@tangle-network/agent-eval 0.100.0 → 0.100.1

package/dist/{chunk-IZCEK2HR.js.map → chunk-2MLIEQSN.js.map}

@@ -1 +1 @@
-{"version":3,"sources":["../src/trace-analyst/prompts.ts","../src/trace-analyst/analyst.ts"],"sourcesContent":["/** Ax RLM prompt for bounded trace discovery and evidence-backed analysis. */\n\nexport const TRACE_ANALYST_ACTOR_DESCRIPTION = `You answer questions about an OTLP-shaped JSONL trace dataset using the trace tools provided in the \\`traces\\` namespace.\n\nDISCOVERY → NARROW → DEEP-READ protocol — follow exactly:\n\n1. ALWAYS call \\`traces.getDatasetOverview({})\\` FIRST without a regex_pattern. The result tells you total_traces, raw_jsonl_bytes, services, agents, models, and sample_trace_ids (real ids — never fabricate one).\n\n2. Use raw_jsonl_bytes to gauge how expensive raw scans will be. \\`filters.regex_pattern\\` is the one scan-heavy filter on getDatasetOverview / queryTraces / countTraces — narrow with indexed fields (has_errors, model_names, service_names, agent_names, time bounds) BEFORE adding a regex on a large dataset.\n\n3. To list more traces than the sample, call \\`traces.queryTraces({ filters?, limit, offset? })\\`. Each summary carries raw_jsonl_bytes — use it to choose between viewTrace and searchTrace BEFORE calling either.\n\n4. Per-trace inspection:\n   - SMALL trace (raw_jsonl_bytes well under 150_000): call \\`traces.viewTrace({ trace_id })\\`. Returns all spans. Per-attribute payloads are head-capped at ~4KB; large \\`input.value\\` / \\`output.value\\` / \\`llm.input_messages\\` will show a \\`[trace-analyst truncated: N bytes]\\` marker.\n   - LARGE trace (raw_jsonl_bytes near or above 150_000, or you saw an \\`oversized\\` response): use \\`traces.searchTrace({ trace_id, regex_pattern })\\` to get bounded SpanMatchRecords (span metadata + matched text + surrounding context). Then call \\`traces.viewSpans({ trace_id, span_ids: [...] })\\` for surgical reads (~16KB cap, 4× higher than discovery), or \\`traces.searchSpan({ trace_id, span_id, regex_pattern })\\` for one large span. Stays bounded regardless of trace size.\n   - Useful regex patterns: \\`STATUS_CODE_ERROR\\` (failures), tool names like \\`grep\\` or \\`view_trace\\`, error strings like \\`MaxTurnsExceeded\\`, model names, attribute keys.\n\n5. ONLY call viewTrace / viewSpans / searchTrace / searchSpan with trace/span ids you have already seen in sample_trace_ids, a queryTraces page, or a previous search result. Never invent ids.\n\n5a. **Result-shape contract** — searchTrace and searchSpan return \\`{ trace_id, hits, total_matches, has_more }\\`. Iterate \\`result.hits\\` (NOT result.matches). Each hit has \\`{ span_id, span_name, span_kind, attribute_path, matched_text, context_before, context_after, match_offset }\\`. viewTrace returns \\`{ trace_id, spans }\\` (or \\`oversized\\`). viewSpans returns \\`{ trace_id, spans, missing_span_ids, truncated_attribute_count }\\`. Never assume a field name — log the result shape first if unsure.\n\n6. If viewTrace returns an \\`oversized\\` summary instead of \\`spans\\`, DO NOT retry the same call. Read the summary's top_span_names, span_count, span_response_bytes_max, error_span_count to plan a follow-up: switch to searchTrace (or searchSpan for one large span), then viewSpans on a smaller, surgical span_ids set.\n\n7. If searchTrace or searchSpan returns has_more=true, REFINE the regex to be more specific rather than blindly raising max_matches.\n\n8. If a tool errors (invalid regex, range error), STOP and reconsider — don't retry with a guessed id or argument. Use the discovery tools above to recover.\n\n9. If a ~4KB-truncated payload from viewTrace / searchTrace matters for your answer, first try viewSpans on that span id (~16KB cap). If a 16KB-truncated payload from viewSpans still matters, narrow further with searchSpan against a more specific regex rather than asking for the full payload again.\n\n10. If maxDepth > 0 and the question splits into independent semantic branches, delegate well-defined subtasks to subagents using \\`await llmQuery(...)\\`. Pass narrow context and a focused query. Examples:\n\n    const reviews = await llmQuery([\n      { query: 'Drill into trace abc123 — what tool calls preceded the failure?', context: { trace_id: 'abc123' } },\n      { query: 'Drill into trace def456 — same failure mode?', context: { trace_id: 'def456' } },\n    ]);\n\nOBSERVABILITY rules:\n- Each non-final actor turn must emit at least one \\`console.log(...)\\` for evidence. Up to 3 logs per turn is fine when correlating multiple data sources (e.g. one log for findings list, one for source-file content, one for derived analysis).