: string; start: number; attrs: Record<string, string | number | boolean> }\n | undefined\n const flushRound = (endMs: number) => {\n if (!pendingRound) return\n out.push(\n make(\n pendingRound.id,\n rootId,\n 'loop.round',\n 'round',\n pendingRound.start,\n endMs,\n pendingRound.attrs,\n ),\n )\n pendingRound = undefined\n }\n\n for (const e of events) {\n const p = rec(e.payload)\n switch (e.kind) {\n case 'loop.plan': {\n flushRound(e.timestamp)\n const id = generateSpanId()\n const roundIdx = num(p.roundIndex) ?? 0\n const attrs: Record<string, string | number | boolean> = {\n [GEN_AI.operation]: 'invoke_workflow',\n 'tangle.loop.round.index': roundIdx,\n 'tangle.loop.move.kind': str(p.moveKind) ?? 'unknown',\n 'tangle.loop.move.round': roundIdx,\n 'tangle.loop.move.width': num(p.plannedCount) ?? 0,\n }\n const r = str(p.rationale)\n if (r) attrs['tangle.loop.move.rationale'] = r\n const parent = num(p.parentIndex)\n if (parent !== undefined) attrs['tangle.loop.move.parent_index'] = parent\n if (Array.isArray(p.childIndices) && p.childIndices.length > 0) {\n attrs['tangle.loop.move.child_indices'] = p.childIndices.map(String).join(',')\n }\n pendingRound = { id, start: e.timestamp, attrs }\n currentRoundId = id\n break\n }\n case 'loop.iteration.started': {\n const idx = num(p.iterationIndex)\n if (idx !== undefined) iterStartTs.set(idx, e.timestamp)\n break\n }\n case 'loop.iteration.dispatch': {\n const idx = num(p.iterationIndex)\n if (idx === undefined) break\n const place: Record<string, string> = {}\n const kind = str(p.placement)\n if (kind) place['tangle.loop.placement.kind'] = kind\n const sid = str(p.sandboxId)\n if (sid) place['tangle.sandbox.id'] = sid\n const fid = str(p.fleetId)\n if (fid) place['tangle.fleet.id'] = fid\n const mid = str(p.machineId)\n if (mid) place['tangle.machine.id'] = mid\n placementByIdx.set(idx, place)\n break\n }\n case 'loop.iteration.ended': {\n const idx = num(p.iterationIndex) ?? 0\n const start = iterStartTs.get(idx) ?? e.timestamp\n const err = str(p.error)\n const attrs: Record<string, string | number | boolean> = {\n [GEN_AI.operation]: 'invoke_agent',\n 'tangle.loop.iteration.index': idx,\n }\n const agent = str(p.agentRunName)\n if (agent) attrs[GEN_AI.agentName] = agent\n const tu = rec(p.tokenUsage)\n const inTok = num(tu.input)\n if (inTok !== undefined) attrs[GEN_AI.inputTokens] = inTok\n const outTok = num(tu.output)\n if (outTok !== undefined) attrs[GEN_AI.outputTokens] = outTok\n const cost = num(p.costUsd)\n if (cost !== undefined) attrs['tangle.cost.usd'] = cost\n const verdict = rec(p.verdict)\n if (typeof verdict.valid === 'boolean') attrs['tangle.loop.verdict.valid'] = verdict.valid\n const score = num(verdict.score)\n if (score !== undefined) attrs['tangle.loop.verdict.score'] = score\n if (err) attrs['tangle.loop.error'] = err\n const gid = num(p.groupId)\n if (gid !== undefined) attrs['tangle.loop.iteration.group_id'] = gid\n const par = num(p.parentIndex)\n if (par !== undefined) attrs['tangle.loop.iteration.parent_index'] = par\n const dur = num(p.durationMs)\n if (dur !== undefined) attrs['tangle.loop.iteration.duration_ms'] = dur\n const preview = str(p.outputPreview)\n if (preview) attrs['tangle.loop.iteration.output_preview'] = preview\n Object.assign(attrs, placementByIdx.get(idx) ?? {})\n out.push(\n make(\n generateSpanId(),\n currentRoundId ?? rootId,\n 'loop.iteration',\n 'branch',\n start,\n e.timestamp,\n attrs,\n err !== undefined,\n ),\n )\n break\n }\n case 'loop.decision': {\n if (pendingRound) {\n const dec = str(p.decision)\n if (dec) pendingRound.attrs['tangle.loop.decision'] = dec\n flushRound(e.timestamp)\n }\n currentRoundId = undefined\n break\n }\n }\n }\n flushRound(rootEnd)\n return out\n}\n\nfunction parseHeadersFromEnv(): Record<string, string> {\n if (typeof process === 'undefined') return {}\n const raw = process.env.OTEL_EXPORTER_OTLP_HEADERS\n if (!raw) return {}\n const out: Record<string, string> = {}\n for (const pair of raw.split(',')) {\n const eq = pair.indexOf('=')\n if (eq < 0) continue\n const key = pair.slice(0, eq).trim()\n const value = pair.slice(eq + 1).trim()\n if (key) out[key] = value\n }\n return out\n}\n\nfunction toAttributes(record: Record<string, string | number | boolean>): OtelAttribute[] {\n return Object.entries(record).map(([key, value]) => ({\n key,\n value:\n typeof value === 'number'\n ? Number.isInteger(value)\n ? { intValue: value.toString() }\n : { doubleValue: value }\n : typeof value === 'boolean'\n ? { boolValue: value }\n : { stringValue: value },\n }))\n}\n\nfunction msToNs(ms: number): string {\n return (BigInt(Math.floor(ms)) * 1_000_000n).toString()\n}\n\nfunction padSpanId(id: string): string {\n const cleaned = id.replace(/-/g, '')\n return cleaned.slice(0, 16).padEnd(16, '0')\n}\n\nfunction padTraceId(id: string): string {\n const cleaned = id.replace(/-/g, '')\n return cleaned.slice(0, 32).padEnd(32, '0')\n}\n\nfunction generateSpanId(): string {\n const bytes = new Uint8Array(8)\n if (typeof globalThis.crypto?.getRandomValues === 'function') {\n globalThis.crypto.getRandomValues(bytes)\n } else {\n for (let i = 0; i < 8; i++) bytes[i] = Math.floor(Math.random() * 256)\n }\n return Array.from(bytes)\n .map((b) => b.toString(16).padStart(2, '0'))\n .join('')\n}\n\n// ─── Eval-run ingest (self-improvement provenance) ───────────────────────────\n//\n// Tangle Intelligence has a first-class, non-trace record for self-improvement\n// runs: POST /v1/ingest/eval-runs (\"Mode D\"). Each generation carries a\n// `surfaceHash` (the proposed-change identity) + arbitrary `surface` provenance;\n// a later `gate-decided` event re-emits the same `runId` (idempotent upsert) with\n// a real `gateDecision` + `holdoutLift`, so proposal→verdict is one diffable\n// record. This is how a consumer's RSI loop records WHAT it changed, WHY, from\n// which evidence — the audit trail behind agentic self-improvement.\n\n/** Wire version the eval-runs ingest enforces (X-Tangle-Wire-Version + body). */\nexport const INTELLIGENCE_WIRE_VERSION = '2026-05-26.v1'\n\nexport interface EvalRunGeneration {\n /** 0-based ordinal of this generation within the run (required by ingest). */\n index: number\n /** Identity of the proposed surface change (content-addressed hash). */\n surfaceHash: string\n /** Arbitrary provenance for this generation (rationale, evidence, source). */\n surface?: unknown\n /** Per-scenario results; empty until the generation is measured. */\n cells?: unknown[]\n /** Mean composite score (0 when unmeasured — pair with labels.measured). */\n compositeMean: number\n costUsd: number\n durationMs: number\n}\n\nexport interface EvalRunEvent {\n runId: string\n runDir: string\n /** ISO timestamp. */\n timestamp: string\n status:\n | 'started'\n | 'baseline-complete'\n | 'generation-complete'\n | 'gate-decided'\n | 'finished'\n | 'errored'\n labels?: Record<string, string>\n baseline?: EvalRunGeneration\n generations?: EvalRunGeneration[]\n gateDecision?: 'ship' | 'hold' | 'need_more_work' | 'model_ceiling' | 'arch_ceiling'\n holdoutLift?: number\n totalCostUsd: number\n totalDurationMs: number\n errorMessage?: string\n}\n\nexport interface EvalRunsExportConfig {\n /** Bearer key — tenant is resolved server-side from it. Reads TANGLE_API_KEY. */\n apiKey?: string\n /** Intelligence base. Reads INTELLIGENCE_BASE env, else prod. */\n base?: string\n /** Idempotency-Key header (e.g. the runId) — safe retries + upsert. */\n idempotencyKey?: string\n}\n\nexport interface EvalRunsExportResult {\n ok: boolean\n status: number\n accepted: number\n rejected: Array<{ index: number; reason: string }>\n}\n\nconst DEFAULT_INTELLIGENCE_BASE = 'https://intelligence.tangle.tools'\n\n/**\n * Ship self-improvement eval-run events to Tangle Intelligence. Unlike the\n * best-effort span exporter, this RESOLVES with the ingest verdict (accepted /\n * rejected per event) so a consumer's loop can assert its provenance landed.\n * Throws only on a missing key or network failure.\n */\nexport async function exportEvalRuns(\n events: EvalRunEvent[],\n config?: EvalRunsExportConfig,\n): Promise<EvalRunsExportResult> {\n if (events.length === 0) return { ok: true, status: 0, accepted: 0, rejected: [] }\n const apiKey =\n config?.apiKey ?? (typeof process !== 'undefined' ? process.env.TANGLE_API_KEY : undefined)\n if (!apiKey)\n throw new Error('exportEvalRuns: apiKey required (pass config.apiKey or set TANGLE_API_KEY)')\n const base =\n config?.base ??\n (typeof process !== 'undefined' ? process.env.INTELLIGENCE_BASE : undefined) ??\n DEFAULT_INTELLIGENCE_BASE\n const url = `${base.replace(/\\/+$/, '')}/v1/ingest/eval-runs`\n const res = await fetch(url, {\n method: 'POST',\n headers: {\n 'content-type': 'application/json',\n authorization: `Bearer ${apiKey}`,\n 'X-Tangle-Wire-Version': INTELLIGENCE_WIRE_VERSION,\n ...(config?.idempotencyKey ? { 'Idempotency-Key': config.idempotencyKey } : {}),\n },\n body: JSON.stringify({ wireVersion: INTELLIGENCE_WIRE_VERSION, events }),\n })\n let parsed: { accepted?: number; rejected?: Array<{ index: number; reason: string }> } = {}\n try {\n parsed = (await res.json()) as typeof parsed\n } catch {\n // non-JSON body (e.g. 5xx HTML) — leave parsed empty\n }\n return {\n ok: res.ok,\n status: res.status,\n accepted: parsed.accepted ?? (res.ok ? events.length : 0),\n rejected: parsed.rejected ?? [],\n }\n}\n"],"mappings":";AA6DA,IAAM,QAAQ,EAAE,MAAM,iCAAiC,SAAS,SAAS;AAQzE,IAAM,SAAS;AAAA,EACb,WAAW;AAAA,EACX,WAAW;AAAA,EACX,gBAAgB;AAAA,EAChB,aAAa;AAAA,EACb,cAAc;AAChB;AAKO,SAAS,mBAAmB,QAAqD;AACtF,QAAM,mBACJ,QAAQ,aACP,OAAO,YAAY,cAAc,QAAQ,IAAI,8BAA8B;AAC9E,MAAI,CAAC,iBAAkB,QAAO;AAC9B,QAAM,WAAmB;AAEzB,QAAM,UAAU,QAAQ,WAAW,oBAAoB;AACvD,QAAM,YAAY,QAAQ,aAAa;AACvC,QAAM,kBAAkB,QAAQ,mBAAmB;AACnD,QAAM,cAAc,QAAQ,eAAe;AAC3C,QAAM,gBAAgB,QAAQ,sBAAsB,CAAC;AAErD,QAAM,UAAsB,CAAC;AAC7B,MAAI;AACJ,MAAI,UAAU;AAEd,QAAM,WAAyB;AAAA,IAC7B,WAAW,MAAsB;AAC/B,UAAI,QAAS;AACb,cAAQ,KAAK,IAAI;AACjB,UAAI,QAAQ,UAAU,WAAW;AAC/B,aAAK,QAAQ;AAAA,MACf;AAAA,IACF;AAAA,IAEA,MAAM,QAAuB;AAC3B,YAAM,QAAQ;AAAA,IAChB;AAAA,IAEA,MAAM,WAA0B;AAC9B,gBAAU;AACV,UAAI,UAAU,QAAW;AACvB,sBAAc,KAAK;AACnB,gBAAQ;AAAA,MACV;AACA,YAAM,QAAQ;AAAA,IAChB;AAAA,EACF;AAEA,UAAQ,YAAY,MAAM;AACxB,QAAI,QAAQ,SAAS,EAAG,MAAK,QAAQ;AAAA,EACvC,GAAG,eAAe;AAClB,MAAI,OAAO,UAAU,YAAY,WAAW,OAAO;AACjD;AAAC,IAAC,MAAyB,MAAM;AAAA,EACnC;AAEA,iBAAe,UAAyB;AACtC,QAAI,QAAQ,WAAW,EAAG;AAC1B,UAAM,QAAQ,QAAQ,OAAO,CAAC;AAC9B,UAAM,OAAmB;AAAA,MACvB,eAAe;AAAA,QACb;AAAA,UACE,UAAU;AAAA,YACR,YAAY,aAAa;AAAA,cACvB,gBAAgB;AAAA,cAChB,GAAG;AAAA,YACL,CAAC;AAAA,UACH;AAAA,UACA,YAAY,CAAC,EAAE,OAAO,OAAO,OAAO,MAAM,CAAC;AAAA,QAC7C;AAAA,MACF;AAAA,IACF;AACA,UAAM,MAAM,GAAG,SAAS,QAAQ,QAAQ,EAAE,CAAC;AAC3C,QAAI;AACF,YAAM,MAAM,KAAK;AAAA,QACf,QAAQ;AAAA,QACR,SAAS,EAAE,gBAAgB,oBAAoB,GAAG,QAAQ;AAAA,QAC1D,MAAM,KAAK,UAAU,IAAI;AAAA,MAC3B,CAAC;AAAA,IACH,QAAQ;AAAA,IAER;AAAA,EACF;AAEA,SAAO;AACT;AAKO,SAAS,oBACd,OAMA,SACA,cACU;AACV,QAAM,SAAS,eAAe;AAC9B,QAAM,QAAmD;AAAA,IACvD,mBAAmB,MAAM;AAAA,IACzB,eAAe,MAAM;AAAA,EACvB;AACA,aAAW,CAAC,GAAG,CAAC,KAAK,OAAO,QAAQ,MAAM,OAAO,GAAG;AAClD,QAAI,OAAO,MAAM,YAAY,OAAO,MAAM,YAAY,OAAO,MAAM,WAAW;AAC5E,YAAM,QAAQ,CAAC,EAAE,IAAI;AAAA,IACvB;AAAA,EACF;AACA,QAAM,KAAK,OAAO,MAAM,SAAS;AACjC,SAAO;AAAA,IACL,SAAS,WAAW,OAAO;AAAA,IAC3B;AAAA,IACA,cAAc,eAAe,UAAU,YAAY,IAAI;AAAA,IACvD,MAAM,MAAM;AAAA,IACZ,MAAM;AAAA,IACN,mBAAmB;AAAA,IACnB,iBAAiB;AAAA,IACjB,YAAY,aAAa,KAAK;AAAA,IAC9B,QAAQ,EAAE,MAAM,EAAE;AAAA,EACpB;AACF;AAUO,SAAS,aACd,MACA,YACA,SACA,aACA,cACU;AACV,QAAM,KAAK,OAAO,WAAW;AAC7B,SAAO;AAAA,IACL,SAAS,WAAW,OAAO;AAAA,IAC3B,QAAQ,eAAe;AAAA,IACvB,cAAc,eAAe,UAAU,YAAY,IAAI;AAAA,IACvD;AAAA,IACA,MAAM;AAAA,IACN,mBAAmB;AAAA,IACnB,iBAAiB;AAAA,IACjB,YAAY,aAAa,UAAU;AAAA,IACnC,QAAQ,EAAE,MAAM,EAAE;AAAA,EACpB;AACF;AAsCO,SAAS,mBACd,QACA,SACA,kBACY;AACZ,QAAM,MAAM,WAAW,OAAO;AAC9B,SAAO,mBAAmB,MAAM,EAAE,IAAI,CAAC,UAAU;AAAA,IAC/C,SAAS;AAAA,IACT,QAAQ,KAAK;AAAA,IACb,cAAc,KAAK,eACf,UAAU,KAAK,YAAY,IAC3B,mBACE,UAAU,gBAAgB,IAC1B;AAAA,IACN,MAAM,KAAK;AAAA,IACX,MAAM;AAAA,IACN,mBAAmB,OAAO,KAAK,OAAO;AAAA,IACtC,iBAAiB,OAAO,KAAK,KAAK;AAAA,IAClC,YAAY,aAAa,KAAK,KAAK;AAAA,IACnC,QAAQ,EAAE,MAAM,KAAK,QAAQ,IAAI,EAAE;AAAA,EACrC,EAAE;AACJ;AAUO,SAAS,mBACd,QACgB;AAChB,MAAI,OAAO,WAAW,EAAG,QAAO,CAAC;AACjC,QAAM,MAAsB,CAAC;AAC7B,QAAM,MAAM,CAAC,MACX,OAAO,MAAM,YAAY,OAAO,SAAS,CAAC,IAAI,IAAI;AACpD,QAAM,MAAM,CAAC,MACX,OAAO,MAAM,YAAY,EAAE,SAAS,IAAI,IAAI;AAC9C,QAAM,MAAM,CAAC,MACX,KAAK,OAAO,MAAM,WAAY,IAAgC,CAAC;AAEjE,QAAM,UAAU,OAAO,KAAK,CAAC,MAAM,EAAE,SAAS,cAAc;AAC5D,QAAM,QAAQ,OAAO,KAAK,CAAC,MAAM,EAAE,SAAS,YAAY;AACxD,QAAM,QAAQ,OAAO,CAAC,GAAG,SAAS;AAClC,QAAM,YAAY,SAAS,aAAa,OAAO,CAAC,EAAG;AACnD,QAAM,UAAU,OAAO,aAAa,OAAO,OAAO,SAAS,CAAC,EAAG;AAC/D,QAAM,SAAS,eAAe;AAE9B,QAAM,OAAO,CACX,QACA,cACA,MACA,MACA,SACA,OACA,OACA,QAAQ,WACU;AAAA,IAClB;AAAA,IACA;AAAA,IACA;AAAA,IACA;AAAA,IACA;AAAA,IACA;AAAA,IACA;AAAA,IACA;AAAA,EACF;AAGA,QAAM,KAAK,IAAI,SAAS,OAAO;AAC/B,QAAM,YAAuD;AAAA,IAC3D,CAAC,OAAO,SAAS,GAAG;AAAA,IACpB,CAAC,OAAO,cAAc,GAAG;AAAA;AAAA;AAAA;AAAA,IAIzB,iBAAiB;AAAA,IACjB,sBAAsB,IAAI,GAAG,MAAM,KAAK;AAAA,EAC1C;AACA,MAAI,MAAM,QAAQ,GAAG,aAAa,KAAK,GAAG,cAAc,SAAS,GAAG;AAClE,cAAU,oBAAoB,IAAI,GAAG,cAAc,IAAI,MAAM,EAAE,KAAK,GAAG;AAIvE,cAAU,oBAAoB,IAAI,OAAO,GAAG,cAAc,CAAC,CAAC;AAAA,EAC9D;AACA,MAAI,OAAO;AACT,UAAM,KAAK,IAAI,MAAM,OAAO;AAC5B,UAAM,MAAM,IAAI,GAAG,oBAAoB;AACvC,QAAI,QAAQ,OAAW,WAAU,oCAAoC,IAAI;AACzE,UAAM,OAAO,IAAI,GAAG,YAAY;AAChC,QAAI,SAAS,OAAW,WAAU,iBAAiB,IAAI;AACvD,UAAM,MAAM,IAAI,GAAG,UAAU;AAC7B,QAAI,QAAQ,OAAW,WAAU,yBAAyB,IAAI;AAC9D,UAAM,QAAQ,IAAI,GAAG,UAAU;AAC/B,QAAI,UAAU,OAAW,WAAU,wBAAwB,IAAI;AAAA,EACjE;AACA,MAAI,KAAK,KAAK,QAAQ,QAAW,QAAQ,QAAQ,WAAW,SAAS,SAAS,CAAC;AAG/E,QAAM,cAAc,oBAAI,IAAoB;AAC5C,QAAM,iBAAiB,oBAAI,IAAoC;AAC/D,MAAI;AACJ,MAAI;AAGJ,QAAM,aAAa,CAAC,UAAkB;AACpC,QAAI,CAAC,aAAc;AACnB,QAAI;AAAA,MACF;AAAA,QACE,aAAa;AAAA,QACb;AAAA,QACA;AAAA,QACA;AAAA,QACA,aAAa;AAAA,QACb;AAAA,QACA,aAAa;AAAA,MACf;AAAA,IACF;AACA,mBAAe;AAAA,EACjB;AAEA,aAAW,KAAK,QAAQ;AACtB,UAAM,IAAI,IAAI,EAAE,OAAO;AACvB,YAAQ,EAAE,MAAM;AAAA,MACd,KAAK,aAAa;AAChB,mBAAW,EAAE,SAAS;AACtB,cAAM,KAAK,eAAe;AAC1B,cAAM,WAAW,IAAI,EAAE,UAAU,KAAK;AACtC,cAAM,QAAmD;AAAA,UACvD,CAAC,OAAO,SAAS,GAAG;AAAA,UACpB,2BAA2B;AAAA,UAC3B,yBAAyB,IAAI,EAAE,QAAQ,KAAK;AAAA,UAC5C,0BAA0B;AAAA,UAC1B,0BAA0B,IAAI,EAAE,YAAY,KAAK;AAAA,QACnD;AACA,cAAM,IAAI,IAAI,EAAE,SAAS;AACzB,YAAI,EAAG,OAAM,4BAA4B,IAAI;AAC7C,cAAM,SAAS,IAAI,EAAE,WAAW;AAChC,YAAI,WAAW,OAAW,OAAM,+BAA+B,IAAI;AACnE,YAAI,MAAM,QAAQ,EAAE,YAAY,KAAK,EAAE,aAAa,SAAS,GAAG;AAC9D,gBAAM,gCAAgC,IAAI,EAAE,aAAa,IAAI,MAAM,EAAE,KAAK,GAAG;AAAA,QAC/E;AACA,uBAAe,EAAE,IAAI,OAAO,EAAE,WAAW,MAAM;AAC/C,yBAAiB;AACjB;AAAA,MACF;AAAA,MACA,KAAK,0BAA0B;AAC7B,cAAM,MAAM,IAAI,EAAE,cAAc;AAChC,YAAI,QAAQ,OAAW,aAAY,IAAI,KAAK,EAAE,SAAS;AACvD;AAAA,MACF;AAAA,MACA,KAAK,2BAA2B;AAC9B,cAAM,MAAM,IAAI,EAAE,cAAc;AAChC,YAAI,QAAQ,OAAW;AACvB,cAAM,QAAgC,CAAC;AACvC,cAAM,OAAO,IAAI,EAAE,SAAS;AAC5B,YAAI,KAAM,OAAM,4BAA4B,IAAI;AAChD,cAAM,MAAM,IAAI,EAAE,SAAS;AAC3B,YAAI,IAAK,OAAM,mBAAmB,IAAI;AACtC,cAAM,MAAM,IAAI,EAAE,OAAO;AACzB,YAAI,IAAK,OAAM,iBAAiB,IAAI;AACpC,cAAM,MAAM,IAAI,EAAE,SAAS;AAC3B,YAAI,IAAK,OAAM,mBAAmB,IAAI;AACtC,uBAAe,IAAI,KAAK,KAAK;AAC7B;AAAA,MACF;AAAA,MACA,KAAK,wBAAwB;AAC3B,cAAM,MAAM,IAAI,EAAE,cAAc,KAAK;AACrC,cAAM,QAAQ,YAAY,IAAI,GAAG,KAAK,EAAE;AACxC,cAAM,MAAM,IAAI,EAAE,KAAK;AACvB,cAAM,QAAmD;AAAA,UACvD,CAAC,OAAO,SAAS,GAAG;AAAA,UACpB,+BAA+B;AAAA,QACjC;AACA,cAAM,QAAQ,IAAI,EAAE,YAAY;AAChC,YAAI,MAAO,OAAM,OAAO,SAAS,IAAI;AACrC,cAAM,KAAK,IAAI,EAAE,UAAU;AAC3B,cAAM,QAAQ,IAAI,GAAG,KAAK;AAC1B,YAAI,UAAU,OAAW,OAAM,OAAO,WAAW,IAAI;AACrD,cAAM,SAAS,IAAI,GAAG,MAAM;AAC5B,YAAI,WAAW,OAAW,OAAM,OAAO,YAAY,IAAI;AACvD,cAAM,OAAO,IAAI,EAAE,OAAO;AAC1B,YAAI,SAAS,OAAW,OAAM,iBAAiB,IAAI;AACnD,cAAM,UAAU,IAAI,EAAE,OAAO;AAC7B,YAAI,OAAO,QAAQ,UAAU,UAAW,OAAM,2BAA2B,IAAI,QAAQ;AACrF,cAAM,QAAQ,IAAI,QAAQ,KAAK;AAC/B,YAAI,UAAU,OAAW,OAAM,2BAA2B,IAAI;AAC9D,YAAI,IAAK,OAAM,mBAAmB,IAAI;AACtC,cAAM,MAAM,IAAI,EAAE,OAAO;AACzB,YAAI,QAAQ,OAAW,OAAM,gCAAgC,IAAI;AACjE,cAAM,MAAM,IAAI,EAAE,WAAW;AAC7B,YAAI,QAAQ,OAAW,OAAM,oCAAoC,IAAI;AACrE,cAAM,MAAM,IAAI,EAAE,UAAU;AAC5B,YAAI,QAAQ,OAAW,OAAM,mCAAmC,IAAI;AACpE,cAAM,UAAU,IAAI,EAAE,aAAa;AACnC,YAAI,QAAS,OAAM,sCAAsC,IAAI;AAC7D,eAAO,OAAO,OAAO,eAAe,IAAI,GAAG,KAAK,CAAC,CAAC;AAClD,YAAI;AAAA,UACF;AAAA,YACE,eAAe;AAAA,YACf,kBAAkB;AAAA,YAClB;AAAA,YACA;AAAA,YACA;AAAA,YACA,EAAE;AAAA,YACF;AAAA,YACA,QAAQ;AAAA,UACV;AAAA,QACF;AACA;AAAA,MACF;AAAA,MACA,KAAK,iBAAiB;AACpB,YAAI,cAAc;AAChB,gBAAM,MAAM,IAAI,EAAE,QAAQ;AAC1B,cAAI,IAAK,cAAa,MAAM,sBAAsB,IAAI;AACtD,qBAAW,EAAE,SAAS;AAAA,QACxB;AACA,yBAAiB;AACjB;AAAA,MACF;AAAA,IACF;AAAA,EACF;AACA,aAAW,OAAO;AAClB,SAAO;AACT;AAEA,SAAS,sBAA8C;AACrD,MAAI,OAAO,YAAY,YAAa,QAAO,CAAC;AAC5C,QAAM,MAAM,QAAQ,IAAI;AACxB,MAAI,CAAC,IAAK,QAAO,CAAC;AAClB,QAAM,MAA8B,CAAC;AACrC,aAAW,QAAQ,IAAI,MAAM,GAAG,GAAG;AACjC,UAAM,KAAK,KAAK,QAAQ,GAAG;AAC3B,QAAI,KAAK,EAAG;AACZ,UAAM,MAAM,KAAK,MAAM,GAAG,EAAE,EAAE,KAAK;AACnC,UAAM,QAAQ,KAAK,MAAM,KAAK,CAAC,EAAE,KAAK;AACtC,QAAI,IAAK,KAAI,GAAG,IAAI;AAAA,EACtB;AACA,SAAO;AACT;AAEA,SAAS,aAAa,QAAoE;AACxF,SAAO,OAAO,QAAQ,MAAM,EAAE,IAAI,CAAC,CAAC,KAAK,KAAK,OAAO;AAAA,IACnD;AAAA,IACA,OACE,OAAO,UAAU,WACb,OAAO,UAAU,KAAK,IACpB,EAAE,UAAU,MAAM,SAAS,EAAE,IAC7B,EAAE,aAAa,MAAM,IACvB,OAAO,UAAU,YACf,EAAE,WAAW,MAAM,IACnB,EAAE,aAAa,MAAM;AAAA,EAC/B,EAAE;AACJ;AAEA,SAAS,OAAO,IAAoB;AAClC,UAAQ,OAAO,KAAK,MAAM,EAAE,CAAC,IAAI,UAAY,SAAS;AACxD;AAEA,SAAS,UAAU,IAAoB;AACrC,QAAM,UAAU,GAAG,QAAQ,MAAM,EAAE;AACnC,SAAO,QAAQ,MAAM,GAAG,EAAE,EAAE,OAAO,IAAI,GAAG;AAC5C;AAEA,SAAS,WAAW,IAAoB;AACtC,QAAM,UAAU,GAAG,QAAQ,MAAM,EAAE;AACnC,SAAO,QAAQ,MAAM,GAAG,EAAE,EAAE,OAAO,IAAI,GAAG;AAC5C;AAEA,SAAS,iBAAyB;AAChC,QAAM,QAAQ,IAAI,WAAW,CAAC;AAC9B,MAAI,OAAO,WAAW,QAAQ,oBAAoB,YAAY;AAC5D,eAAW,OAAO,gBAAgB,KAAK;AAAA,EACzC,OAAO;AACL,aAAS,IAAI,GAAG,IAAI,GAAG,IAAK,OAAM,CAAC,IAAI,KAAK,MAAM,KAAK,OAAO,IAAI,GAAG;AAAA,EACvE;AACA,SAAO,MAAM,KAAK,KAAK,EACpB,IAAI,CAAC,MAAM,EAAE,SAAS,EAAE,EAAE,SAAS,GAAG,GAAG,CAAC,EAC1C,KAAK,EAAE;AACZ;AAaO,IAAM,4BAA4B;AAuDzC,IAAM,4BAA4B;AAQlC,eAAsB,eACpB,QACA,QAC+B;AAC/B,MAAI,OAAO,WAAW,EAAG,QAAO,EAAE,IAAI,MAAM,QAAQ,GAAG,UAAU,GAAG,UAAU,CAAC,EAAE;AACjF,QAAM,SACJ,QAAQ,WAAW,OAAO,YAAY,cAAc,QAAQ,IAAI,iBAAiB;AACnF,MAAI,CAAC;AACH,UAAM,IAAI,MAAM,4EAA4E;AAC9F,QAAM,OACJ,QAAQ,SACP,OAAO,YAAY,cAAc,QAAQ,IAAI,oBAAoB,WAClE;AACF,QAAM,MAAM,GAAG,KAAK,QAAQ,QAAQ,EAAE,CAAC;AACvC,QAAM,MAAM,MAAM,MAAM,KAAK;AAAA,IAC3B,QAAQ;AAAA,IACR,SAAS;AAAA,MACP,gBAAgB;AAAA,MAChB,eAAe,UAAU,MAAM;AAAA,MAC/B,yBAAyB;AAAA,MACzB,GAAI,QAAQ,iBAAiB,EAAE,mBAAmB,OAAO,eAAe,IAAI,CAAC;AAAA,IAC/E;AAAA,IACA,MAAM,KAAK,UAAU,EAAE,aAAa,2BAA2B,OAAO,CAAC;AAAA,EACzE,CAAC;AACD,MAAI,SAAqF,CAAC;AAC1F,MAAI;AACF,aAAU,MAAM,IAAI,KAAK;AAAA,EAC3B,QAAQ;AAAA,EAER;AACA,SAAO;AAAA,IACL,IAAI,IAAI;AAAA,IACR,QAAQ,IAAI;AAAA,IACZ,UAAU,OAAO,aAAa,IAAI,KAAK,OAAO,SAAS;AAAA,IACvD,UAAU,OAAO,YAAY,CAAC;AAAA,EAChC;AACF;","names":[]}
1
+ {"version":3,"sources":["../src/otel-export.ts"],"sourcesContent":["/**\n * OTEL span exporter — streams LoopTraceEvents to an OTLP/HTTP collector.\n *\n * Reads OTEL_EXPORTER_OTLP_ENDPOINT + OTEL_EXPORTER_OTLP_HEADERS from env\n * when no explicit config is given. Keeps the runtime dep-free from\n * @opentelemetry/sdk-trace-base — minimal OTLP/JSON serializer.\n *\n * The exporter accepts both raw OtelSpan objects and LoopTraceEvents\n * (which get converted to OTLP spans automatically).\n */\n\nexport interface OtelExportConfig {\n /** OTLP endpoint. Reads OTEL_EXPORTER_OTLP_ENDPOINT env by default. */\n endpoint?: string\n /** OTLP headers. Reads OTEL_EXPORTER_OTLP_HEADERS env by default. */\n headers?: Record<string, string>\n /** Batch size before flush. Default 64. */\n batchSize?: number\n /** Flush interval ms. Default 5000. */\n flushIntervalMs?: number\n /** Resource attributes stamped on every export. */\n resourceAttributes?: Record<string, string | number | boolean>\n /** Service name. Default 'agent-runtime'. */\n serviceName?: string\n}\n\nexport interface OtelExporter {\n /** Export a span. */\n exportSpan(span: OtelSpan): void\n /** Force flush pending spans. */\n flush(): Promise<void>\n /** Shutdown cleanly. */\n shutdown(): Promise<void>\n}\n\nexport interface OtelSpan {\n traceId: string\n spanId: string\n parentSpanId?: string\n name: string\n kind?: number\n startTimeUnixNano: string\n endTimeUnixNano: string\n attributes?: OtelAttribute[]\n status?: { code: number; message?: string }\n}\n\nexport interface OtelAttribute {\n key: string\n value: { stringValue?: string; intValue?: string; doubleValue?: number; boolValue?: boolean }\n}\n\ninterface OtlpResourceSpans {\n resource: { attributes: OtelAttribute[] }\n scopeSpans: Array<{ scope: { name: string; version: string }; spans: OtelSpan[] }>\n}\n\ninterface OtlpExport {\n resourceSpans: OtlpResourceSpans[]\n}\n\nconst SCOPE = { name: '@tangle-network/agent-runtime', version: '0.83.0' }\n\n/**\n * Current (non-deprecated) OpenTelemetry GenAI semantic-convention keys.\n * Registry: https://opentelemetry.io/docs/specs/semconv/registry/attributes/gen-ai/\n * NB: `gen_ai.system` / `gen_ai.usage.prompt_tokens` / `completion_tokens` are\n * DEPRECATED — do not emit them. We use `provider.name` + `input/output_tokens`.\n */\nconst GEN_AI = {\n operation: 'gen_ai.operation.name',\n agentName: 'gen_ai.agent.name',\n conversationId: 'gen_ai.conversation.id',\n inputTokens: 'gen_ai.usage.input_tokens',\n outputTokens: 'gen_ai.usage.output_tokens',\n} as const\n\n/**\n * Create an OTEL exporter. Returns undefined when no endpoint is configured.\n */\nexport function createOtelExporter(config?: OtelExportConfig): OtelExporter | undefined {\n const resolvedEndpoint =\n config?.endpoint ??\n (typeof process !