\n- Do NOT combine \\`console.log\\` with \\`final(...)\\` or \\`askClarification(...)\\` in the same turn — finish gathering data first, then call final on its own turn.\n- Reuse runtime variables across turns; don't recompute.\n- When done, call \\`await final(answer)\\` with the fully-formed report. The responder rewrites the answer into output fields; if you only pass a vague summary string the responder has nothing concrete to format.\n\nCRITICAL — \\`final()\\` payload contract for evidence-grounded analysis tasks:\n- Pass a STRUCTURED object as the second arg with the actual data the responder needs to format the answer. Do NOT pass abstract instructions; pass evidence.\n- Example for per-item verdict tasks:\n  \\`\\`\\`js\n  await final(\"Format the per-item verdict report from the evidence below.\", {\n    findings: [\n      { id: 'sub-1-finding-1', claim: '...', verdict: 'TRUE-POSITIVE', evidence: 'lines 42-45 of contracts/X.sol show ...' },\n      ...all items\n    ],\n    systemic_summary: '3 sentences I wrote based on the evidence above'\n  });\n  \\`\\`\\`\n- Calling \\`final(\"answer\", {})\\` with no evidence is a failure mode — the responder will hallucinate or echo back the field names. Always include the gathered data.\n- Premature final after a single viewSpans call is INSUFFICIENT for per-finding analysis tasks. Read the requested attributes (e.g. \\`spans[i].attributes['redteam.finding.title']\\`), and for each one perform the requested cross-reference (e.g. read the source SPAN's \\`attributes['source.content']\\`).\n\nOUTPUT contract — your final answer must include:\n- A clear prose conclusion answering the user's question.\n- Trace ids and span ids cited as evidence for each claim.\n- Failure modes named in the user's domain language, with frequency and concrete examples.\n\nDo NOT invent trace ids, span ids, error messages, or model names. Every fact must be traceable to a tool result.`\n\nexport const TRACE_ANALYST_ACTOR_DESCRIPTION_VERSION = 'trace-analyst-actor-v5-2026-05-06'\n\n/** Subagent prompt for focused trace-inspection subtasks. */\nexport const TRACE_ANALYST_SUBAGENT_DESCRIPTION = `You are a trace-analyst subagent. Your parent has delegated a focused trace-inspection question. Use the same DISCOVERY → NARROW → DEEP-READ protocol but stay tightly scoped: do exactly what was asked, return a concise compact answer, do NOT spawn further subagents unless the parent's question is genuinely multi-branch.\n\nCite trace ids and span ids for every claim. Do NOT invent ids.`\n","import { type AxActorTurn, type AxAIService, type AxFunction, AxJSRuntime, agent } from '@ax-llm/ax'\nimport {\n  TRACE_ANALYST_ACTOR_DESCRIPTION,\n  TRACE_ANALYST_ACTOR_DESCRIPTION_VERSION,\n  TRACE_ANALYST_SUBAGENT_DESCRIPTION,\n} from './prompts'\nimport type { TraceAnalysisStore } from './store'\nimport { OtlpFileTraceStore, TraceFileMissingError } from './store-otlp'\nimport { buildTraceAnalystTools } from './tools'\n\nexport interface AnalyzeTracesInput {\n  /** The user-facing question. Domain framing belongs here, not in the\n   *  actor description. */\n  question: string\n}\n\nexport interface AnalyzeTracesResult {\n  /** The responder's prose answer. */\n  answer: string\n  /** Bulleted findings extracted from the responder's structured output. */\n  findings: string[]\n  /** Per-actor-turn snapshots captured via `actorTurnCallback`. */\n  turns: AnalyzeTracesTurnSnapshot[]\n  /** Total turns the actor took. */\n  turnCount: number\n  /** Token usage by role. */\n  usage: TraceAnalystUsage\n  /** Full system + assistant + tool message log by role. */\n  chatLog: TraceAnalystChatLog\n  /** Prompt version that produced this run. */\n  actorPromptVersion: string\n}\n\nexport interface TraceAnalystUsage {\n  actor: TraceAnalystUsageEntry[]\n  responder: TraceAnalystUsageEntry[]\n}\n\nexport interface TraceAnalystUsageEntry {\n  [key: string]: unknown\n}\n\nexport interface TraceAnalystChatLog {\n  actor: TraceAnalystChatMessage[]\n  responder: TraceAnalystChatMessage[]\n}\n\nexport interface TraceAnalystChatMessage {\n  [key: string]: unknown\n}\n\nexport interface AnalyzeTracesTurnSnapshot {\n  turn: number\n  isError: boolean\n  /** The JS code the actor produced for this turn. */\n  code: string\n  /** The formatted action-log entry the actor sees on the next turn. */\n  output: string\n  /** Provider thought (when `actorOptions.showThoughts` is true and the\n   *  provider returns it). */\n  thought?: string\n}\n\nexport interface AnalyzeTracesOptions {\n  /** Trace data source. Pass either an OTLP-JSONL path or a custom store. */\n  source: string | TraceAnalysisStore\n  /** Caller-provided AxAIService. */\n  ai: AxAIService\n  /** Model id forwarded to actor + responder. */\n  model?: string\n  /** Recursion depth. 0 = no sub-agent dispatch. Default 1. */\n  maxDepth?: number\n  /** Maximum actor turns. Default 12. */\n  maxTurns?: number\n  /** Maximum parallel sub-agent calls in batched llmQuery. Default 2. */\n  maxParallelSubagents?: number\n  /** Override the actor description. */\n  actorDescription?: string\n  /** Override the subagent description. */\n  subagentDescription?: string\n  /** Per-turn observability hook. */\n  onTurn?: (turn: AnalyzeTracesTurnSnapshot) => void | Promise<void>\n  /** Override max runtime characters per turn. Default 6000. */\n  maxRuntimeChars?: number\n  /** When set, every turn's snapshot is appended to this JSONL file\n   *  immediately. If the analyst crashes mid-loop (provider 503,\n   *  network error, validator reject) the partial reasoning is still\n   *  on disk. Replay the file with the responder afterward to recover\n   *  evidence. */\n  progressLogPath?: string\n}\n\n/**\n * Run the trace analyst.\n *\n * Throws:\n *   - `TraceFileMissingError` if `source` is a path and doesn't exist.\n *   - `AxAgentClarificationError` if the analyst asks for clarification.\n *   - Provider errors (auth, rate limits) propagate from the AI service.\n */\nexport async function analyzeTraces(\n  input: AnalyzeTracesInput,\n  options: AnalyzeTracesOptions,\n): Promise<AnalyzeTracesResult> {\n  if (!input.question || typeof input.question !== 'string') {\n    throw new TypeError('analyzeTraces: input.question must be a non-empty string')\n  }\n\n  const store: TraceAnalysisStore =\n    typeof options.source === 'string'\n      ? new OtlpFileTraceStore({ path: options.source })\n      : options.source\n\n  // Pre-warm file stores so missing inputs fail before the RLM starts.\n  if (store instanceof OtlpFileTraceStore) {\n    await store.ensureIndexed()\n  }\n\n  const tools: AxFunction[] = buildTraceAnalystTools({ store })\n  const turns: AnalyzeTracesTurnSnapshot[] = []\n\n  // Persist each turn as JSONL so interrupted analyst runs keep useful evidence.\n  let progressFs: import('node:fs').WriteStream | undefined\n  if (options.progressLogPath) {\n    const { createWriteStream } = await import('node:fs')\n    const { mkdir } = await import('node:fs/promises')\n    const { dirname } = await import('node:path')\n    await mkdir(dirname(options.progressLogPath), { recursive: true })\n    progressFs = createWriteStream(options.progressLogPath, { flags: 'a' })\n  }\n\n  const actorTurnCallback = async (turn: AxActorTurn): Promise<void> => {\n    const snap: AnalyzeTracesTurnSnapshot = {\n      turn: turn.turn,\n      isError: turn.isError,\n      code: turn.code,\n      output: turn.output,\n      thought: turn.thought,\n    }\n    turns.push(snap)\n    if (progressFs) {\n      try {\n        progressFs.write(`${JSON.stringify({ ...snap, ts: Date.now() })}\\n`)\n      } catch {\n        // Progress logging must never fail the analyst.\n      }\n    }\n    if (options.onTurn) await options.onTurn(snap)\n  }\n\n  const maxDepth = options.maxDepth ?? 1\n  const maxTurns = options.maxTurns ?? 12\n  const maxParallelSubagents = options.maxParallelSubagents ?? 2\n  const maxRuntimeChars = options.maxRuntimeChars ?? 6000\n\n  const analyst = agent<{ question: string }, { answer: string; findings: string[] }>(\n    // `reasoning!` is an internal (Ax `!`) scratchpad field: generated first to\n    // force reason-before-conclude, stripped from the returned output — so the\n    // consumed shape stays { answer, findings }. Brings the trace-analyst to the\n    // same prose-first CoT ordering the kind-factory gets from its `report` field.\n    'question:string -> reasoning!:string, answer:string, findings:string[]',\n    {\n      agentIdentity: {\n        name: 'TraceAnalyst',\n        description:\n          'Analyzes OTLP-shaped JSONL traces using bounded discovery tools to identify systemic failure modes.',\n      },\n      contextFields: ['question'],\n      runtime: new AxJSRuntime({\n        permissions: [],\n        blockDynamicImport: true,\n        allowedModules: [],\n        freezeIntrinsics: true,\n        blockShadowRealm: true,\n        // RLM stdout mode relies on runtime bindings persisting across turns.\n        preventGlobalThisExtensions: false,\n      }),\n      mode: maxDepth > 0 ? 'advanced' : 'simple',\n      recursionOptions: maxDepth > 0 ? { maxDepth } : undefined,\n      maxTurns,\n      maxRuntimeChars,\n      maxBatchedLlmQueryConcurrency: maxParallelSubagents,\n      promptLevel: 'detailed',\n      // Trace analysis depends on exact prior tool results and runtime variables.