== 'undefined' ? process.env.OTEL_EXPORTER_OTLP_ENDPOINT : undefined)\n if (!resolvedEndpoint) return undefined\n const endpoint: string = resolvedEndpoint\n\n const headers = config?.headers ?? parseHeadersFromEnv()\n const batchSize = config?.batchSize ?? 64\n const flushIntervalMs = config?.flushIntervalMs ?? 5000\n const serviceName = config?.serviceName ?? 'agent-runtime'\n const resourceAttrs = config?.resourceAttributes ?? {}\n\n const pending: OtelSpan[] = []\n let timer: ReturnType<typeof setInterval> | undefined\n let stopped = false\n\n const exporter: OtelExporter = {\n exportSpan(span: OtelSpan): void {\n if (stopped) return\n pending.push(span)\n if (pending.length >= batchSize) {\n void doFlush()\n }\n },\n\n async flush(): Promise<void> {\n await doFlush()\n },\n\n async shutdown(): Promise<void> {\n stopped = true\n if (timer !== undefined) {\n clearInterval(timer)\n timer = undefined\n }\n await doFlush()\n },\n }\n\n timer = setInterval(() => {\n if (pending.length > 0) void doFlush()\n }, flushIntervalMs)\n if (typeof timer === 'object' && 'unref' in timer) {\n ;(timer as NodeJS.Timeout).unref()\n }\n\n async function doFlush(): Promise<void> {\n if (pending.length === 0) return\n const batch = pending.splice(0)\n const body: OtlpExport = {\n resourceSpans: [\n {\n resource: {\n attributes: toAttributes({\n 'service.name': serviceName,\n ...resourceAttrs,\n }),\n },\n scopeSpans: [{ scope: SCOPE, spans: batch }],\n },\n ],\n }\n const url = `${endpoint.replace(/\\/+$/, '')}/v1/traces`\n try {\n await fetch(url, {\n method: 'POST',\n headers: { 'content-type': 'application/json', ...headers },\n body: JSON.stringify(body),\n })\n } catch {\n // Best-effort — telemetry export must not crash the runtime.\n }\n }\n\n return exporter\n}\n\n/**\n * Convert a LoopTraceEvent into an OtelSpan for export.\n */\nexport function loopEventToOtelSpan(\n event: {\n kind: string\n runId: string\n timestamp: number\n payload: object\n },\n traceId: string,\n parentSpanId?: string,\n): OtelSpan {\n const spanId = generateSpanId()\n const attrs: Record<string, string | number | boolean> = {\n 'loop.event_kind': event.kind,\n 'loop.run_id': event.runId,\n }\n for (const [k, v] of Object.entries(event.payload)) {\n if (typeof v === 'string' || typeof v === 'number' || typeof v === 'boolean') {\n attrs[`loop.${k}`] = v\n }\n }\n const ts = msToNs(event.timestamp)\n return {\n traceId: padTraceId(traceId),\n spanId,\n parentSpanId: parentSpanId ? padSpanId(parentSpanId) : undefined,\n name: event.kind,\n kind: 1,\n startTimeUnixNano: ts,\n endTimeUnixNano: ts,\n attributes: toAttributes(attrs),\n status: { code: 1 },\n }\n}\n\n/**\n * Build a single flat OtelSpan whose attribute keys are emitted VERBATIM — no\n * `loop.` namespace. Use for non-loop spans (e.g. the intelligence per-run\n * span) whose keys ARE the downstream contract (`gen_ai.request.model`,\n * `tangle.sessionId`) and are exact-matched by readers. `loopEventToOtelSpan`\n * namespaces payload keys because loop payload fields are free-form; do not\n * reuse it for spans with contract keys.\n */\nexport function flatOtelSpan(\n name: string,\n attributes: Record<string, string | number | boolean>,\n traceId: string,\n timestampMs: number,\n parentSpanId?: string,\n): OtelSpan {\n const ts = msToNs(timestampMs)\n return {\n traceId: padTraceId(traceId),\n spanId: generateSpanId(),\n parentSpanId: parentSpanId ? padSpanId(parentSpanId) : undefined,\n name,\n kind: 1,\n startTimeUnixNano: ts,\n endTimeUnixNano: ts,\n attributes: toAttributes(attributes),\n status: { code: 1 },\n }\n}\n\n/**\n * Sink-neutral node in a reconstructed loop span tree. The root node's\n * `parentSpanId` is `undefined` — sinks decide how to parent it (the OTEL\n * mapper attaches the inherited delegation span; the delegation journal\n * leaves it as the tree root).\n */\nexport interface LoopSpanNode {\n spanId: string\n parentSpanId?: string\n /** `'loop'` | `'loop.round'` | `'loop.iteration'`. */\n name: string\n /** Topology level: loop root, plan round, or iteration branch. */\n kind: 'loop' | 'round' | 'branch'\n startMs: number\n endMs: number\n attrs: Record<string, string | number | boolean>\n /** True when the iteration carried an error — maps to OTEL status code 2. */\n error: boolean\n}\n\n/**\n * Build a nested, real-duration OTLP span tree for ONE loop run from its full\n * ordered `LoopTraceEvent` stream. Unlike `loopEventToOtelSpan` (one flat,\n * zero-duration span per event), this reconstructs the topology hierarchy a\n * GenAI trace viewer renders natively:\n *\n * loop (invoke_workflow)\n * └─ loop.round[k] (invoke_workflow) ← tangle.loop.move.{kind,width,rationale}\n * ├─ loop.iteration[i] (invoke_agent) ← gen_ai.agent.name + usage + verdict + placement\n * └─ …\n *\n * Attributes follow the current GenAI semconv (`gen_ai.*`) where they apply and\n * a namespaced `tangle.loop.*` / `tangle.cost.usd` extension for topology /\n * verdict / placement / cost (not yet standardized). Pure: feed it a buffered\n * per-runId event array (e.g. flushed on `loop.ended`) and export the result.\n */\nexport function buildLoopOtelSpans(\n events: ReadonlyArray<{ kind: string; runId: string; timestamp: number; payload: object }>,\n traceId: string,\n rootParentSpanId?: string,\n): OtelSpan[] {\n const tid = padTraceId(traceId)\n return buildLoopSpanNodes(events).map((node) => ({\n traceId: tid,\n spanId: node.spanId,\n parentSpanId: node.parentSpanId\n ? padSpanId(node.parentSpanId)\n : rootParentSpanId\n ? padSpanId(rootParentSpanId)\n : undefined,\n name: node.name,\n kind: 1,\n startTimeUnixNano: msToNs(node.startMs),\n endTimeUnixNano: msToNs(node.endMs),\n attributes: toAttributes(node.attrs),\n status: { code: node.error ? 2 : 1 },\n }))\n}\n\n/**\n * Sink-neutral core behind {@link buildLoopOtelSpans}: reconstruct the\n * loop → round → branch span tree from one run's ordered `LoopTraceEvent`\n * stream. Consumed by the OTEL mapper above and by the MCP delegation\n * journal's compact trace tee — one topology reconstruction, two sinks.\n * Tolerates partial streams (a run that never reached `loop.ended` closes\n * at the last observed event's timestamp).\n */\nexport function buildLoopSpanNodes(\n events: ReadonlyArray<{ kind: string; runId: string; timestamp: number; payload: object }>,\n): LoopSpanNode[] {\n if (events.length === 0) return []\n const out: LoopSpanNode[] = []\n const num = (v: unknown): number | undefined =>\n typeof v === 'number' && Number.isFinite(v) ? v : undefined\n const str = (v: unknown): string | undefined =>\n typeof v === 'string' && v.length > 0 ? v : undefined\n const rec = (v: unknown): Record<string, unknown> =>\n v && typeof v === 'object' ? (v as Record<string, unknown>) : {}\n\n const started = events.find((e) => e.kind === 'loop.started')\n const ended = events.find((e) => e.kind === 'loop.ended')\n const runId = events[0]?.runId ?? ''\n const rootStart = started?.timestamp ?? events[0]!.timestamp\n const rootEnd = ended?.timestamp ?? events[events.length - 1]!.timestamp\n const rootId = generateSpanId()\n\n const make = (\n spanId: string,\n parentSpanId: string | undefined,\n name: string,\n kind: LoopSpanNode['kind'],\n startMs: number,\n endMs: number,\n attrs: Record<string, string | number | boolean>,\n error = false,\n ): LoopSpanNode => ({\n spanId,\n parentSpanId,\n name,\n kind,\n startMs,\n endMs,\n attrs,\n error,\n })\n\n // root\n const sp = rec(started?.payload)\n const rootAttrs: Record<string, string | number | boolean> = {\n [GEN_AI.operation]: 'invoke_workflow',\n [GEN_AI.conversationId]: runId,\n // Explicit run identity under the tangle namespace so a consuming system of\n // record (Tangle Intelligence's run spine) maps this span tree to one run\n // deterministically, instead of inferring it from the conversation id.\n 'tangle.run.id': runId,\n 'tangle.loop.driver': str(sp.driver) ?? 'driver',\n }\n if (Array.isArray(sp.agentRunNames) && sp.agentRunNames.length > 0) {\n rootAttrs['tangle.loop.agents'] = sp.agentRunNames.map(String).join(',')\n // Subject grain: the primary agent/profile this run drove. Lets the spine\n // group runs under the thing being worked on rather than one service-wide\n // bucket. `agentRunNames[0]` is the lead agent on a fan-out.\n rootAttrs['tangle.subject.key'] = String(sp.agentRunNames[0])\n }\n if (ended) {\n const ep = rec(ended.payload)\n const win = num(ep.winnerIterationIndex)\n if (win !== undefined) rootAttrs['tangle.loop.winner.iteration_index'] = win\n const cost = num(ep.totalCostUsd)\n if (cost !== undefined) rootAttrs['tangle.cost.usd'] = cost\n const dur = num(ep.durationMs)\n if (dur !== undefined) rootAttrs['tangle.loop.duration_ms'] = dur\n const iters = num(ep.iterations)\n if (iters !== undefined) rootAttrs['tangle.loop.iterations'] = iters\n }\n out.push(make(rootId, undefined, 'loop', 'loop', rootStart, rootEnd, rootAttrs))\n\n // rounds + iterations\n const iterStartTs = new Map<number, number>()\n const placementByIdx = new Map<number, Record<string, string>>()\n let currentRoundId: string | undefined\n let pendingRound:\n | { id: string; start: number; attrs: Record<string, string | number | boolean> }\n | undefined\n const flushRound = (endMs: number) => {\n if (!pendingRound) return\n out.push(\n make(\n pendingRound.id,\n rootId,\n 'loop.round',\n 'round',\n pendingRound.start,\n endMs,\n pendingRound.attrs,\n ),\n )\n pendingRound = undefined\n }\n\n for (const e of events) {\n const p = rec(e.payload)\n switch (e.kind) {\n case 'loop.plan': {\n flushRound(e.timestamp)\n const id = generateSpanId()\n const roundIdx = num(p.roundIndex) ?? 0\n const attrs: Record<string, string | number | boolean> = {\n [GEN_AI.operation]: 'invoke_workflow',\n 'tangle.loop.round.index': roundIdx,\n 'tangle.loop.move.kind': str(p.moveKind) ?? 'unknown',\n 'tangle.loop.move.round': roundIdx,\n 'tangle.loop.move.width': num(p.plannedCount) ?? 0,\n }\n const r = str(p.rationale)\n if (r) attrs['tangle.loop.move.rationale'] = r\n const parent = num(p.parentIndex)\n if (parent !== undefined) attrs['tangle.loop.move.parent_index'] = parent\n if (Array.isArray(p.childIndices) && p.childIndices.length > 0) {\n attrs['tangle.loop.move.child_indices'] = p.childIndices.map(String).join(',')\n }\n pendingRound = { id, start: e.timestamp, attrs }\n currentRoundId = id\n break\n }\n case 'loop.iteration.started': {\n const idx = num(p.iterationIndex)\n if (idx !