\n      contextPolicy: { preset: 'full', budget: 'balanced' },\n      functions: { local: tools },\n      actorOptions: {\n        description: options.actorDescription ?? TRACE_ANALYST_ACTOR_DESCRIPTION,\n        ...(options.model ? { model: options.model } : {}),\n        // Keep actor messages tool-call/content shaped across reasoning models.\n        showThoughts: false,\n        thinkingTokenBudget: 'none',\n      },\n      responderOptions: {\n        ...(options.model ? { model: options.model } : {}),\n        description: options.subagentDescription ?? TRACE_ANALYST_SUBAGENT_DESCRIPTION,\n        showThoughts: false,\n      },\n      actorTurnCallback,\n      bubbleErrors: [TraceFileMissingError],\n    },\n  )\n\n  let result: { answer: unknown; findings: unknown }\n  try {\n    result = await analyst.forward(options.ai, { question: input.question })\n  } finally {\n    if (progressFs) {\n      await new Promise<void>((resolve) => progressFs!.end(() => resolve()))\n    }\n  }\n\n  return {\n    answer: typeof result.answer === 'string' ? result.answer : String(result.answer ?? ''),\n    findings: Array.isArray(result.findings)\n      ? result.findings.filter((s): s is string => typeof s === 'string')\n      : [],\n    turns,\n    turnCount: turns.length,\n    usage: normalizeRoleArrays(analyst.getUsage()),\n    chatLog: normalizeRoleArrays(analyst.getChatLog()),\n    actorPromptVersion: TRACE_ANALYST_ACTOR_DESCRIPTION_VERSION,\n  }\n}\n\nfunction normalizeRoleArrays(value: unknown): {\n  actor: Record<string, unknown>[]\n  responder: Record<string, unknown>[]\n} {\n  const record = value && typeof value === 'object' ? (value as Record<string, unknown>) : {}\n  return {\n    actor: normalizeRecordArray(record.actor),\n    responder: normalizeRecordArray(record.responder),\n  }\n}\n\nfunction normalizeRecordArray(value: unknown): Record<string, unknown>[] {\n  if (!Array.isArray(value)) return []\n  return value.map((item) =>\n    item && typeof item === 'object' ? { ...(item as Record<string, unknown>) } : { value: item },\n  )\n}\n"],"mappings":";;;;;;;AAEO,IAAM,kCAAkC;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AA8DxC,IAAM,0CAA0C;AAGhD,IAAM,qCAAqC;AAAA;AAAA;;;ACnElD,SAA8D,aAAa,aAAa;AAoGxF,eAAsB,cACpB,OACA,SAC8B;AAC9B,MAAI,CAAC,MAAM,YAAY,OAAO,MAAM,aAAa,UAAU;AACzD,UAAM,IAAI,UAAU,0DAA0D;AAAA,EAChF;AAEA,QAAM,QACJ,OAAO,QAAQ,WAAW,WACtB,IAAI,mBAAmB,EAAE,MAAM,QAAQ,OAAO,CAAC,IAC/C,QAAQ;AAGd,MAAI,iBAAiB,oBAAoB;AACvC,UAAM,MAAM,cAAc;AAAA,EAC5B;AAEA,QAAM,QAAsB,uBAAuB,EAAE,MAAM,CAAC;AAC5D,QAAM,QAAqC,CAAC;AAG5C,MAAI;AACJ,MAAI,QAAQ,iBAAiB;AAC3B,UAAM,EAAE,kBAAkB,IAAI,MAAM,OAAO,IAAS;AACpD,UAAM,EAAE,MAAM,IAAI,MAAM,OAAO,aAAkB;AACjD,UAAM,EAAE,QAAQ,IAAI,MAAM,OAAO,MAAW;AAC5C,UAAM,MAAM,QAAQ,QAAQ,eAAe,GAAG,EAAE,WAAW,KAAK,CAAC;AACjE,iBAAa,kBAAkB,QAAQ,iBAAiB,EAAE,OAAO,IAAI,CAAC;AAAA,EACxE;AAEA,QAAM,oBAAoB,OAAO,SAAqC;AACpE,UAAM,OAAkC;AAAA,MACtC,MAAM,KAAK;AAAA,MACX,SAAS,KAAK;AAAA,MACd,MAAM,KAAK;AAAA,MACX,QAAQ,KAAK;AAAA,MACb,SAAS,KAAK;AAAA,IAChB;AACA,UAAM,KAAK,IAAI;AACf,QAAI,YAAY;AACd,UAAI;AACF,mBAAW,MAAM,GAAG,KAAK,UAAU,EAAE,GAAG,MAAM,IAAI,KAAK,IAAI,EAAE,CAAC,CAAC;AAAA,CAAI;AAAA,MACrE,QAAQ;AAAA,MAER;AAAA,IACF;AACA,QAAI,QAAQ,OAAQ,OAAM,QAAQ,OAAO,IAAI;AAAA,EAC/C;AAEA,QAAM,WAAW,QAAQ,YAAY;AACrC,QAAM,WAAW,QAAQ,YAAY;AACrC,QAAM,uBAAuB,QAAQ,wBAAwB;AAC7D,QAAM,kBAAkB,QAAQ,mBAAmB;AAEnD,QAAM,UAAU;AAAA;AAAA;AAAA;AAAA;AAAA,IAKd;AAAA,IACA;AAAA,MACE,eAAe;AAAA,QACb,MAAM;AAAA,QACN,aACE;AAAA,MACJ;AAAA,MACA,eAAe,CAAC,UAAU;AAAA,MAC1B,SAAS,IAAI,YAAY;AAAA,QACvB,aAAa,CAAC;AAAA,QACd,oBAAoB;AAAA,QACpB,gBAAgB,CAAC;AAAA,QACjB,kBAAkB;AAAA,QAClB,kBAAkB;AAAA;AAAA,QAElB,6BAA6B;AAAA,MAC/B,CAAC;AAAA,MACD,MAAM,WAAW,IAAI,aAAa;AAAA,MAClC,kBAAkB,WAAW,IAAI,EAAE,SAAS,IAAI;AAAA,MAChD;AAAA,MACA;AAAA,MACA,+BAA+B;AAAA,MAC/B,aAAa;AAAA;AAAA,MAEb,eAAe,EAAE,QAAQ,QAAQ,QAAQ,WAAW;AAAA,MACpD,WAAW,EAAE,OAAO,MAAM;AAAA,MAC1B,cAAc;AAAA,QACZ,aAAa,QAAQ,oBAAoB;AAAA,QACzC,GAAI,QAAQ,QAAQ,EAAE,OAAO,QAAQ,MAAM,IAAI,CAAC;AAAA;AAAA,QAEhD,cAAc;AAAA,QACd,qBAAqB;AAAA,MACvB;AAAA,MACA,kBAAkB;AAAA,QAChB,GAAI,QAAQ,QAAQ,EAAE,OAAO,QAAQ,MAAM,IAAI,CAAC;AAAA,QAChD,aAAa,QAAQ,uBAAuB;AAAA,QAC5C,cAAc;AAAA,MAChB;AAAA,MACA;AAAA,MACA,cAAc,CAAC,qBAAqB;AAAA,IACtC;AAAA,EACF;AAEA,MAAI;AACJ,MAAI;AACF,aAAS,MAAM,QAAQ,QAAQ,QAAQ,IAAI,EAAE,UAAU,MAAM,SAAS,CAAC;AAAA,EACzE,UAAE;AACA,QAAI,YAAY;AACd,YAAM,IAAI,QAAc,CAAC,YAAY,WAAY,IAAI,MAAM,QAAQ,CAAC,CAAC;AAAA,IACvE;AAAA,EACF;AAEA,SAAO;AAAA,IACL,QAAQ,OAAO,OAAO,WAAW,WAAW,OAAO,SAAS,OAAO,OAAO,UAAU,EAAE;AAAA,IACtF,UAAU,MAAM,QAAQ,OAAO,QAAQ,IACnC,OAAO,SAAS,OAAO,CAAC,MAAmB,OAAO,MAAM,QAAQ,IAChE,CAAC;AAAA,IACL;AAAA,IACA,WAAW,MAAM;AAAA,IACjB,OAAO,oBAAoB,QAAQ,SAAS,CAAC;AAAA,IAC7C,SAAS,oBAAoB,QAAQ,WAAW,CAAC;AAAA,IACjD,oBAAoB;AAAA,EACtB;AACF;AAEA,SAAS,oBAAoB,OAG3B;AACA,QAAM,SAAS,SAAS,OAAO,UAAU,WAAY,QAAoC,CAAC;AAC1F,SAAO;AAAA,IACL,OAAO,qBAAqB,OAAO,KAAK;AAAA,IACxC,WAAW,qBAAqB,OAAO,SAAS;AAAA,EAClD;AACF;AAEA,SAAS,qBAAqB,OAA2C;AACvE,MAAI,CAAC,MAAM,QAAQ,KAAK,EAAG,QAAO,CAAC;AACnC,SAAO,MAAM;AAAA,IAAI,CAAC,SAChB,QAAQ,OAAO,SAAS,WAAW,EAAE,GAAI,KAAiC,IAAI,EAAE,OAAO,KAAK;AAAA,EAC9F;AACF;","names":[]}
+{"version":3,"sources":["../src/trace-analyst/prompts.ts","../src/trace-analyst/analyst.ts"],"sourcesContent":["/** Ax RLM prompt for bounded trace discovery and evidence-backed analysis. */\n\nexport const TRACE_ANALYST_ACTOR_DESCRIPTION = `You answer questions about an OTLP-shaped JSONL trace dataset using the trace tools provided in the \\`traces\\` namespace.\n\nDISCOVERY → NARROW → DEEP-READ protocol — follow exactly:\n\n1. ALWAYS call \\`traces.getDatasetOverview({})\\` FIRST without a regex_pattern. The result tells you total_traces, raw_jsonl_bytes, services, agents, models, and sample_trace_ids (real ids — never fabricate one).\n\n2. Use raw_jsonl_bytes to gauge how expensive raw scans will be. \\`filters.regex_pattern\\` is the one scan-heavy filter on getDatasetOverview / queryTraces / countTraces — narrow with indexed fields (has_errors, model_names, service_names, agent_names, time bounds) BEFORE adding a regex on a large dataset.\n\n3. To list more traces than the sample, call \\`traces.queryTraces({ filters?, limit, offset? })\\`. Each summary carries raw_jsonl_bytes — use it to choose between viewTrace and searchTrace BEFORE calling either.\n\n4. Per-trace inspection:\n   - SMALL trace (raw_jsonl_bytes well under 150_000): call \\`traces.viewTrace({ trace_id })\\`. Returns all spans. Per-attribute payloads are head-capped at ~4KB; large \\`input.value\\` / \\`output.value\\` / \\`llm.input_messages\\` will show a \\`[trace-analyst truncated: N bytes]\\` marker.\n   - LARGE trace (raw_jsonl_bytes near or above 150_000, or you saw an \\`oversized\\` response): use \\`traces.searchTrace({ trace_id, regex_pattern })\\` to get bounded SpanMatchRecords (span metadata + matched text + surrounding context). Then call \\`traces.viewSpans({ trace_id, span_ids: [...] })\\` for surgical reads (~16KB cap, 4× higher than discovery), or \\`traces.searchSpan({ trace_id, span_id, regex_pattern })\\` for one large span. Stays bounded regardless of trace size.\n   - Useful regex patterns: \\`STATUS_CODE_ERROR\\` (failures), tool names like \\`grep\\` or \\`view_trace\\`, error strings like \\`MaxTurnsExceeded\\`, model names, attribute keys.\n\n5. ONLY call viewTrace / viewSpans / searchTrace / searchSpan with trace/span ids you have already seen in sample_trace_ids, a queryTraces page, or a previous search result. Never invent ids.\n\n5a. **Result-shape contract** — searchTrace and searchSpan return \\`{ trace_id, hits, total_matches, has_more }\\`. Iterate \\`result.hits\\` (NOT result.matches). Each hit has \\`{ span_id, span_name, span_kind, attribute_path, matched_text, context_before, context_after, match_offset }\\`. viewTrace returns \\`{ trace_id, spans }\\` (or \\`oversized\\`). viewSpans returns \\`{ trace_id, spans, missing_span_ids, truncated_attribute_count }\\`. Never assume a field name — log the result shape first if unsure.\n\n6. If viewTrace returns an \\`oversized\\` summary instead of \\`spans\\`, DO NOT retry the same call. Read the summary's top_span_names, span_count, span_response_bytes_max, error_span_count to plan a follow-up: switch to searchTrace (or searchSpan for one large span), then viewSpans on a smaller, surgical span_ids set.\n\n7. If searchTrace or searchSpan returns has_more=true, REFINE the regex to be more specific rather than blindly raising max_matches.\n\n8. If a tool errors (invalid regex, range error), STOP and reconsider — don't retry with a guessed id or argument. Use the discovery tools above to recover.\n\n9. If a ~4KB-truncated payload from viewTrace / searchTrace matters for your answer, first try viewSpans on that span id (~16KB cap). If a 16KB-truncated payload from viewSpans still matters, narrow further with searchSpan against a more specific regex rather than asking for the full payload again.\n\n10. If maxDepth > 0 and the question splits into independent semantic branches, delegate well-defined subtasks to subagents using \\`await llmQuery(...)\\`. Pass narrow context and a focused query. Examples:\n\n    const reviews = await llmQuery([\n      { query: 'Drill into trace abc123 — what tool calls preceded the failure?', context: { trace_id: 'abc123' } },\n      { query: 'Drill into trace def456 — same failure mode?', context: { trace_id: 'def456' } },\n    ]);\n\nOBSERVABILITY rules:\n- Each non-final actor turn must emit at least one \\`console.log(...)