== undefined) iterStartTs.set(idx, e.timestamp)\n break\n }\n case 'loop.iteration.dispatch': {\n const idx = num(p.iterationIndex)\n if (idx === undefined) break\n const place: Record<string, string> = {}\n const kind = str(p.placement)\n if (kind) place['tangle.loop.placement.kind'] = kind\n const sid = str(p.sandboxId)\n if (sid) place['tangle.sandbox.id'] = sid\n const fid = str(p.fleetId)\n if (fid) place['tangle.fleet.id'] = fid\n const mid = str(p.machineId)\n if (mid) place['tangle.machine.id'] = mid\n placementByIdx.set(idx, place)\n break\n }\n case 'loop.iteration.ended': {\n const idx = num(p.iterationIndex) ?? 0\n const start = iterStartTs.get(idx) ?? e.timestamp\n const err = str(p.error)\n const attrs: Record<string, string | number | boolean> = {\n [GEN_AI.operation]: 'invoke_agent',\n 'tangle.loop.iteration.index': idx,\n }\n const agent = str(p.agentRunName)\n if (agent) attrs[GEN_AI.agentName] = agent\n const tu = rec(p.tokenUsage)\n const inTok = num(tu.input)\n if (inTok !== undefined) attrs[GEN_AI.inputTokens] = inTok\n const outTok = num(tu.output)\n if (outTok !== undefined) attrs[GEN_AI.outputTokens] = outTok\n const cost = num(p.costUsd)\n if (cost !== undefined) attrs['tangle.cost.usd'] = cost\n const verdict = rec(p.verdict)\n if (typeof verdict.valid === 'boolean') attrs['tangle.loop.verdict.valid'] = verdict.valid\n const score = num(verdict.score)\n if (score !== undefined) attrs['tangle.loop.verdict.score'] = score\n if (err) attrs['tangle.loop.error'] = err\n const gid = num(p.groupId)\n if (gid !== undefined) attrs['tangle.loop.iteration.group_id'] = gid\n const par = num(p.parentIndex)\n if (par !== undefined) attrs['tangle.loop.iteration.parent_index'] = par\n const dur = num(p.durationMs)\n if (dur !== undefined) attrs['tangle.loop.iteration.duration_ms'] = dur\n const preview = str(p.outputPreview)\n if (preview) attrs['tangle.loop.iteration.output_preview'] = preview\n Object.assign(attrs, placementByIdx.get(idx) ?? {})\n out.push(\n make(\n generateSpanId(),\n currentRoundId ?? rootId,\n 'loop.iteration',\n 'branch',\n start,\n e.timestamp,\n attrs,\n err !== undefined,\n ),\n )\n break\n }\n case 'loop.decision': {\n if (pendingRound) {\n const dec = str(p.decision)\n if (dec) pendingRound.attrs['tangle.loop.decision'] = dec\n flushRound(e.timestamp)\n }\n currentRoundId = undefined\n break\n }\n }\n }\n flushRound(rootEnd)\n return out\n}\n\nfunction parseHeadersFromEnv(): Record<string, string> {\n if (typeof process === 'undefined') return {}\n const raw = process.env.OTEL_EXPORTER_OTLP_HEADERS\n if (!raw) return {}\n const out: Record<string, string> = {}\n for (const pair of raw.split(',')) {\n const eq = pair.indexOf('=')\n if (eq < 0) continue\n const key = pair.slice(0, eq).trim()\n const value = pair.slice(eq + 1).trim()\n if (key) out[key] = value\n }\n return out\n}\n\nfunction toAttributes(record: Record<string, string | number | boolean>): OtelAttribute[] {\n return Object.entries(record).map(([key, value]) => ({\n key,\n value:\n typeof value === 'number'\n ? Number.isInteger(value)\n ? { intValue: value.toString() }\n : { doubleValue: value }\n : typeof value === 'boolean'\n ? { boolValue: value }\n : { stringValue: value },\n }))\n}\n\nfunction msToNs(ms: number): string {\n return (BigInt(Math.floor(ms)) * 1_000_000n).toString()\n}\n\nfunction padSpanId(id: string): string {\n const cleaned = id.replace(/-/g, '')\n return cleaned.slice(0, 16).padEnd(16, '0')\n}\n\nfunction padTraceId(id: string): string {\n const cleaned = id.replace(/-/g, '')\n return cleaned.slice(0, 32).padEnd(32, '0')\n}\n\nfunction generateSpanId(): string {\n const bytes = new Uint8Array(8)\n if (typeof globalThis.crypto?.getRandomValues === 'function') {\n globalThis.crypto.getRandomValues(bytes)\n } else {\n for (let i = 0; i < 8; i++) bytes[i] = Math.floor(Math.random() * 256)\n }\n return Array.from(bytes)\n .map((b) => b.toString(16).padStart(2, '0'))\n .join('')\n}\n\n// ─── Eval-run ingest (self-improvement provenance) ───────────────────────────\n//\n// Tangle Intelligence has a first-class, non-trace record for self-improvement\n// runs: POST /v1/ingest/eval-runs (\"Mode D\"). Each generation carries a\n// `surfaceHash` (the proposed-change identity) + arbitrary `surface` provenance;\n// a later `gate-decided` event re-emits the same `runId` (idempotent upsert) with\n// a real `gateDecision` + `holdoutLift`, so proposal→verdict is one diffable\n// record. This is how a consumer's RSI loop records WHAT it changed, WHY, from\n// which evidence — the audit trail behind agentic self-improvement.\n\n/** Wire version the eval-runs ingest enforces (X-Tangle-Wire-Version + body). */\nexport const INTELLIGENCE_WIRE_VERSION = '2026-05-26.v1'\n\nexport interface EvalRunGeneration {\n /** 0-based ordinal of this generation within the run (required by ingest). */\n index: number\n /** Identity of the proposed surface change (content-addressed hash). */\n surfaceHash: string\n /** Arbitrary provenance for this generation (rationale, evidence, source). */\n surface?: unknown\n /** Per-scenario results; empty until the generation is measured. */\n cells?: unknown[]\n /** Mean composite score (0 when unmeasured — pair with labels.measured). */\n compositeMean: number\n costUsd: number\n durationMs: number\n}\n\nexport interface EvalRunEvent {\n runId: string\n runDir: string\n /** ISO timestamp. */\n timestamp: string\n status:\n | 'started'\n | 'baseline-complete'\n | 'generation-complete'\n | 'gate-decided'\n | 'finished'\n | 'errored'\n labels?: Record<string, string>\n baseline?: EvalRunGeneration\n generations?: EvalRunGeneration[]\n gateDecision?: 'ship' | 'hold' | 'need_more_work' | 'model_ceiling' | 'arch_ceiling'\n holdoutLift?: number\n totalCostUsd: number\n totalDurationMs: number\n errorMessage?: string\n}\n\nexport interface EvalRunsExportConfig {\n /** Bearer key — tenant is resolved server-side from it. Reads TANGLE_API_KEY. */\n apiKey?: string\n /** Intelligence base. Reads TANGLE_INTELLIGENCE_URL env, else prod. */\n base?: string\n /** Idempotency-Key header (e.g. the runId) — safe retries + upsert. */\n idempotencyKey?: string\n}\n\nexport interface EvalRunsExportResult {\n ok: boolean\n status: number\n accepted: number\n rejected: Array<{ index: number; reason: string }>\n}\n\nconst DEFAULT_INTELLIGENCE_BASE = 'https://intelligence.tangle.tools'\n\n/**\n * Ship self-improvement eval-run events to Tangle Intelligence. Unlike the\n * best-effort span exporter, this RESOLVES with the ingest verdict (accepted /\n * rejected per event) so a consumer's loop can assert its provenance landed.\n * Throws only on a missing key or network failure.\n */\nexport async function exportEvalRuns(\n events: EvalRunEvent[],\n config?: EvalRunsExportConfig,\n): Promise<EvalRunsExportResult> {\n if (events.length === 0) return { ok: true, status: 0, accepted: 0, rejected: [] }\n const apiKey =\n config?.apiKey ?? (typeof process !== 'undefined' ? process.env.TANGLE_API_KEY : undefined)\n if (!apiKey)\n throw new Error('exportEvalRuns: apiKey required (pass config.apiKey or set TANGLE_API_KEY)')\n const base =\n config?.base ??\n (typeof process !== 'undefined' ? process.env.TANGLE_INTELLIGENCE_URL : undefined) ??\n DEFAULT_INTELLIGENCE_BASE\n const url = `${base.replace(/\\/+$/, '')}/v1/ingest/eval-runs`\n const res = await fetch(url, {\n method: 'POST',\n headers: {\n 'content-type': 'application/json',\n authorization: `Bearer ${apiKey}`,\n 'X-Tangle-Wire-Version': INTELLIGENCE_WIRE_VERSION,\n ...(config?.idempotencyKey ? { 'Idempotency-Key': config.idempotencyKey } : {}),\n },\n body: JSON.stringify({ wireVersion: INTELLIGENCE_WIRE_VERSION, events }),\n })\n let parsed: { accepted?: number; rejected?: Array<{ index: number; reason: string }> } = {}\n try {\n parsed = (await res.json()) as typeof parsed\n } catch {\n // non-JSON body (e.g. 5xx HTML) — leave parsed empty\n }\n return {\n ok: res.ok,\n status: res.status,\n accepted: parsed.accepted ?? (res.ok ? events.length : 0),\n rejected: parsed.rejected ?? [],\n }\n}\n"],"mappings":";AA6DA,IAAM,QAAQ,EAAE,MAAM,iCAAiC,SAAS,SAAS;AAQzE,IAAM,SAAS;AAAA,EACb,WAAW;AAAA,EACX,WAAW;AAAA,EACX,gBAAgB;AAAA,EAChB,aAAa;AAAA,EACb,cAAc;AAChB;AAKO,SAAS,mBAAmB,QAAqD;AACtF,QAAM,mBACJ,QAAQ,aACP,OAAO,YAAY,cAAc,QAAQ,IAAI,8BAA8B;AAC9E,MAAI,CAAC,iBAAkB,QAAO;AAC9B,QAAM,WAAmB;AAEzB,QAAM,UAAU,QAAQ,WAAW,oBAAoB;AACvD,QAAM,YAAY,QAAQ,aAAa;AACvC,QAAM,kBAAkB,QAAQ,mBAAmB;AACnD,QAAM,cAAc,QAAQ,eAAe;AAC3C,QAAM,gBAAgB,QAAQ,sBAAsB,CAAC;AAErD,QAAM,UAAsB,CAAC;AAC7B,MAAI;AACJ,MAAI,UAAU;AAEd,QAAM,WAAyB;AAAA,IAC7B,WAAW,MAAsB;AAC/B,UAAI,QAAS;AACb,cAAQ,KAAK,IAAI;AACjB,UAAI,QAAQ,UAAU,WAAW;AAC/B,aAAK,QAAQ;AAAA,MACf;AAAA,IACF;AAAA,IAEA,MAAM,QAAuB;AAC3B,YAAM,QAAQ;AAAA,IAChB;AAAA,IAEA,MAAM,WAA0B;AAC9B,gBAAU;AACV,UAAI,UAAU,QAAW;AACvB,sBAAc,KAAK;AACnB,gBAAQ;AAAA,MACV;AACA,YAAM,QAAQ;AAAA,IAChB;AAAA,EACF;AAEA,UAAQ,YAAY,MAAM;AACxB,QAAI,QAAQ,SAAS,EAAG,MAAK,QAAQ;AAAA,EACvC,GAAG,eAAe;AAClB,MAAI,OAAO,UAAU,YAAY,WAAW,OAAO;AACjD;AAAC,IAAC,MAAyB,MAAM;AAAA,EACnC;AAEA,iBAAe,UAAyB;AACtC,QAAI,QAAQ,WAAW,EAAG;AAC1B,UAAM,QAAQ,QAAQ,OAAO,CAAC;AAC9B,UAAM,OAAmB;AAAA,MACvB,eAAe;AAAA,QACb;AAAA,UACE,UAAU;AAAA,YACR,YAAY,aAAa;AAAA,cACvB,gBAAgB;AAAA,cAChB,GAAG;AAAA,YACL,CAAC;AAAA,UACH;AAAA,UACA,YAAY,CAAC,EAAE,OAAO,OAAO,OAAO,MAAM,CAAC;AAAA,QAC7C;AAAA,MACF;AAAA,IACF;AACA,UAAM,MAAM,GAAG,SAAS,QAAQ,QAAQ,EAAE,CAAC;AAC3C,QAAI;AACF,YAAM,MAAM,KAAK;AAAA,QACf,QAAQ;AAAA,QACR,SAAS,EAAE,gBAAgB,oBAAoB,GAAG,QAAQ;AAAA,QAC1D,MAAM,KAAK,UAAU,IAAI;AAAA,MAC3B,CAAC;AAAA,IACH,QAAQ;AAAA,IAER;AAAA,EACF;AAEA,SAAO;AACT;AAKO,SAAS,oBACd,OAMA,SACA,cACU;AACV,QAAM,SAAS,eAAe;AAC9B,QAAM,QAAmD;AAAA,IACvD,mBAAmB,MAAM;AAAA,IACzB,eAAe,MAAM;AAAA,EACvB;AACA,aAAW,CAAC,GAAG,CAAC,KAAK,OAAO,QAAQ,MAAM,OAAO,GAAG;AAClD,QAAI,OAAO,MAAM,YAAY,OAAO,MAAM,YAAY,OAAO,MAAM,WAAW;AAC5E,YAAM,QAAQ,CAAC,EAAE,IAAI;AAAA,IACvB;AAAA,EACF;AACA,QAAM,KAAK,OAAO,MAAM,SAAS;AACjC,SAAO;AAAA,IACL,SAAS,WAAW,OAAO;AAAA,IAC3B;AAAA,IACA,cAAc,eAAe,UAAU,YAAY,IAAI;AAAA,IACvD,MAAM,MAAM;AAAA,IACZ,MAAM;AAAA,IACN,mBAAmB;AAAA,IACnB,iBAAiB;AAAA,IACjB,YAAY,aAAa,KAAK;AAAA,IAC9B,QAAQ,EAAE,MAAM,EAAE;AAAA,EACpB;AACF;AAUO,SAAS,aACd,MACA,YACA,SACA,aACA,cACU;AACV,QAAM,KAAK,OAAO,WAAW;AAC7B,SAAO;AAAA,IACL,SAAS,WAAW,OAAO;AAAA,IAC3B,QAAQ,eAAe;AAAA,IACvB,cAAc,eAAe,UAAU,YAAY,IAAI;AAAA,IACvD;AAAA,IACA,MAAM;AAAA,IACN,mBAAmB;AAAA,IACnB,iBAAiB;AAAA,IACjB,YAAY,aAAa,UAAU;AAAA,IACnC,QAAQ,EAAE,MAAM,EAAE;AAAA,EACpB;AACF;AAsCO,SAAS,mBACd,QACA,SACA,kBACY;AACZ,QAAM,MAAM,WAAW,OAAO;AAC9B,SAAO,mBAAmB,MAAM,EAAE,IAAI,CAAC,UAAU;AAAA,IAC/C,SAAS;AAAA,IACT,QAAQ,KAAK;AAAA,IACb,cAAc,KAAK,eACf,UAAU,KAAK,YAAY,IAC3B,mBACE,UAAU,gBAAgB,IAC1B;AAAA,IACN,MAAM,KAAK;AAAA,IACX,MAAM;AAAA,IACN,mBAAmB,OAAO,KAAK,OAAO;AAAA,IACtC,iBAAiB,OAAO,KAAK,KAAK;AAAA,IAClC,YAAY,aAAa,KAAK,KAAK;AAAA,IACnC,QAAQ,EAAE,MAAM,KAAK,QAAQ,IAAI,EAAE;AAAA,EACrC,EAAE;AACJ;AAUO,SAAS,mBACd,QACgB;AAChB,MAAI,OAAO,WAAW,EAAG,QAAO