\\` for evidence. Up to 3 logs per turn is fine when correlating multiple data sources (e.g. one log for findings list, one for source-file content, one for derived analysis).\n- Do NOT combine \\`console.log\\` with \\`final(...)\\` or \\`askClarification(...)\\` in the same turn — finish gathering data first, then call final on its own turn.\n- Reuse runtime variables across turns; don't recompute.\n- When done, call \\`await final(answer)\\` with the fully-formed report. The responder rewrites the answer into output fields; if you only pass a vague summary string the responder has nothing concrete to format.\n\nCRITICAL — \\`final()\\` payload contract for evidence-grounded analysis tasks:\n- Pass a STRUCTURED object as the second arg with the actual data the responder needs to format the answer. Do NOT pass abstract instructions; pass evidence.\n- Example for per-item verdict tasks:\n  \\`\\`\\`js\n  await final(\"Format the per-item verdict report from the evidence below.\", {\n    findings: [\n      { id: 'sub-1-finding-1', claim: '...', verdict: 'TRUE-POSITIVE', evidence: 'lines 42-45 of contracts/X.sol show ...' },\n      ...all items\n    ],\n    systemic_summary: '3 sentences I wrote based on the evidence above'\n  });\n  \\`\\`\\`\n- Calling \\`final(\"answer\", {})\\` with no evidence is a failure mode — the responder will hallucinate or echo back the field names. Always include the gathered data.\n- Premature final after a single viewSpans call is INSUFFICIENT for per-finding analysis tasks. Read the requested attributes (e.g. \\`spans[i].attributes['redteam.finding.title']\\`), and for each one perform the requested cross-reference (e.g. read the source SPAN's \\`attributes['source.content']\\`).\n\nOUTPUT contract — your final answer must include:\n- A clear prose conclusion answering the user's question.\n- Trace ids and span ids cited as evidence for each claim.\n- Failure modes named in the user's domain language, with frequency and concrete examples.\n\nDo NOT invent trace ids, span ids, error messages, or model names. Every fact must be traceable to a tool result.`\n\nexport const TRACE_ANALYST_ACTOR_DESCRIPTION_VERSION = 'trace-analyst-actor-v5-2026-05-06'\n\n/** Subagent prompt for focused trace-inspection subtasks. */\nexport const TRACE_ANALYST_SUBAGENT_DESCRIPTION = `You are a trace-analyst subagent. Your parent has delegated a focused trace-inspection question. Use the same DISCOVERY → NARROW → DEEP-READ protocol but stay tightly scoped: do exactly what was asked, return a concise compact answer, do NOT spawn further subagents unless the parent's question is genuinely multi-branch.\n\nCite trace ids and span ids for every claim. Do NOT invent ids.`\n","import { type AxActorTurn, type AxAIService, type AxFunction, AxJSRuntime, agent } from '@ax-llm/ax'\nimport {\n  TRACE_ANALYST_ACTOR_DESCRIPTION,\n  TRACE_ANALYST_ACTOR_DESCRIPTION_VERSION,\n  TRACE_ANALYST_SUBAGENT_DESCRIPTION,\n} from './prompts'\nimport type { TraceAnalysisStore } from './store'\nimport { OtlpFileTraceStore, TraceFileMissingError } from './store-otlp'\nimport { buildTraceAnalystTools } from './tools'\n\nexport interface AnalyzeTracesInput {\n  /** The user-facing question. Domain framing belongs here, not in the\n   *  actor description. */\n  question: string\n}\n\nexport interface AnalyzeTracesResult {\n  /** The responder's prose answer. */\n  answer: string\n  /** Bulleted findings extracted from the responder's structured output. */\n  findings: string[]\n  /** Per-actor-turn snapshots captured via `actorTurnCallback`. */\n  turns: AnalyzeTracesTurnSnapshot[]\n  /** Total turns the actor took. */\n  turnCount: number\n  /** Token usage by role. */\n  usage: TraceAnalystUsage\n  /** Full system + assistant + tool message log by role. */\n  chatLog: TraceAnalystChatLog\n  /** Prompt version that produced this run. */\n  actorPromptVersion: string\n}\n\nexport interface TraceAnalystUsage {\n  actor: TraceAnalystUsageEntry[]\n  responder: TraceAnalystUsageEntry[]\n}\n\nexport interface TraceAnalystUsageEntry {\n  [key: string]: unknown\n}\n\nexport interface TraceAnalystChatLog {\n  actor: TraceAnalystChatMessage[]\n  responder: TraceAnalystChatMessage[]\n}\n\nexport interface TraceAnalystChatMessage {\n  [key: string]: unknown\n}\n\nexport interface AnalyzeTracesTurnSnapshot {\n  turn: number\n  isError: boolean\n  /** The JS code the actor produced for this turn. */\n  code: string\n  /** The formatted action-log entry the actor sees on the next turn. */\n  output: string\n  /** Provider thought (when `actorOptions.showThoughts` is true and the\n   *  provider returns it). */\n  thought?: string\n}\n\nexport interface AnalyzeTracesOptions {\n  /** Trace data source. Pass either an OTLP-JSONL path or a custom store. */\n  source: string | TraceAnalysisStore\n  /** Caller-provided AxAIService. */\n  ai: AxAIService\n  /** Model id forwarded to actor + responder. */\n  model?: string\n  /** Recursion depth. 0 = no sub-agent dispatch. Default 1. */\n  maxDepth?: number\n  /** Maximum actor turns. Default 12. */\n  maxTurns?: number\n  /** Maximum parallel sub-agent calls in batched llmQuery. Default 2. */\n  maxParallelSubagents?: number\n  /** Override the actor description. */\n  actorDescription?: string\n  /** Override the subagent description. */\n  subagentDescription?: string\n  /** Per-turn observability hook. */\n  onTurn?: (turn: AnalyzeTracesTurnSnapshot) => void | Promise<void>\n  /** Override max runtime characters per turn. Default 6000. */\n  maxRuntimeChars?: number\n  /** When set, every turn's snapshot is appended to this JSONL file\n   *  immediately. If the analyst crashes mid-loop (provider 503,\n   *  network error, validator reject) the partial reasoning is still\n   *  on disk. Replay the file with the responder afterward to recover\n   *  evidence. */\n  progressLogPath?: string\n}\n\n/**\n * Run the trace analyst.\n *\n * Throws:\n *   - `TraceFileMissingError` if `source` is a path and doesn't exist.\n *   - `AxAgentClarificationError` if the analyst asks for clarification.\n *   - Provider errors (auth, rate limits) propagate from the AI service.\n */\nexport async function analyzeTraces(\n  input: AnalyzeTracesInput,\n  options: AnalyzeTracesOptions,\n): Promise<AnalyzeTracesResult> {\n  if (!input.question || typeof input.question !== 'string') {\n    throw new TypeError('analyzeTraces: input.question must be a non-empty string')\n  }\n\n  const store: TraceAnalysisStore =\n    typeof options.source === 'string'\n      ? new OtlpFileTraceStore({ path: options.source })\n      : options.source\n\n  // Pre-warm file stores so missing inputs fail before the RLM starts.\n  if (store instanceof OtlpFileTraceStore) {\n    await store.ensureIndexed()\n  }\n\n  const tools: AxFunction[] = buildTraceAnalystTools({ store })\n  const turns: AnalyzeTracesTurnSnapshot[] = []\n\n  // Persist each turn as JSONL so interrupted analyst runs keep useful evidence.\n  let progressFs: import('node:fs').WriteStream | undefined\n  if (options.progressLogPath) {\n    const { createWriteStream } = await import('node:fs')\n    const { mkdir } = await import('node:fs/promises')\n    const { dirname } = await import('node:path')\n    await mkdir(dirname(options.progressLogPath), { recursive: true })\n    progressFs = createWriteStream(options.progressLogPath, { flags: 'a' })\n  }\n\n  const actorTurnCallback = async (turn: AxActorTurn): Promise<void> => {\n    const snap: AnalyzeTracesTurnSnapshot = {\n      turn: turn.turn,\n      isError: turn.isError,\n      code: turn.code,\n      output: turn.output,\n      thought: turn.thought,\n    }\n    turns.push(snap)\n    if (progressFs) {\n      try {\n        progressFs.write(`${JSON.stringify({ ...snap, ts: Date.now() })}\\n`)\n      } catch {\n        // Progress logging must never fail the analyst.\n      }\n    }\n    if (options.onTurn) await options.onTurn(snap)\n  }\n\n  const maxDepth = options.maxDepth ?? 1\n  const maxTurns = options.maxTurns ?? 12\n  const maxParallelSubagents = options.maxParallelSubagents ?? 2\n  const maxRuntimeChars = options.maxRuntimeChars ?? 6000\n  const functions = tools as unknown as NonNullable<Parameters<typeof agent>[1]>['functions']\n\n  const analyst = agent<{ question: string }, { answer: string; findings: string[] }>(\n    // `reasoning!` is an internal (Ax `!`) scratchpad field: generated first to\n    // force reason-before-conclude, stripped from the returned output — so the\n    // consumed shape stays { answer, findings }. Brings the trace-analyst to the\n    // same prose-first CoT ordering the kind-factory gets from its `report` field.\n    'question:string -> reasoning!:string, answer:string, findings:string[]',\n    {\n      agentIdentity: {\n        name: 'TraceAnalyst',\n        description:\n          'Analyzes OTLP-shaped JSONL traces using bounded discovery tools to identify systemic failure modes.',\n      },\n      contextFields: ['question'],\n      runtime: new AxJSRuntime({\n        permissions: [],\n        blockDynamicImport: true,\n        allowedModules: [],\n        freezeIntrinsics: true,\n        blockShadowRealm: true,\n        // RLM stdout mode relies on runtime bindings persisting across turns.