,CAAC;AACjC,QAAM,MAAsB,CAAC;AAC7B,QAAM,MAAM,CAAC,MACX,OAAO,MAAM,YAAY,OAAO,SAAS,CAAC,IAAI,IAAI;AACpD,QAAM,MAAM,CAAC,MACX,OAAO,MAAM,YAAY,EAAE,SAAS,IAAI,IAAI;AAC9C,QAAM,MAAM,CAAC,MACX,KAAK,OAAO,MAAM,WAAY,IAAgC,CAAC;AAEjE,QAAM,UAAU,OAAO,KAAK,CAAC,MAAM,EAAE,SAAS,cAAc;AAC5D,QAAM,QAAQ,OAAO,KAAK,CAAC,MAAM,EAAE,SAAS,YAAY;AACxD,QAAM,QAAQ,OAAO,CAAC,GAAG,SAAS;AAClC,QAAM,YAAY,SAAS,aAAa,OAAO,CAAC,EAAG;AACnD,QAAM,UAAU,OAAO,aAAa,OAAO,OAAO,SAAS,CAAC,EAAG;AAC/D,QAAM,SAAS,eAAe;AAE9B,QAAM,OAAO,CACX,QACA,cACA,MACA,MACA,SACA,OACA,OACA,QAAQ,WACU;AAAA,IAClB;AAAA,IACA;AAAA,IACA;AAAA,IACA;AAAA,IACA;AAAA,IACA;AAAA,IACA;AAAA,IACA;AAAA,EACF;AAGA,QAAM,KAAK,IAAI,SAAS,OAAO;AAC/B,QAAM,YAAuD;AAAA,IAC3D,CAAC,OAAO,SAAS,GAAG;AAAA,IACpB,CAAC,OAAO,cAAc,GAAG;AAAA;AAAA;AAAA;AAAA,IAIzB,iBAAiB;AAAA,IACjB,sBAAsB,IAAI,GAAG,MAAM,KAAK;AAAA,EAC1C;AACA,MAAI,MAAM,QAAQ,GAAG,aAAa,KAAK,GAAG,cAAc,SAAS,GAAG;AAClE,cAAU,oBAAoB,IAAI,GAAG,cAAc,IAAI,MAAM,EAAE,KAAK,GAAG;AAIvE,cAAU,oBAAoB,IAAI,OAAO,GAAG,cAAc,CAAC,CAAC;AAAA,EAC9D;AACA,MAAI,OAAO;AACT,UAAM,KAAK,IAAI,MAAM,OAAO;AAC5B,UAAM,MAAM,IAAI,GAAG,oBAAoB;AACvC,QAAI,QAAQ,OAAW,WAAU,oCAAoC,IAAI;AACzE,UAAM,OAAO,IAAI,GAAG,YAAY;AAChC,QAAI,SAAS,OAAW,WAAU,iBAAiB,IAAI;AACvD,UAAM,MAAM,IAAI,GAAG,UAAU;AAC7B,QAAI,QAAQ,OAAW,WAAU,yBAAyB,IAAI;AAC9D,UAAM,QAAQ,IAAI,GAAG,UAAU;AAC/B,QAAI,UAAU,OAAW,WAAU,wBAAwB,IAAI;AAAA,EACjE;AACA,MAAI,KAAK,KAAK,QAAQ,QAAW,QAAQ,QAAQ,WAAW,SAAS,SAAS,CAAC;AAG/E,QAAM,cAAc,oBAAI,IAAoB;AAC5C,QAAM,iBAAiB,oBAAI,IAAoC;AAC/D,MAAI;AACJ,MAAI;AAGJ,QAAM,aAAa,CAAC,UAAkB;AACpC,QAAI,CAAC,aAAc;AACnB,QAAI;AAAA,MACF;AAAA,QACE,aAAa;AAAA,QACb;AAAA,QACA;AAAA,QACA;AAAA,QACA,aAAa;AAAA,QACb;AAAA,QACA,aAAa;AAAA,MACf;AAAA,IACF;AACA,mBAAe;AAAA,EACjB;AAEA,aAAW,KAAK,QAAQ;AACtB,UAAM,IAAI,IAAI,EAAE,OAAO;AACvB,YAAQ,EAAE,MAAM;AAAA,MACd,KAAK,aAAa;AAChB,mBAAW,EAAE,SAAS;AACtB,cAAM,KAAK,eAAe;AAC1B,cAAM,WAAW,IAAI,EAAE,UAAU,KAAK;AACtC,cAAM,QAAmD;AAAA,UACvD,CAAC,OAAO,SAAS,GAAG;AAAA,UACpB,2BAA2B;AAAA,UAC3B,yBAAyB,IAAI,EAAE,QAAQ,KAAK;AAAA,UAC5C,0BAA0B;AAAA,UAC1B,0BAA0B,IAAI,EAAE,YAAY,KAAK;AAAA,QACnD;AACA,cAAM,IAAI,IAAI,EAAE,SAAS;AACzB,YAAI,EAAG,OAAM,4BAA4B,IAAI;AAC7C,cAAM,SAAS,IAAI,EAAE,WAAW;AAChC,YAAI,WAAW,OAAW,OAAM,+BAA+B,IAAI;AACnE,YAAI,MAAM,QAAQ,EAAE,YAAY,KAAK,EAAE,aAAa,SAAS,GAAG;AAC9D,gBAAM,gCAAgC,IAAI,EAAE,aAAa,IAAI,MAAM,EAAE,KAAK,GAAG;AAAA,QAC/E;AACA,uBAAe,EAAE,IAAI,OAAO,EAAE,WAAW,MAAM;AAC/C,yBAAiB;AACjB;AAAA,MACF;AAAA,MACA,KAAK,0BAA0B;AAC7B,cAAM,MAAM,IAAI,EAAE,cAAc;AAChC,YAAI,QAAQ,OAAW,aAAY,IAAI,KAAK,EAAE,SAAS;AACvD;AAAA,MACF;AAAA,MACA,KAAK,2BAA2B;AAC9B,cAAM,MAAM,IAAI,EAAE,cAAc;AAChC,YAAI,QAAQ,OAAW;AACvB,cAAM,QAAgC,CAAC;AACvC,cAAM,OAAO,IAAI,EAAE,SAAS;AAC5B,YAAI,KAAM,OAAM,4BAA4B,IAAI;AAChD,cAAM,MAAM,IAAI,EAAE,SAAS;AAC3B,YAAI,IAAK,OAAM,mBAAmB,IAAI;AACtC,cAAM,MAAM,IAAI,EAAE,OAAO;AACzB,YAAI,IAAK,OAAM,iBAAiB,IAAI;AACpC,cAAM,MAAM,IAAI,EAAE,SAAS;AAC3B,YAAI,IAAK,OAAM,mBAAmB,IAAI;AACtC,uBAAe,IAAI,KAAK,KAAK;AAC7B;AAAA,MACF;AAAA,MACA,KAAK,wBAAwB;AAC3B,cAAM,MAAM,IAAI,EAAE,cAAc,KAAK;AACrC,cAAM,QAAQ,YAAY,IAAI,GAAG,KAAK,EAAE;AACxC,cAAM,MAAM,IAAI,EAAE,KAAK;AACvB,cAAM,QAAmD;AAAA,UACvD,CAAC,OAAO,SAAS,GAAG;AAAA,UACpB,+BAA+B;AAAA,QACjC;AACA,cAAM,QAAQ,IAAI,EAAE,YAAY;AAChC,YAAI,MAAO,OAAM,OAAO,SAAS,IAAI;AACrC,cAAM,KAAK,IAAI,EAAE,UAAU;AAC3B,cAAM,QAAQ,IAAI,GAAG,KAAK;AAC1B,YAAI,UAAU,OAAW,OAAM,OAAO,WAAW,IAAI;AACrD,cAAM,SAAS,IAAI,GAAG,MAAM;AAC5B,YAAI,WAAW,OAAW,OAAM,OAAO,YAAY,IAAI;AACvD,cAAM,OAAO,IAAI,EAAE,OAAO;AAC1B,YAAI,SAAS,OAAW,OAAM,iBAAiB,IAAI;AACnD,cAAM,UAAU,IAAI,EAAE,OAAO;AAC7B,YAAI,OAAO,QAAQ,UAAU,UAAW,OAAM,2BAA2B,IAAI,QAAQ;AACrF,cAAM,QAAQ,IAAI,QAAQ,KAAK;AAC/B,YAAI,UAAU,OAAW,OAAM,2BAA2B,IAAI;AAC9D,YAAI,IAAK,OAAM,mBAAmB,IAAI;AACtC,cAAM,MAAM,IAAI,EAAE,OAAO;AACzB,YAAI,QAAQ,OAAW,OAAM,gCAAgC,IAAI;AACjE,cAAM,MAAM,IAAI,EAAE,WAAW;AAC7B,YAAI,QAAQ,OAAW,OAAM,oCAAoC,IAAI;AACrE,cAAM,MAAM,IAAI,EAAE,UAAU;AAC5B,YAAI,QAAQ,OAAW,OAAM,mCAAmC,IAAI;AACpE,cAAM,UAAU,IAAI,EAAE,aAAa;AACnC,YAAI,QAAS,OAAM,sCAAsC,IAAI;AAC7D,eAAO,OAAO,OAAO,eAAe,IAAI,GAAG,KAAK,CAAC,CAAC;AAClD,YAAI;AAAA,UACF;AAAA,YACE,eAAe;AAAA,YACf,kBAAkB;AAAA,YAClB;AAAA,YACA;AAAA,YACA;AAAA,YACA,EAAE;AAAA,YACF;AAAA,YACA,QAAQ;AAAA,UACV;AAAA,QACF;AACA;AAAA,MACF;AAAA,MACA,KAAK,iBAAiB;AACpB,YAAI,cAAc;AAChB,gBAAM,MAAM,IAAI,EAAE,QAAQ;AAC1B,cAAI,IAAK,cAAa,MAAM,sBAAsB,IAAI;AACtD,qBAAW,EAAE,SAAS;AAAA,QACxB;AACA,yBAAiB;AACjB;AAAA,MACF;AAAA,IACF;AAAA,EACF;AACA,aAAW,OAAO;AAClB,SAAO;AACT;AAEA,SAAS,sBAA8C;AACrD,MAAI,OAAO,YAAY,YAAa,QAAO,CAAC;AAC5C,QAAM,MAAM,QAAQ,IAAI;AACxB,MAAI,CAAC,IAAK,QAAO,CAAC;AAClB,QAAM,MAA8B,CAAC;AACrC,aAAW,QAAQ,IAAI,MAAM,GAAG,GAAG;AACjC,UAAM,KAAK,KAAK,QAAQ,GAAG;AAC3B,QAAI,KAAK,EAAG;AACZ,UAAM,MAAM,KAAK,MAAM,GAAG,EAAE,EAAE,KAAK;AACnC,UAAM,QAAQ,KAAK,MAAM,KAAK,CAAC,EAAE,KAAK;AACtC,QAAI,IAAK,KAAI,GAAG,IAAI;AAAA,EACtB;AACA,SAAO;AACT;AAEA,SAAS,aAAa,QAAoE;AACxF,SAAO,OAAO,QAAQ,MAAM,EAAE,IAAI,CAAC,CAAC,KAAK,KAAK,OAAO;AAAA,IACnD;AAAA,IACA,OACE,OAAO,UAAU,WACb,OAAO,UAAU,KAAK,IACpB,EAAE,UAAU,MAAM,SAAS,EAAE,IAC7B,EAAE,aAAa,MAAM,IACvB,OAAO,UAAU,YACf,EAAE,WAAW,MAAM,IACnB,EAAE,aAAa,MAAM;AAAA,EAC/B,EAAE;AACJ;AAEA,SAAS,OAAO,IAAoB;AAClC,UAAQ,OAAO,KAAK,MAAM,EAAE,CAAC,IAAI,UAAY,SAAS;AACxD;AAEA,SAAS,UAAU,IAAoB;AACrC,QAAM,UAAU,GAAG,QAAQ,MAAM,EAAE;AACnC,SAAO,QAAQ,MAAM,GAAG,EAAE,EAAE,OAAO,IAAI,GAAG;AAC5C;AAEA,SAAS,WAAW,IAAoB;AACtC,QAAM,UAAU,GAAG,QAAQ,MAAM,EAAE;AACnC,SAAO,QAAQ,MAAM,GAAG,EAAE,EAAE,OAAO,IAAI,GAAG;AAC5C;AAEA,SAAS,iBAAyB;AAChC,QAAM,QAAQ,IAAI,WAAW,CAAC;AAC9B,MAAI,OAAO,WAAW,QAAQ,oBAAoB,YAAY;AAC5D,eAAW,OAAO,gBAAgB,KAAK;AAAA,EACzC,OAAO;AACL,aAAS,IAAI,GAAG,IAAI,GAAG,IAAK,OAAM,CAAC,IAAI,KAAK,MAAM,KAAK,OAAO,IAAI,GAAG;AAAA,EACvE;AACA,SAAO,MAAM,KAAK,KAAK,EACpB,IAAI,CAAC,MAAM,EAAE,SAAS,EAAE,EAAE,SAAS,GAAG,GAAG,CAAC,EAC1C,KAAK,EAAE;AACZ;AAaO,IAAM,4BAA4B;AAuDzC,IAAM,4BAA4B;AAQlC,eAAsB,eACpB,QACA,QAC+B;AAC/B,MAAI,OAAO,WAAW,EAAG,QAAO,EAAE,IAAI,MAAM,QAAQ,GAAG,UAAU,GAAG,UAAU,CAAC,EAAE;AACjF,QAAM,SACJ,QAAQ,WAAW,OAAO,YAAY,cAAc,QAAQ,IAAI,iBAAiB;AACnF,MAAI,CAAC;AACH,UAAM,IAAI,MAAM,4EAA4E;AAC9F,QAAM,OACJ,QAAQ,SACP,OAAO,YAAY,cAAc,QAAQ,IAAI,0BAA0B,WACxE;AACF,QAAM,MAAM,GAAG,KAAK,QAAQ,QAAQ,EAAE,CAAC;AACvC,QAAM,MAAM,MAAM,MAAM,KAAK;AAAA,IAC3B,QAAQ;AAAA,IACR,SAAS;AAAA,MACP,gBAAgB;AAAA,MAChB,eAAe,UAAU,MAAM;AAAA,MAC/B,yBAAyB;AAAA,MACzB,GAAI,QAAQ,iBAAiB,EAAE,mBAAmB,OAAO,eAAe,IAAI,CAAC;AAAA,IAC/E;AAAA,IACA,MAAM,KAAK,UAAU,EAAE,aAAa,2BAA2B,OAAO,CAAC;AAAA,EACzE,CAAC;AACD,MAAI,SAAqF,CAAC;AAC1F,MAAI;AACF,aAAU,MAAM,IAAI,KAAK;AAAA,EAC3B,QAAQ;AAAA,EAER;AACA,SAAO;AAAA,IACL,IAAI,IAAI;AAAA,IACR,QAAQ,IAAI;AAAA,IACZ,UAAU,OAAO,aAAa,IAAI,KAAK,OAAO,SAAS;AAAA,IACvD,UAAU,OAAO,YAAY,CAAC;AAAA,EAChC;AACF;","names":[]}
@@ -1,9 +1,14 @@
1
1
  import {
2
2
  runLocalHarness
3
- } from "./chunk-7LO5GMAO.js";
3
+ } from "./chunk-CMYMTRGA.js";
4
4
 
5
5
  // src/improvement/agentic-generator.ts
6
6
  import { spawnSync } from "child_process";
7
+ import { existsSync, readFileSync } from "fs";
8
+ import { join } from "path";
9
+ var RAW_TRACE_ANALYST_ID = "raw-trace-distiller";
10
+ var RAW_TRACE_AREA = "raw-trace-context";
11
+ var RAW_TRACE_DIAGNOSIS_PATH = ".improve/raw-trace-diagnosis.md";
7
12
  function agenticGenerator(opts = {}) {
8
13
  const harness = opts.harness ?? "claude";
9
14
  const buildPrompt = opts.buildPrompt ?? defaultBuildPrompt;
@@ -12,8 +17,15 @@ function agenticGenerator(opts = {}) {
12
17
  const verify = opts.verify;
13
18
  return {
14
19
  kind: `agentic:${harness}`,
20
+ // The seed repo + (in rawTraceContext mode) the raw-trace filesystem context
21
+ // are the change signal — an agentic coder proposes from them even when the
22
+ // distiller yielded zero findings. Without this, the improvementDriver's
23
+ // empty-findings guard short-circuits and generates ZERO candidates on the
24
+ // first (and, for a single-generation run, only) proposal round.
25
+ proposesWithoutFindings: true,
15
26
  async generate({ worktreePath, report, findings, maxShots, signal }) {
16
27
  const basePrompt = buildPrompt({ report, findings });
28
+ const needsRawTraceEvidence = requiresRawTraceEvidence(findings);
17
29
  const shots = Math.max(1, maxShots);
18
30
  let attemptNote = "";
19
31
  for (let shot = 0; shot < shots; shot++) {
@@ -24,6 +36,10 @@ function agenticGenerator(opts = {}) {
24
36
  taskPrompt: attemptNote ? `${basePrompt}
25
37
 
26
38
  ${attemptNote}` : basePrompt,
39
+ // The candidate worktree is isolated and must be editable without an
40
+ // interactive permission prompt. Other runLocalHarness callers remain
41
+ // permission-safe by default.
42
+ dangerouslySkipPermissions: harness === "claude",
27
43
  timeoutMs: opts.timeoutMs,
28
44
  signal
29
45
  });
@@ -31,6 +47,13 @@ ${attemptNote}` : basePrompt,
31
47
  attemptNote = EMPTY_TREE_NOTE;
32
48
  continue;
33
49
  }
50
+ if (needsRawTraceEvidence) {
51
+ const problem = rawTraceEvidenceProblem(worktreePath, findings);
52
+ if (problem) {
53
+ attemptNote = problem;
54
+ continue;
55
+ }
56
+ }
34
57
  if (!verify) {
35
58
  return { applied: true, summary: summarize(findings) };
36
59
  }
@@ -57,6 +80,16 @@ function defaultBuildPrompt(args) {
57
80
  lines.push(`- (${f.severity})${where} ${f.claim}`);
58
81
  if (f.recommended_action) lines.push(` \u2192 ${f.recommended_action}`);
59
82
  }
83
+ if (requiresRawTraceEvidence(args.findings)) {
84
+ lines.push(
85
+ "",
86
+ "Raw trace evidence requirement:",
87
+ `- Inspect at least one raw trace path named above before editing.`,
88
+ `- Write ${RAW_TRACE_DIAGNOSIS_PATH} in this worktree.`,
89
+ "- Include the exact trace path(s) inspected, the failure mechanism, and the code change made.",
90
+ "- A candidate without this file, or with only this file changed, is discarded."
91
+ );
92
+ }
60
93
  return lines.join("\n");
61
94
  }
62
95
  var EMPTY_TREE_NOTE = "NOTE: your previous attempt left the working tree unchanged. Make the concrete file edits now.";
@@ -69,6 +102,51 @@ function failureNote(feedback) {
69
102
  ${truncate(detail, 4e3)}` : "No verifier detail was captured."
70
103
  ].join("\n");
71
104
  }
105
+ function rawTraceEvidenceProblem(worktreePath, findings) {
106
+ const changedPaths = worktreeChangedPaths(worktreePath);
107
+ const substantive = changedPaths.filter((path) => path !== RAW_TRACE_DIAGNOSIS_PATH);
108
+ if (substantive.length === 0) {
109
+ return [
110
+ `NOTE: raw-trace mode requires a real code/config edit in addition to ${RAW_TRACE_DIAGNOSIS_PATH}.`,
111
+ "Your previous attempt only changed the diagnosis artifact. Inspect the cited traces and make the causal code change."
112
+ ].join("\n");
113
+ }
114
+ const diagnosisPath = join(worktreePath, RAW_TRACE_DIAGNOSIS_PATH);
115
+ if (!existsSync(diagnosisPath)) {
116
+ return [
117
+ `NOTE: raw-trace mode requires ${RAW_TRACE_DIAGNOSIS_PATH}.`,
118
+ "Before retrying, inspect at least one cited spans.jsonl/cached-result.json/artifact path, then write the diagnosis file with the exact path, failure mechanism, and code change."
119
+ ].join("\n");
120
+ }
121
+ const body = readFileSync(diagnosisPath, "utf8");
122
+ const evidencePaths = traceEvidencePaths(findings);
123
+ if (evidencePaths.length > 0 && !evidencePaths.some((path) => body.includes(path))) {
124
+ return [
125
+ `${RAW_TRACE_DIAGNOSIS_PATH} exists, but it does not cite any exact raw trace path from the findings.`,
126
+ `Cite at least one of these inspected paths exactly: ${evidencePaths.slice(0, 5).join(", ")}`
127
+ ].join("\n");
128
+ }
129
+ return null;
130
+ }
131
+ function requiresRawTraceEvidence(findings) {
132
+ return findings.some((finding) => {
133
+ const f = finding;
134
+ return f.analyst_id === RAW_TRACE_ANALYST_ID || f.area === RAW_TRACE_AREA;
135
+ });
136
+ }
137
+ function traceEvidencePaths(findings) {
138
+ const out = [];
139
+ for (const finding of findings) {
140
+ const refs = finding.evidence_refs;
141
+ if (!Array.isArray(refs)) continue;
142
+ for (const ref of refs) {
143
+ if (!ref || typeof ref !== "object") continue;
144
+ const uri = ref.uri;
145
+ if (typeof uri === "string" && uri.length > 0) out.push(uri);
146
+ }
147
+ }
148
+ return [...new Set(out)];
149
+ }
72
150
  function commandVerifier(command, args = [], timeoutMs = 3e5) {
73
151
  return (worktreePath) => {
74
152
  const result = spawnSync(command, args, {
@@ -105,7 +183,10 @@ function truncate(s, n) {
105
183
  return s.length <= n ? s : `${s.slice(0, n - 1)}\u2026`;
106
184
  }
107
185
  function worktreeDirty(worktreePath) {
108
- const result = spawnSync("git", ["status", "--porcelain"], {
186
+ return worktreeChangedPaths(worktreePath).length > 0;
187
+ }
188
+ function worktreeChangedPaths(worktreePath) {
189
+ const result = spawnSync("git", ["status", "--porcelain", "--untracked-files=all"], {
109
190
  cwd: worktreePath,
110
191
  encoding: "utf-8"
111
192
  });
@@ -119,7 +200,7 @@ function worktreeDirty(worktreePath) {
119
200
  `agenticGenerator: git status exited ${result.status} in ${worktreePath}: ${result.stderr.trim()}`
120
201
  );
121
202
  }
122
- return result.stdout.trim().length > 0;
203
+ return result.stdout.split("\n").map((line) => line.trim()).filter((line) => line.length > 0).map((line) => line.slice(3).trim());
123
204
  }
124
205
 
125
206
  // src/improvement/build-prompts.ts
@@ -263,4 +344,4 @@ export {
263
344
  mcpBuildPrompt,
264
345
  mcpServeVerifier
265
346
  };
266
- //# sourceMappingURL=chunk-VNOOH22O.js.map
347
+ //# sourceMappingURL=chunk-OOL3675H.js.map
@@ -0,0 +1 @@
1
+ {"version":3,"sources":["../src/improvement/agentic-generator.ts","../src/improvement/build-prompts.ts","../src/improvement/mcp-serve-verifier.ts"],"sourcesContent":["/**\n *\n * `agenticGenerator` — the full-agentic `CandidateGenerator`: the\n * `shots=N, sandbox=on` setting of the one `improvementDriver`. It runs a real\n * coding harness (claude / codex / opencode) inside the candidate worktree the\n * driver already created, letting the agent read the codebase + the research\n * report and make the change in place. The driver then commits the worktree\n * into a `CodeSurface`.\n *\n * Mechanism: identical to the proven Phase-2.8 in-process executor — spawn the\n * harness as a subprocess with `cwd` = the worktree, on the same filesystem,\n * so edits land in place (no sandbox-mount round-trip). `runLocalHarness` is\n * the verified primitive. The OUTER sandbox is the improvement loop's own\n * execution context; the generator does not nest a second sandbox per\n * candidate (which would reintroduce a host↔sandbox worktree-transport\n * problem that does not need solving here).\n *\n * `maxShots` is the DEPTH dial — a multi-shot verify-in-session loop, NOT the\n * kernel `runLoop`. Each shot runs one full harness session in the (persistent)\n * worktree; between shots the loop refines based on what the last shot produced:\n * - empty tree → \"you changed nothing, make the edits\" → retry\n * - dirty + `verify` fails → feed the verifier's failure into the next shot\n * (the worktree persists, so the harness RESUMES atop its own failing\n * edits with the error in hand — no `--resume` session plumbing needed,\n * and harness-agnostic across claude/codex/opencode)\n * - dirty + `verify` ok (or no verifier configured) → return the candidate\n * A candidate that never verifies within `maxShots` is discarded (`applied:\n * false`), never shipped — if you configured a verifier, a non-passing tree is\n * not a candidate. With no verifier the legacy behavior holds: first dirty shot\n * is the candidate.\n *\n * @experimental\n */\n\nimport { spawnSync } from 'node:child_process'\nimport { existsSync, readFileSync } from 'node:fs'\nimport { join } from 'node:path'\nimport type { AnalystFinding } from '@tangle-network/agent-eval'\nimport { type LocalHarness, runLocalHarness } from '../mcp/local-harness'\nimport type { CandidateGenerator } from './improvement-driver'\n\nconst RAW_TRACE_ANALYST_ID = 'raw-trace-distiller'\nconst RAW_TRACE_AREA = 'raw-trace-context'\nconst RAW_TRACE_DIAGNOSIS_PATH = '.improve/raw-trace-diagnosis.md'\n\n/** Outcome of verifying a candidate worktree. `feedback` (compiler errors,\n * failing test output) is fed into the next shot when `ok` is false. */\nexport interface VerifyResult {\n ok: boolean\n feedback?: string\n}\n\n/** Verifies the edited worktree. Sync or async; throws only on a setup fault\n * (a candidate that fails verification returns `{ok:false}`, it does not\n * throw). */\nexport type Verifier = (worktreePath: string) => Promise<VerifyResult> | VerifyResult\n\nexport interface AgenticGeneratorOptions {\n /** Local coding harness to run in the worktree. Default `claude`. */\n harness?: LocalHarness\n /** Per-shot wall-clock timeout (ms). Default = `runLocalHarness` default (5m). */\n timeoutMs?: number\n /** Build the harness task prompt from the report + findings. Override for\n * domain phrasing; the default turns findings into a concrete coder task. */\n buildPrompt?: (args: { report: unknown; findings: AnalystFinding[] }) => string\n /** Verify the worktree after each dirtying shot. When set, a candidate that\n * fails verification is NOT returned — the failure feeds the next shot\n * (verify-in-session), up to `maxShots`; a candidate that never verifies is\n * discarded (`applied:false`), never shipped. Omitted ⇒ legacy behavior:\n * the first dirty shot is the candidate. See `commandVerifier`. */\n verify?: Verifier\n /** Test seam — inject the harness runner (defaults to `runLocalHarness`). */\n runHarness?: typeof runLocalHarness\n /** Test seam — inject the worktree-dirty check (defaults to `git status`). */\n isDirty?: (worktreePath: string) => boolean\n}\n\n/** Full-agentic `CandidateGenerator` (the `shots=N, sandbox=on` setting): run a real coding harness inside the candidate worktree so the agent makes the change in place. */\nexport function agenticGenerator(opts: AgenticGeneratorOptions = {}): CandidateGenerator {\n const harness = opts.harness ?? 'claude'\n const buildPrompt = opts.buildPrompt ?? defaultBuildPrompt\n const run = opts.runHarness ?? runLocalHarness\n const dirty = opts.isDirty ?? worktreeDirty\n const verify = opts.verify\n\n return {\n kind: `agentic:${harness}`,\n // The seed repo + (in rawTraceContext mode) the raw-trace filesystem context\n // are the change signal — an agentic coder proposes from them even when the\n // distiller yielded zero findings. Without this, the improvementDriver's\n // empty-findings guard short-circuits and generates ZERO candidates on the\n // first (and, for a single-generation run, only) proposal round.\n proposesWithoutFindings: true,\n async generate({ worktreePath, report, findings, maxShots, signal }) {\n const basePrompt = buildPrompt({ report, findings })\n const needsRawTraceEvidence = requiresRawTraceEvidence(findings)\n const shots = Math.max(1, maxShots)\n // Feedback appended to the base prompt for the NEXT shot — empty on shot 0.\n let attemptNote = ''\n\n for (let shot = 0; shot < shots; shot++) {\n if (signal.aborted) break\n await run({\n harness,\n cwd: worktreePath,\n taskPrompt: attemptNote ? `${basePrompt}\\n\\n${attemptNote}` : basePrompt,\n // The candidate worktree is isolated and must be editable without an\n // interactive permission prompt. Other runLocalHarness callers remain\n // permission-safe by default.\n dangerouslySkipPermissions: harness === 'claude',\n timeoutMs: opts.timeoutMs,\n signal,\n })\n\n // The worktree IS the signal: no edits ⇒ tell the next shot to act.\n if (!dirty(worktreePath)) {\n attemptNote = EMPTY_TREE_NOTE\n continue\n }\n\n if (needsRawTraceEvidence) {\n const problem = rawTraceEvidenceProblem(worktreePath, findings)\n if (problem) {\n attemptNote = problem\n continue\n }\n }\n\n // Dirty: with no verifier the diff IS the candidate (we trust the diff,\n // not the harness's stdout). With a verifier the candidate must pass it.\n if (!verify) {\n return { applied: true, summary: summarize(findings) }\n }\n const result = await verify(worktreePath)\n if (result.ok) {\n return { applied: true, summary: summarize(findings) }\n }\n // Dirty but failing — resume next shot atop these edits with the error.\n attemptNote = failureNote(result.feedback)\n }\n\n // Shots exhausted: no verified candidate (or, sans verifier, no edits).\n return { applied: false, summary: '' }\n },\n }\n}\n\n/** Turn the analyst's findings (+ optional report) into a concrete coder task. */\nfunction defaultBuildPrompt(args: { report: unknown; findings: AnalystFinding[] }): string {\n const lines: string[] = [\n 'You are improving this codebase based on an evaluation analysis.',\n 'Make the smallest set of edits that addresses the findings below, then stop.',\n 'Do not change unrelated code. Do not commit — leave changes in the working tree.',\n '',\n 'Findings:',\n ]\n for (const f of args.findings) {\n const where = f.subject ? ` [${f.subject}]` : ''\n lines.push(`- (${f.severity})${where} ${f.claim}`)\n if (f.recommended_action) lines.push(` → ${f.recommended_action}`)\n }\n if (requiresRawTraceEvidence(args.findings)) {\n lines.push(\n '',\n 'Raw trace evidence requirement:',\n `- Inspect at least one raw trace path named above before editing.`,\n `- Write ${RAW_TRACE_DIAGNOSIS_PATH} in this worktree.`,\n '- Include the exact trace path(s) inspected, the failure mechanism, and the code change made.',\n '- A candidate without this file, or with only this file changed, is discarded.',\n )\n }\n return lines.join('\\n')\n}\n\nconst EMPTY_TREE_NOTE =\n 'NOTE: your previous attempt left the working tree unchanged. Make the concrete file edits now.'\n\n/** Next-shot feedback when the worktree is dirty but failed verification. The\n * edits persist on disk, so the harness resumes atop them — tell it to fix in\n * place, not start over. Verifier detail is truncated to keep the prompt bounded. */\nfunction failureNote(feedback?: string): string {\n const detail = feedback?.trim()\n return [\n 'NOTE: your edits are in the working tree but verification FAILED.',\n 'Fix the problem in place — build on your existing edits, do not revert them.',\n detail ? `Verifier output:\\n${truncate(detail, 4000)}` : 'No verifier detail was captured.',\n ].join('\\n')\n}\n\nfunction rawTraceEvidenceProblem(worktreePath: string, findings: AnalystFinding[]): string | null {\n const changedPaths = worktreeChangedPaths(worktreePath)\n const substantive = changedPaths.filter((path) => path !== RAW_TRACE_DIAGNOSIS_PATH)\n if (substantive.length === 0) {\n return [\n `NOTE: raw-trace mode requires a real code/config edit in addition to ${RAW_TRACE_DIAGNOSIS_PATH}.`,\n 'Your previous attempt only changed the diagnosis artifact. Inspect the cited traces and make the causal code change.',\n ].join('\\n')\n }\n\n const diagnosisPath = join(worktreePath, RAW_TRACE_DIAGNOSIS_PATH)\n if (!existsSync(diagnosisPath)) {\n return [\n `NOTE: raw-trace mode requires ${RAW_TRACE_DIAGNOSIS_PATH}.`,\n 'Before retrying, inspect at least one cited spans.jsonl/cached-result.json/artifact path, then write the diagnosis file with the exact path, failure mechanism, and code change.',\n ].join('\\n')\n }\n\n const body = readFileSync(diagnosisPath, 'utf8')\n const evidencePaths = traceEvidencePaths(findings)\n if (evidencePaths.length > 0 && !evidencePaths.some((path) => body.includes(path))) {\n return [\n `${RAW_TRACE_DIAGNOSIS_PATH} exists, but it does not cite any exact raw trace path from the findings.`,\n `Cite at least one of these inspected paths exactly: ${evidencePaths.slice(0, 5).join(', ')}`,\n ].join('\\n')\n }\n\n return null\n}\n\nfunction requiresRawTraceEvidence(findings: AnalystFinding[]): boolean {\n return findings.some((finding) => {\n const f = finding as unknown as Record<string, unknown>\n return f.analyst_id === RAW_TRACE_ANALYST_ID || f.area === RAW_TRACE_AREA\n })\n}\n\nfunction traceEvidencePaths(findings: AnalystFinding[]): string[] {\n const out: string[] = []\n for (const finding of findings) {\n const refs = (finding as unknown as { evidence_refs?: unknown }).evidence_refs\n if (!Array.isArray(refs)) continue\n for (const ref of refs) {\n if (!ref || typeof ref !== 'object') continue\n const uri = (ref as { uri?: unknown }).uri\n if (typeof uri === 'string' && uri.length > 0) out.push(uri)\n }\n }\n return [...new Set(out)]\n}\n\n/** A `Verifier` that runs a command in the worktree: exit 0 ⇒ ok, any other\n * exit ⇒ failed with stdout+stderr as feedback. The common case — verify by\n * `tsc --noEmit`, `pnpm build`, or a test command. A timeout is treated as a\n * FAILED candidate (a change that hangs the build is a bad change); a missing\n * binary or spawn fault throws (a setup bug, not a failed candidate — no\n * silent fallback). */\nexport function commandVerifier(\n command: string,\n args: string[] = [],\n timeoutMs = 300_000,\n): Verifier {\n return (worktreePath: string): VerifyResult => {\n const result = spawnSync(command, args, {\n cwd: worktreePath,\n encoding: 'utf-8',\n timeout: timeoutMs,\n })\n if (result.signal) {\n return {\n ok: false,\n feedback: `verifier '${command}' killed by ${result.signal} (likely timeout after ${timeoutMs}ms)`,\n }\n }\n if (result.error) {\n const code = (result.error as NodeJS.ErrnoException).code\n if (code === 'ENOENT') {\n throw new Error(\n `commandVerifier: '${command}' not found in PATH (setup bug, not a failed candidate)`,\n )\n }\n throw new Error(`commandVerifier: '${command}' failed to spawn: ${result.error.message}`)\n }\n if (result.status === 0) return { ok: true }\n const out = `${result.stdout ?? ''}${result.stderr ?? ''}`.trim()\n return { ok: false, feedback: out.length > 0 ? out : `exit ${result.status}` }\n }\n}\n\n/** A one-line summary for the commit message, derived from the findings. */\nfunction summarize(findings: AnalystFinding[]): string {\n if (findings.length === 0) return 'agentic improvement'\n if (findings.length === 1) return `agentic: ${truncate(findings[0]!.claim, 64)}`\n return `agentic: ${findings.length} findings addressed`\n}\n\nfunction truncate(s: string, n: number): string {\n return s.length <= n ? s : `${s.slice(0, n - 1)}…`\n}\n\n/** Non-empty `git status --porcelain` ⇒ the harness changed the worktree.\n * Fails loud: the worktree is a fresh checkout, so a git error here means\n * something is genuinely broken (git missing, corrupt index, killed mid-run).\n * Folding that into `false` would silently discard a candidate and mask the\n * real failure — forbidden by the no-silent-fallbacks doctrine. */\nfunction worktreeDirty(worktreePath: string): boolean {\n return worktreeChangedPaths(worktreePath).length > 0\n}\n\nfunction worktreeChangedPaths(worktreePath: string): string[] {\n const result = spawnSync('git', ['status', '--porcelain', '--untracked-files=all'], {\n cwd: worktreePath,\n encoding: 'utf-8',\n })\n if (result.error) {\n throw new Error(\n `agenticGenerator: git status failed to spawn in ${worktreePath}: ${result.error.message}`,\n )\n }\n if (result.status !== 0) {\n throw new Error(\n `agenticGenerator: git status exited ${result.status} in ${worktreePath}: ${result.stderr.trim()}`,\n )\n }\n return result.stdout\n .split('\\n')\n .map((line) => line.trim())\n .filter((line) => line.length > 0)\n .map((line) => line.slice(3).trim())\n}\n","/**\n * Build-prompt starting points for the two buildable artifact types. There is\n * NO `toolGenerator`/`mcpGenerator` wrapper — the factory is `agenticGenerator`\n * + a verifier (docs/artifact-lifecycle-frontier.md), so a tool or an MCP\n * server is built by composing the pieces directly:\n *\n * // a tool:\n * agenticGenerator({ buildPrompt: toolBuildPrompt, verify: commandVerifier('pnpm', ['test']) })\n * // an MCP server:\n * agenticGenerator({ buildPrompt: mcpBuildPrompt, verify: mcpServeVerifier({ command: 'node', args: ['server.mjs'] }) })\n *\n * These are the only type-specific bit (the phrasing that points the agent at a\n * tool vs. an MCP); the worktree, resume-on-failure loop, and improvement-loop\n * wrapper are shared. MCP is the load-bearing target — it is how a harness\n * acquires tools; raw tools matter where we control the loader.\n */\n\nimport type { AnalystFinding } from '@tangle-network/agent-eval'\n\ntype FindingsArg = { report: unknown; findings: AnalystFinding[] }\n\nfunction findingLines(findings: AnalystFinding[]): string[] {\n return findings.map((f) => {\n const where = f.subject ? ` [${f.subject}]` : ''\n const action = f.recommended_action ? ` → ${f.recommended_action}` : ''\n return `- (${f.severity})${where} ${f.claim}${action}`\n })\n}\n\n/** Build the starting instruction for a coder agent tasked with implementing a new tool. */\nexport function toolBuildPrompt(args: FindingsArg): string {\n return [\n 'You are building a new TOOL for this codebase to address the gaps below.',\n 'Write the tool as a small, self-contained module PLUS tests that exercise it.',\n 'The tool must compile and its tests must pass — they will be run automatically;',\n 'if verification fails you will get the error and another attempt. Do not commit;',\n 'leave the changes in the working tree.',\n '',\n 'Gaps the tool should close:',\n ...findingLines(args.findings),\n ].join('\\n')\n}\n\n/** Build the starting instruction for a coder agent tasked with implementing a new MCP server. */\nexport function mcpBuildPrompt(args: FindingsArg): string {\n return [\n 'You are building a new MCP SERVER (Model Context Protocol) that exposes',\n 'tool(s) addressing the gaps below, so any harness can mount it.',\n 'Requirements that WILL be checked by booting the server:',\n '- it starts over stdio and answers the MCP `initialize` handshake,',\n '- `tools/list` returns at least one tool with a valid input schema.',\n 'Newline-delimited JSON-RPC 2.0, protocol version 2024-11-05. Include a start',\n 'command (e.g. a package.json `start` script or a clear entrypoint). If the',\n 'boot-and-probe fails you will get the error and another attempt. Do not',\n 'commit; leave the changes in the working tree.',\n '',\n 'Capabilities the server should provide:',\n ...findingLines(args.findings),\n ].join('\\n')\n}\n","/**\n * `mcpServeVerifier` — the intrinsic verifier for a built MCP server: the\n * boot-and-probe checker named in docs/artifact-lifecycle-frontier.md. A\n * generated MCP server is only a candidate if it actually *serves* — so this\n * boots it over stdio (the default local MCP transport) and runs the real\n * handshake: `initialize` → `notifications/initialized` → `tools/list`, and\n * asserts the server answers with at least `minTools` tools.\n *\n * Outcomes follow the `Verifier` contract: a server that fails to start, exits\n * early, errors the handshake, times out, or exposes no tools is a FAILED\n * candidate (`{ok:false}`, fed back into the next generation shot); a missing\n * start binary or spawn fault THROWS (a setup bug, never a silent fallback).\n *\n * Protocol matches the runtime's own stdio MCP server (src/mcp/server.ts):\n * newline-delimited JSON-RPC 2.0, protocol version 2024-11-05.\n */\n\nimport { spawn } from 'node:child_process'\nimport { createInterface } from 'node:readline'\nimport type { Verifier, VerifyResult } from './agentic-generator'\n\nconst PROTOCOL_VERSION = '2024-11-05'\n\nexport interface McpServeSpec {\n /** Command that starts the built MCP server in the worktree (stdio transport). */\n command: string\n args?: string[]\n /** Extra env for the server process (merged over `process.env`). */\n env?: Record<string, string>\n /** Handshake timeout (ms). Default 30s. */\n timeoutMs?: number\n /** Minimum tools the server must expose to pass. Default 1. */\n minTools?: number\n}\n\ninterface JsonRpcResponse {\n jsonrpc?: string\n id?: number | string | null\n result?: unknown\n error?: { code: number; message: string }\n}\n\n/** Build a `Verifier` that boots a generated MCP server over stdio and checks it exposes tools. */\nexport function mcpServeVerifier(spec: McpServeSpec): Verifier {\n const timeoutMs = spec.timeoutMs ?? 30_000\n const minTools = spec.minTools ?? 1\n\n return (worktreePath: string): Promise<VerifyResult> =>\n new Promise<VerifyResult>((resolve, reject) => {\n const child = spawn(spec.command, spec.args ?? [], {\n cwd: worktreePath,\n stdio: ['pipe', 'pipe', 'pipe'],\n env: { ...process.env, ...spec.env },\n })\n\n const stderr: string[] = []\n let settled = false\n let nextId = 1\n const initId = nextId++\n let listId = -1\n\n const settle = (fn: () => void) => {\n if (settled) return\n settled = true\n clearTimeout(timer)\n rl.close()\n child.kill('SIGKILL')\n fn()\n }\n const withStderr = (msg: string) =>\n stderr.length > 0 ? `${msg}\\nstderr:\\n${stderr.join('').slice(-2000)}` : msg\n const pass = () => settle(() => resolve({ ok: true }))\n const failCandidate = (msg: string) =>\n settle(() => resolve({ ok: false, feedback: withStderr(msg) }))\n const setupFault = (err: Error) => settle(() => reject(err))\n\n const send = (msg: Record<string, unknown>): boolean => {\n try {\n child.stdin.write(`${JSON.stringify(msg)}\\n`)\n return true\n } catch (err) {\n // EPIPE: the server died mid-handshake — a failed candidate, not a fault.\n failCandidate(`writing to MCP server stdin failed: ${(err as Error).message}`)\n return false\n }\n }\n\n child.on('error', (err) => {\n const code = (err as NodeJS.ErrnoException).code\n setupFault(\n code === 'ENOENT'\n ? new Error(\n `mcpServeVerifier: '${spec.command}' not found in PATH (setup bug, not a failed candidate)`,\n )\n : new Error(`mcpServeVerifier: '${spec.command}' failed to spawn: ${err.message}`),\n )\n })\n child.on('exit', (code, signal) => {\n // An exit before the handshake completes is a failed candidate (the\n // server crashed on boot); after we settle, our own SIGKILL fires here.\n failCandidate(`MCP server exited (code ${code}, signal ${signal}) before serving`)\n })\n child.stderr.on('data', (d) => stderr.push(String(d)))\n\n const rl = createInterface({ input: child.stdout })\n rl.on('line', (line) => {\n let msg: JsonRpcResponse | undefined\n try {\n msg = JSON.parse(line) as JsonRpcResponse\n } catch {\n return // servers log to stdout too; skip non-JSON lines\n }\n if (!msg || typeof msg !== 'object') return\n\n if (msg.id === initId) {\n if (msg.error) return failCandidate(`initialize errored: ${JSON.stringify(msg.error)}`)\n if (!send({ jsonrpc: '2.0', method: 'notifications/initialized' })) return\n listId = nextId++\n send({ jsonrpc: '2.0', id: listId, method: 'tools/list' })\n return\n }\n if (msg.id === listId) {\n if (msg.error) return failCandidate(`tools/list errored: ${JSON.stringify(msg.error)}`)\n const tools = (msg.result as { tools?: unknown[] } | undefined)?.tools\n if (!Array.isArray(tools)) return failCandidate('tools/list result has no tools array')\n if (tools.length < minTools) {\n return failCandidate(`tools/list returned ${tools.length} tool(s), need >= ${minTools}`)\n }\n return pass()\n }\n })\n\n const timer = setTimeout(\n () => failCandidate(`MCP server did not complete the handshake within ${timeoutMs}ms`),\n timeoutMs,\n )\n\n send({\n jsonrpc: '2.0',\n id: initId,\n method: 'initialize',\n params: {\n protocolVersion: PROTOCOL_VERSION,\n capabilities: {},\n clientInfo: { name: 'agent-runtime-mcp-verify', version: '0' },\n },\n })\n })\n}\n"],"mappings":";;;;;AAkCA,SAAS,iBAAiB;AAC1B,SAAS,YAAY,oBAAoB;AACzC,SAAS,YAAY;AAKrB,IAAM,uBAAuB;AAC7B,IAAM,iBAAiB;AACvB,IAAM,2BAA2B;AAmC1B,SAAS,iBAAiB,OAAgC,CAAC,GAAuB;AACvF,QAAM,UAAU,KAAK,WAAW;AAChC,QAAM,cAAc,KAAK,eAAe;AACxC,QAAM,MAAM,KAAK,cAAc;AAC/B,QAAM,QAAQ,KAAK,WAAW;AAC9B,QAAM,SAAS,KAAK;AAEpB,SAAO;AAAA,IACL,MAAM,WAAW,OAAO;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA,IAMxB,yBAAyB;AAAA,IACzB,MAAM,SAAS,EAAE,cAAc,QAAQ,UAAU,UAAU,OAAO,GAAG;AACnE,YAAM,aAAa,YAAY,EAAE,QAAQ,SAAS,CAAC;AACnD,YAAM,wBAAwB,yBAAyB,QAAQ;AAC/D,YAAM,QAAQ,KAAK,IAAI,GAAG,QAAQ;AAElC,UAAI,cAAc;AAElB,eAAS,OAAO,GAAG,OAAO,OAAO,QAAQ;AACvC,YAAI,OAAO,QAAS;AACpB,cAAM,IAAI;AAAA,UACR;AAAA,UACA,KAAK;AAAA,UACL,YAAY,cAAc,GAAG,UAAU;AAAA;AAAA,EAAO,WAAW,KAAK;AAAA;AAAA;AAAA;AAAA,UAI9D,4BAA4B,YAAY;AAAA,UACxC,WAAW,KAAK;AAAA,UAChB;AAAA,QACF,CAAC;AAGD,YAAI,CAAC,MAAM,YAAY,GAAG;AACxB,wBAAc;AACd;AAAA,QACF;AAEA,YAAI,uBAAuB;AACzB,gBAAM,UAAU,wBAAwB,cAAc,QAAQ;AAC9D,cAAI,SAAS;AACX,0BAAc;AACd;AAAA,UACF;AAAA,QACF;AAIA,YAAI,CAAC,QAAQ;AACX,iBAAO,EAAE,SAAS,MAAM,SAAS,UAAU,QAAQ,EAAE;AAAA,QACvD;AACA,cAAM,SAAS,MAAM,OAAO,YAAY;AACxC,YAAI,OAAO,IAAI;AACb,iBAAO,EAAE,SAAS,MAAM,SAAS,UAAU,QAAQ,EAAE;AAAA,QACvD;AAEA,sBAAc,YAAY,OAAO,QAAQ;AAAA,MAC3C;AAGA,aAAO,EAAE,SAAS,OAAO,SAAS,GAAG;AAAA,IACvC;AAAA,EACF;AACF;AAGA,SAAS,mBAAmB,MAA+D;AACzF,QAAM,QAAkB;AAAA,IACtB;AAAA,IACA;AAAA,IACA;AAAA,IACA;AAAA,IACA;AAAA,EACF;AACA,aAAW,KAAK,KAAK,UAAU;AAC7B,UAAM,QAAQ,EAAE,UAAU,KAAK,EAAE,OAAO,MAAM;AAC9C,UAAM,KAAK,MAAM,EAAE,QAAQ,IAAI,KAAK,IAAI,EAAE,KAAK,EAAE;AACjD,QAAI,EAAE,mBAAoB,OAAM,KAAK,cAAS,EAAE,kBAAkB,EAAE;AAAA,EACtE;AACA,MAAI,yBAAyB,KAAK,QAAQ,GAAG;AAC3C,UAAM;AAAA,MACJ;AAAA,MACA;AAAA,MACA;AAAA,MACA,WAAW,wBAAwB;AAAA,MACnC;AAAA,MACA;AAAA,IACF;AAAA,EACF;AACA,SAAO,MAAM,KAAK,IAAI;AACxB;AAEA,IAAM,kBACJ;AAKF,SAAS,YAAY,UAA2B;AAC9C,QAAM,SAAS,UAAU,KAAK;AAC9B,SAAO;AAAA,IACL;AAAA,IACA;AAAA,IACA,SAAS;AAAA,EAAqB,SAAS,QAAQ,GAAI,CAAC,KAAK;AAAA,EAC3D,EAAE,KAAK,IAAI;AACb;AAEA,SAAS,wBAAwB,cAAsB,UAA2C;AAChG,QAAM,eAAe,qBAAqB,YAAY;AACtD,QAAM,cAAc,aAAa,OAAO,CAAC,SAAS,SAAS,wBAAwB;AACnF,MAAI,YAAY,WAAW,GAAG;AAC5B,WAAO;AAAA,MACL,wEAAwE,wBAAwB;AAAA,MAChG;AAAA,IACF,EAAE,KAAK,IAAI;AAAA,EACb;AAEA,QAAM,gBAAgB,KAAK,cAAc,wBAAwB;AACjE,MAAI,CAAC,WAAW,aAAa,GAAG;AAC9B,WAAO;AAAA,MACL,iCAAiC,wBAAwB;AAAA,MACzD;AAAA,IACF,EAAE,KAAK,IAAI;AAAA,EACb;AAEA,QAAM,OAAO,aAAa,eAAe,MAAM;AAC/C,QAAM,gBAAgB,mBAAmB,QAAQ;AACjD,MAAI,cAAc,SAAS,KAAK,CAAC,cAAc,KAAK,CAAC,SAAS,KAAK,SAAS,IAAI,CAAC,GAAG;AAClF,WAAO;AAAA,MACL,GAAG,wBAAwB;AAAA,MAC3B,uDAAuD,cAAc,MAAM,GAAG,CAAC,EAAE,KAAK,IAAI,CAAC;AAAA,IAC7F,EAAE,KAAK,IAAI;AAAA,EACb;AAEA,SAAO;AACT;AAEA,SAAS,yBAAyB,UAAqC;AACrE,SAAO,SAAS,KAAK,CAAC,YAAY;AAChC,UAAM,IAAI;AACV,WAAO,EAAE,eAAe,wBAAwB,EAAE,SAAS;AAAA,EAC7D,CAAC;AACH;AAEA,SAAS,mBAAmB,UAAsC;AAChE,QAAM,MAAgB,CAAC;AACvB,aAAW,WAAW,UAAU;AAC9B,UAAM,OAAQ,QAAmD;AACjE,QAAI,CAAC,MAAM,QAAQ,IAAI,EAAG;AAC1B,eAAW,OAAO,MAAM;AACtB,UAAI,CAAC,OAAO,OAAO,QAAQ,SAAU;AACrC,YAAM,MAAO,IAA0B;AACvC,UAAI,OAAO,QAAQ,YAAY,IAAI,SAAS,EAAG,KAAI,KAAK,GAAG;AAAA,IAC7D;AAAA,EACF;AACA,SAAO,CAAC,GAAG,IAAI,IAAI,GAAG,CAAC;AACzB;AAQO,SAAS,gBACd,SACA,OAAiB,CAAC,GAClB,YAAY,KACF;AACV,SAAO,CAAC,iBAAuC;AAC7C,UAAM,SAAS,UAAU,SAAS,MAAM;AAAA,MACtC,KAAK;AAAA,MACL,UAAU;AAAA,MACV,SAAS;AAAA,IACX,CAAC;AACD,QAAI,OAAO,QAAQ;AACjB,aAAO;AAAA,QACL,IAAI;AAAA,QACJ,UAAU,aAAa,OAAO,eAAe,OAAO,MAAM,0BAA0B,SAAS;AAAA,MAC/F;AAAA,IACF;AACA,QAAI,OAAO,OAAO;AAChB,YAAM,OAAQ,OAAO,MAAgC;AACrD,UAAI,SAAS,UAAU;AACrB,cAAM,IAAI;AAAA,UACR,qBAAqB,OAAO;AAAA,QAC9B;AAAA,MACF;AACA,YAAM,IAAI,MAAM,qBAAqB,OAAO,sBAAsB,OAAO,MAAM,OAAO,EAAE;AAAA,IAC1F;AACA,QAAI,OAAO,WAAW,EAAG,QAAO,EAAE,IAAI,KAAK;AAC3C,UAAM,MAAM,GAAG,OAAO,UAAU,EAAE,GAAG,OAAO,UAAU,EAAE,GAAG,KAAK;AAChE,WAAO,EAAE,IAAI,OAAO,UAAU,IAAI,SAAS,IAAI,MAAM,QAAQ,OAAO,MAAM,GAAG;AAAA,EAC/E;AACF;AAGA,SAAS,UAAU,UAAoC;AACrD,MAAI,SAAS,WAAW,EAAG,QAAO;AAClC,MAAI,SAAS,WAAW,EAAG,QAAO,YAAY,SAAS,SAAS,CAAC,EAAG,OAAO,EAAE,CAAC;AAC9E,SAAO,YAAY,SAAS,MAAM;AACpC;AAEA,SAAS,SAAS,GAAW,GAAmB;AAC9C,SAAO,EAAE,UAAU,IAAI,IAAI,GAAG,EAAE,MAAM,GAAG,IAAI,CAAC,CAAC;AACjD;AAOA,SAAS,cAAc,cAA+B;AACpD,SAAO,qBAAqB,YAAY,EAAE,SAAS;AACrD;AAEA,SAAS,qBAAqB,cAAgC;AAC5D,QAAM,SAAS,UAAU,OAAO,CAAC,UAAU,eAAe,uBAAuB,GAAG;AAAA,IAClF,KAAK;AAAA,IACL,UAAU;AAAA,EACZ,CAAC;AACD,MAAI,OAAO,OAAO;AAChB,UAAM,IAAI;AAAA,MACR,mDAAmD,YAAY,KAAK,OAAO,MAAM,OAAO;AAAA,IAC1F;AAAA,EACF;AACA,MAAI,OAAO,WAAW,GAAG;AACvB,UAAM,IAAI;AAAA,MACR,uCAAuC,OAAO,MAAM,OAAO,YAAY,KAAK,OAAO,OAAO,KAAK,CAAC;AAAA,IAClG;AAAA,EACF;AACA,SAAO,OAAO,OACX,MAAM,IAAI,EACV,IAAI,CAAC,SAAS,KAAK,KAAK,CAAC,EACzB,OAAO,CAAC,SAAS,KAAK,SAAS,CAAC,EAChC,IAAI,CAAC,SAAS,KAAK,MAAM,CAAC,EAAE,KAAK,CAAC;AACvC;;;ACzSA,SAAS,aAAa,UAAsC;AAC1D,SAAO,SAAS,IAAI,CAAC,MAAM;AACzB,UAAM,QAAQ,EAAE,UAAU,KAAK,EAAE,OAAO,MAAM;AAC9C,UAAM,SAAS,EAAE,qBAAqB,WAAM,EAAE,kBAAkB,KAAK;AACrE,WAAO,MAAM,EAAE,QAAQ,IAAI,KAAK,IAAI,EAAE,KAAK,GAAG,MAAM;AAAA,EACtD,CAAC;AACH;AAGO,SAAS,gBAAgB,MAA2B;AACzD,SAAO;AAAA,IACL;AAAA,IACA;AAAA,IACA;AAAA,IACA;AAAA,IACA;AAAA,IACA;AAAA,IACA;AAAA,IACA,GAAG,aAAa,KAAK,QAAQ;AAAA,EAC/B,EAAE,KAAK,IAAI;AACb;AAGO,SAAS,eAAe,MAA2B;AACxD,SAAO;AAAA,IACL;AAAA,IACA;AAAA,IACA;AAAA,IACA;AAAA,IACA;AAAA,IACA;AAAA,IACA;AAAA,IACA;AAAA,IACA;AAAA,IACA;AAAA,IACA;AAAA,IACA,GAAG,aAAa,KAAK,QAAQ;AAAA,EAC/B,EAAE,KAAK,IAAI;AACb;;;AC1CA,SAAS,aAAa;AACtB,SAAS,uBAAuB;AAGhC,IAAM,mBAAmB;AAsBlB,SAAS,iBAAiB,MAA8B;AAC7D,QAAM,YAAY,KAAK,aAAa;AACpC,QAAM,WAAW,KAAK,YAAY;AAElC,SAAO,CAAC,iBACN,IAAI,QAAsB,CAAC,SAAS,WAAW;AAC7C,UAAM,QAAQ,MAAM,KAAK,SAAS,KAAK,QAAQ,CAAC,GAAG;AAAA,MACjD,KAAK;AAAA,MACL,OAAO,CAAC,QAAQ,QAAQ,MAAM;AAAA,MAC9B,KAAK,EAAE,GAAG,QAAQ,KAAK,GAAG,KAAK,IAAI;AAAA,IACrC,CAAC;AAED,UAAM,SAAmB,CAAC;AAC1B,QAAI,UAAU;AACd,QAAI,SAAS;AACb,UAAM,SAAS;AACf,QAAI,SAAS;AAEb,UAAM,SAAS,CAAC,OAAmB;AACjC,UAAI,QAAS;AACb,gBAAU;AACV,mBAAa,KAAK;AAClB,SAAG,MAAM;AACT,YAAM,KAAK,SAAS;AACpB,SAAG;AAAA,IACL;AACA,UAAM,aAAa,CAAC,QAClB,OAAO,SAAS,IAAI,GAAG,GAAG;AAAA;AAAA,EAAc,OAAO,KAAK,EAAE,EAAE,MAAM,IAAK,CAAC,KAAK;AAC3E,UAAM,OAAO,MAAM,OAAO,MAAM,QAAQ,EAAE,IAAI,KAAK,CAAC,CAAC;AACrD,UAAM,gBAAgB,CAAC,QACrB,OAAO,MAAM,QAAQ,EAAE,IAAI,OAAO,UAAU,WAAW,GAAG,EAAE,CAAC,CAAC;AAChE,UAAM,aAAa,CAAC,QAAe,OAAO,MAAM,OAAO,GAAG,CAAC;AAE3D,UAAM,OAAO,CAAC,QAA0C;AACtD,UAAI;AACF,cAAM,MAAM,MAAM,GAAG,KAAK,UAAU,GAAG,CAAC;AAAA,CAAI;AAC5C,eAAO;AAAA,MACT,SAAS,KAAK;AAEZ,sBAAc,uCAAwC,IAAc,OAAO,EAAE;AAC7E,eAAO;AAAA,MACT;AAAA,IACF;AAEA,UAAM,GAAG,SAAS,CAAC,QAAQ;AACzB,YAAM,OAAQ,IAA8B;AAC5C;AAAA,QACE,SAAS,WACL,IAAI;AAAA,UACF,sBAAsB,KAAK,OAAO;AAAA,QACpC,IACA,IAAI,MAAM,sBAAsB,KAAK,OAAO,sBAAsB,IAAI,OAAO,EAAE;AAAA,MACrF;AAAA,IACF,CAAC;AACD,UAAM,GAAG,QAAQ,CAAC,MAAM,WAAW;AAGjC,oBAAc,2BAA2B,IAAI,YAAY,MAAM,kBAAkB;AAAA,IACnF,CAAC;AACD,UAAM,OAAO,GAAG,QAAQ,CAAC,MAAM,OAAO,KAAK,OAAO,CAAC,CAAC,CAAC;AAErD,UAAM,KAAK,gBAAgB,EAAE,OAAO,MAAM,OAAO,CAAC;AAClD,OAAG,GAAG,QAAQ,CAAC,SAAS;AACtB,UAAI;AACJ,UAAI;AACF,cAAM,KAAK,MAAM,IAAI;AAAA,MACvB,QAAQ;AACN;AAAA,MACF;AACA,UAAI,CAAC,OAAO,OAAO,QAAQ,SAAU;AAErC,UAAI,IAAI,OAAO,QAAQ;AACrB,YAAI,IAAI,MAAO,QAAO,cAAc,uBAAuB,KAAK,UAAU,IAAI,KAAK,CAAC,EAAE;AACtF,YAAI,CAAC,KAAK,EAAE,SAAS,OAAO,QAAQ,4BAA4B,CAAC,EAAG;AACpE,iBAAS;AACT,aAAK,EAAE,SAAS,OAAO,IAAI,QAAQ,QAAQ,aAAa,CAAC;AACzD;AAAA,MACF;AACA,UAAI,IAAI,OAAO,QAAQ;AACrB,YAAI,IAAI,MAAO,QAAO,cAAc,uBAAuB,KAAK,UAAU,IAAI,KAAK,CAAC,EAAE;AACtF,cAAM,QAAS,IAAI,QAA8C;AACjE,YAAI,CAAC,MAAM,QAAQ,KAAK,EAAG,QAAO,cAAc,sCAAsC;AACtF,YAAI,MAAM,SAAS,UAAU;AAC3B,iBAAO,cAAc,uBAAuB,MAAM,MAAM,qBAAqB,QAAQ,EAAE;AAAA,QACzF;AACA,eAAO,KAAK;AAAA,MACd;AAAA,IACF,CAAC;AAED,UAAM,QAAQ;AAAA,MACZ,MAAM,cAAc,oDAAoD,SAAS,IAAI;AAAA,MACrF;AAAA,IACF;AAEA,SAAK;AAAA,MACH,SAAS;AAAA,MACT,IAAI;AAAA,MACJ,QAAQ;AAAA,MACR,QAAQ;AAAA,QACN,iBAAiB;AAAA,QACjB,cAAc,CAAC;AAAA,QACf,YAAY,EAAE,MAAM,4BAA4B,SAAS,IAAI;AAAA,MAC/D;AAAA,IACF,CAAC;AAAA,EACH,CAAC;AACL;","names":[]}
@@ -8,7 +8,7 @@ import {
8
8
  DELEGATION_STATUS_DESCRIPTION,
9
9
  DELEGATION_STATUS_INPUT_SCHEMA,
10
10
  DELEGATION_STATUS_TOOL_NAME
11
- } from "./chunk-N7EJV7N3.js";
11
+ } from "./chunk-RYBVU4M3.js";
12
12
 
13
13
  // src/mcp/openai-tools.ts
14
14
  function buildTool(name, description, parameters) {
@@ -45,4 +45,4 @@ export {
45
45
  mcpToolsForRuntimeMcp,
46
46
  mcpToolsForRuntimeMcpSubset
47
47
  };
48
- //# sourceMappingURL=chunk-5AVV7KAH.js.map
48
+ //# sourceMappingURL=chunk-QK4DV5PR.js.map
@@ -0,0 +1,214 @@
1
+ import {
2
+ researcherProfile
3
+ } from "./chunk-5QOB7H74.js";
4
+ import {
5
+ supervise
6
+ } from "./chunk-RYBVU4M3.js";
7
+
8
+ // src/knowledge/improvement-job.ts
9
+ import {
10
+ evaluateKnowledgeBaseReadiness,
11
+ improveKnowledgeBase
12
+ } from "@tangle-network/agent-knowledge";
13
+
14
+ // src/knowledge/supervised-update.ts
15
+ var RESEARCH_SUPERVISOR_SYSTEM_PROMPT = [
16
+ "You are a research supervisor. You do not answer the research question yourself.",
17
+ "You create and manage researcher workers that improve one shared knowledge base until it is ready.",
18
+ "",
19
+ "Each round:",
20
+ "1. Read the goal and the gaps the readiness check still reports.",
21
+ "2. Decompose the open work into independent sub-topics.",
22
+ "3. Spawn one researcher per sub-topic; split by sub-topic, never duplicate work.",
23
+ "4. Wait for researchers to settle, steer or re-spawn thin sub-topics, then stop when the check passes.",
24
+ "",
25
+ "Conserve the shared budget. Prefer a small number of well-scoped researchers over a large fanout that re-covers the same ground."
26
+ ].join("\n");
27
+ function knowledgeReadinessDeliverable(options) {
28
+ return {
29
+ describe: `knowledge base at ${options.root} is ready for: ${options.goal}`,
30
+ async check() {
31
+ const result = await options.readiness({
32
+ root: options.root,
33
+ goal: options.goal,
34
+ readinessSpecs: options.readinessSpecs,
35
+ readinessTaskId: options.readinessTaskId,
36
+ readiness: options.readinessOptions
37
+ });
38
+ return typeof result === "boolean" ? result : result.ready;
39
+ }
40
+ };
41
+ }
42
+ function createSupervisedKnowledgeUpdater(options) {
43
+ return (input) => runSupervisedKnowledgeUpdate({
44
+ ...options,
45
+ root: input.candidateRoot ?? input.root ?? options.root,
46
+ goal: input.goal ?? options.goal,
47
+ findings: input.findings ?? options.findings,
48
+ metadata: { ...options.metadata, ...input.metadata }
49
+ });
50
+ }
51
+ async function runSupervisedKnowledgeUpdate(options) {
52
+ const { profile: workerProfile } = researcherProfile({ harness: options.harness });
53
+ const baseInstructions = options.supervisorSystemPrompt ?? RESEARCH_SUPERVISOR_SYSTEM_PROMPT;
54
+ const workerContract = workerProfile.prompt?.systemPrompt;
55
+ const systemPrompt = workerContract ? `${baseInstructions}
56
+
57
+ Each researcher worker you spawn follows this contract:
58
+ ${workerContract}` : baseInstructions;
59
+ const profile = {
60
+ name: "knowledge-research-supervisor",
61
+ model: options.supervisorModel,
62
+ systemPrompt
63
+ };
64
+ const run = options.runSupervised ?? supervise;
65
+ const task = formatSupervisedKnowledgeTask(options);
66
+ const supervised = await run(profile, task, {
67
+ ...options.superviseOptions,
68
+ budget: options.budget,
69
+ backend: options.backend,
70
+ deliverable: knowledgeReadinessDeliverable(options),
71
+ makeWorkerAgent: options.makeWorkerAgent,
72
+ allowedModels: options.allowedModels
73
+ });
74
+ return {
75
+ applied: supervised.kind === "winner",
76
+ summary: supervised.kind === "winner" ? "research supervisor completed and the knowledge base passed readiness" : `research supervisor stopped without a ready knowledge base: ${supervised.reason}`,
77
+ supervised,
78
+ metadata: {
79
+ supervised: true,
80
+ root: options.root,
81
+ goal: options.goal,
82
+ result: supervised.kind
83
+ }
84
+ };
85
+ }
86
+ function formatSupervisedKnowledgeTask(options) {
87
+ const sections = [
88
+ `Goal: ${options.goal}`,
89
+ `Knowledge base root: ${options.root}`,
90
+ options.readinessTaskId ? `Readiness task id: ${options.readinessTaskId}` : void 0,
91
+ options.readinessSpecs?.length ? `Readiness specs:
92
+ ${JSON.stringify(options.readinessSpecs, null, 2)}` : void 0,
93
+ options.findings?.length ? `Current findings:
94
+ ${JSON.stringify(options.findings, null, 2)}` : void 0,
95
+ options.metadata && Object.keys(options.metadata).length > 0 ? `Metadata:
96
+ ${JSON.stringify(options.metadata, null, 2)}` : void 0
97
+ ].filter((section) => Boolean(section));
98
+ return `${sections.join("\n\n")}
99
+
100
+ Update files under the knowledge base root only. Stop when the readiness check passes.`;
101
+ }
102
+
103
+ // src/knowledge/improvement-job.ts
104
+ function createAgentKnowledgeReadinessCheck(options) {
105
+ return async (input) => {
106
+ const readiness = await evaluateKnowledgeBaseReadiness({
107
+ root: input.root,
108
+ goal: input.goal ?? options.goal,
109
+ readinessSpecs: input.readinessSpecs ?? options.readinessSpecs,
110
+ readinessTaskId: input.readinessTaskId ?? options.readinessTaskId,
111
+ readiness: input.readiness ?? options.readiness,
112
+ strict: options.strict,
113
+ kbQuality: options.kbQuality
114
+ });
115
+ return {
116
+ ready: readiness.ready,
117
+ summary: readiness.summary,
118
+ metadata: {
119
+ dimensions: readiness.dimensions,
120
+ validationOk: readiness.validation.ok,
121
+ kbQualityOk: readiness.kbQuality.ok,
122
+ blockingMissing: readiness.readiness?.report.blockingMissingRequirements.length ?? 0
123
+ }
124
+ };
125
+ };
126
+ }
127
+ async function runKnowledgeImprovementJob(options) {
128
+ const {
129
+ allowedModels,
130
+ backend,
131
+ budget,
132
+ harness,
133
+ makeWorkerAgent,
134
+ onMeasurement,
135
+ readinessCheck,
136
+ runSupervised,
137
+ supervisorModel,
138
+ supervisorSystemPrompt,
139
+ superviseOptions,
140
+ ...knowledgeOptions
141
+ } = options;
142
+ const startedAtMs = Date.now();
143
+ const startedAt = new Date(startedAtMs).toISOString();
144
+ const supervisedSpent = emptySpent();
145
+ let updateCalls = 0;
146
+ let updateDurationMs = 0;
147
+ const readiness = readinessCheck ?? createAgentKnowledgeReadinessCheck(options);
148
+ const updateKnowledge = createSupervisedKnowledgeUpdater({
149
+ root: options.root,
150
+ goal: options.goal,
151
+ readiness,
152
+ readinessSpecs: options.readinessSpecs,
153
+ readinessTaskId: options.readinessTaskId,
154
+ readinessOptions: options.readiness,
155
+ budget,
156
+ backend,
157
+ makeWorkerAgent,
158
+ harness,
159
+ supervisorModel,
160
+ supervisorSystemPrompt,
161
+ superviseOptions,
162
+ allowedModels,
163
+ runSupervised
164
+ });
165
+ const resolvedImprovement = await improveKnowledgeBase({
166
+ ...knowledgeOptions,
167
+ updateKnowledge: async (input) => {
168
+ const updateStartedAt = Date.now();
169
+ updateCalls += 1;
170
+ const result = await updateKnowledge(input);
171
+ updateDurationMs += Date.now() - updateStartedAt;
172
+ addSpent(supervisedSpent, result.supervised);
173
+ return result;
174
+ }
175
+ });
176
+ const finishedAtMs = Date.now();
177
+ const measurement = {
178
+ startedAt,
179
+ finishedAt: new Date(finishedAtMs).toISOString(),
180
+ durationMs: finishedAtMs - startedAtMs,
181
+ updateCalls,
182
+ updateDurationMs,
183
+ supervisedSpent
184
+ };
185
+ await onMeasurement?.(measurement);
186
+ return {
187
+ improvement: resolvedImprovement,
188
+ measurement,
189
+ promoted: resolvedImprovement.promoted,
190
+ blocked: resolvedImprovement.blocked
191
+ };
192
+ }
193
+ function emptySpent() {
194
+ return { iterations: 0, inputTokens: 0, outputTokens: 0, usd: 0, ms: 0 };
195
+ }
196
+ function addSpent(target, result) {
197
+ const spent = result.spentTotal;
198
+ target.iterations += spent.iterations ?? 0;
199
+ target.inputTokens += spent.tokens?.input ?? 0;
200
+ target.outputTokens += spent.tokens?.output ?? 0;
201
+ target.usd += spent.usd ?? 0;
202
+ target.ms += spent.ms ?? 0;
203
+ }
204
+
205
+ export {
206
+ RESEARCH_SUPERVISOR_SYSTEM_PROMPT,
207
+ knowledgeReadinessDeliverable,
208
+ createSupervisedKnowledgeUpdater,
209
+ runSupervisedKnowledgeUpdate,
210
+ formatSupervisedKnowledgeTask,
211
+ createAgentKnowledgeReadinessCheck,
212
+ runKnowledgeImprovementJob
213
+ };
214
+ //# sourceMappingURL=chunk-R2VAJGR3.js.map