\n        preventGlobalThisExtensions: false,\n      }),\n      mode: maxDepth > 0 ? 'advanced' : 'simple',\n      recursionOptions: maxDepth > 0 ? { maxDepth } : undefined,\n      maxTurns,\n      maxRuntimeChars,\n      maxBatchedLlmQueryConcurrency: maxParallelSubagents,\n      promptLevel: 'detailed',\n      // Trace analysis depends on exact prior tool results and runtime variables.\n      contextPolicy: { preset: 'full', budget: 'balanced' },\n      functions,\n      actorOptions: {\n        description: options.actorDescription ?? TRACE_ANALYST_ACTOR_DESCRIPTION,\n        ...(options.model ? { model: options.model } : {}),\n        // Keep actor messages tool-call/content shaped across reasoning models.\n        showThoughts: false,\n        thinkingTokenBudget: 'none',\n      },\n      responderOptions: {\n        ...(options.model ? { model: options.model } : {}),\n        description: options.subagentDescription ?? TRACE_ANALYST_SUBAGENT_DESCRIPTION,\n        showThoughts: false,\n      },\n      actorTurnCallback,\n      bubbleErrors: [TraceFileMissingError],\n    },\n  )\n\n  let result: { answer: unknown; findings: unknown }\n  try {\n    result = await analyst.forward(options.ai, { question: input.question })\n  } finally {\n    if (progressFs) {\n      await new Promise<void>((resolve) => progressFs!.end(() => resolve()))\n    }\n  }\n\n  return {\n    answer: typeof result.answer === 'string' ? result.answer : String(result.answer ?? ''),\n    findings: Array.isArray(result.findings)\n      ? result.findings.filter((s): s is string => typeof s === 'string')\n      : [],\n    turns,\n    turnCount: turns.length,\n    usage: normalizeRoleArrays(analyst.getUsage()),\n    chatLog: normalizeRoleArrays(analyst.getChatLog()),\n    actorPromptVersion: TRACE_ANALYST_ACTOR_DESCRIPTION_VERSION,\n  }\n}\n\nfunction normalizeRoleArrays(value: unknown): {\n  actor: Record<string, unknown>[]\n  responder: Record<string, unknown>[]\n} {\n  const record = value && typeof value === 'object' ? (value as Record<string, unknown>) : {}\n  return {\n    actor: normalizeRecordArray(record.actor),\n    responder: normalizeRecordArray(record.responder),\n  }\n}\n\nfunction normalizeRecordArray(value: unknown): Record<string, unknown>[] {\n  if (!Array.isArray(value)) return []\n  return value.map((item) =>\n    item && typeof item === 'object' ? { ...(item as Record<string, unknown>) } : { value: item },\n  )\n}\n"],"mappings":";;;;;;;AAEO,IAAM,kCAAkC;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AA8DxC,IAAM,0CAA0C;AAGhD,IAAM,qCAAqC;AAAA;AAAA;;;ACnElD,SAA8D,aAAa,aAAa;AAoGxF,eAAsB,cACpB,OACA,SAC8B;AAC9B,MAAI,CAAC,MAAM,YAAY,OAAO,MAAM,aAAa,UAAU;AACzD,UAAM,IAAI,UAAU,0DAA0D;AAAA,EAChF;AAEA,QAAM,QACJ,OAAO,QAAQ,WAAW,WACtB,IAAI,mBAAmB,EAAE,MAAM,QAAQ,OAAO,CAAC,IAC/C,QAAQ;AAGd,MAAI,iBAAiB,oBAAoB;AACvC,UAAM,MAAM,cAAc;AAAA,EAC5B;AAEA,QAAM,QAAsB,uBAAuB,EAAE,MAAM,CAAC;AAC5D,QAAM,QAAqC,CAAC;AAG5C,MAAI;AACJ,MAAI,QAAQ,iBAAiB;AAC3B,UAAM,EAAE,kBAAkB,IAAI,MAAM,OAAO,IAAS;AACpD,UAAM,EAAE,MAAM,IAAI,MAAM,OAAO,aAAkB;AACjD,UAAM,EAAE,QAAQ,IAAI,MAAM,OAAO,MAAW;AAC5C,UAAM,MAAM,QAAQ,QAAQ,eAAe,GAAG,EAAE,WAAW,KAAK,CAAC;AACjE,iBAAa,kBAAkB,QAAQ,iBAAiB,EAAE,OAAO,IAAI,CAAC;AAAA,EACxE;AAEA,QAAM,oBAAoB,OAAO,SAAqC;AACpE,UAAM,OAAkC;AAAA,MACtC,MAAM,KAAK;AAAA,MACX,SAAS,KAAK;AAAA,MACd,MAAM,KAAK;AAAA,MACX,QAAQ,KAAK;AAAA,MACb,SAAS,KAAK;AAAA,IAChB;AACA,UAAM,KAAK,IAAI;AACf,QAAI,YAAY;AACd,UAAI;AACF,mBAAW,MAAM,GAAG,KAAK,UAAU,EAAE,GAAG,MAAM,IAAI,KAAK,IAAI,EAAE,CAAC,CAAC;AAAA,CAAI;AAAA,MACrE,QAAQ;AAAA,MAER;AAAA,IACF;AACA,QAAI,QAAQ,OAAQ,OAAM,QAAQ,OAAO,IAAI;AAAA,EAC/C;AAEA,QAAM,WAAW,QAAQ,YAAY;AACrC,QAAM,WAAW,QAAQ,YAAY;AACrC,QAAM,uBAAuB,QAAQ,wBAAwB;AAC7D,QAAM,kBAAkB,QAAQ,mBAAmB;AACnD,QAAM,YAAY;AAElB,QAAM,UAAU;AAAA;AAAA;AAAA;AAAA;AAAA,IAKd;AAAA,IACA;AAAA,MACE,eAAe;AAAA,QACb,MAAM;AAAA,QACN,aACE;AAAA,MACJ;AAAA,MACA,eAAe,CAAC,UAAU;AAAA,MAC1B,SAAS,IAAI,YAAY;AAAA,QACvB,aAAa,CAAC;AAAA,QACd,oBAAoB;AAAA,QACpB,gBAAgB,CAAC;AAAA,QACjB,kBAAkB;AAAA,QAClB,kBAAkB;AAAA;AAAA,QAElB,6BAA6B;AAAA,MAC/B,CAAC;AAAA,MACD,MAAM,WAAW,IAAI,aAAa;AAAA,MAClC,kBAAkB,WAAW,IAAI,EAAE,SAAS,IAAI;AAAA,MAChD;AAAA,MACA;AAAA,MACA,+BAA+B;AAAA,MAC/B,aAAa;AAAA;AAAA,MAEb,eAAe,EAAE,QAAQ,QAAQ,QAAQ,WAAW;AAAA,MACpD;AAAA,MACA,cAAc;AAAA,QACZ,aAAa,QAAQ,oBAAoB;AAAA,QACzC,GAAI,QAAQ,QAAQ,EAAE,OAAO,QAAQ,MAAM,IAAI,CAAC;AAAA;AAAA,QAEhD,cAAc;AAAA,QACd,qBAAqB;AAAA,MACvB;AAAA,MACA,kBAAkB;AAAA,QAChB,GAAI,QAAQ,QAAQ,EAAE,OAAO,QAAQ,MAAM,IAAI,CAAC;AAAA,QAChD,aAAa,QAAQ,uBAAuB;AAAA,QAC5C,cAAc;AAAA,MAChB;AAAA,MACA;AAAA,MACA,cAAc,CAAC,qBAAqB;AAAA,IACtC;AAAA,EACF;AAEA,MAAI;AACJ,MAAI;AACF,aAAS,MAAM,QAAQ,QAAQ,QAAQ,IAAI,EAAE,UAAU,MAAM,SAAS,CAAC;AAAA,EACzE,UAAE;AACA,QAAI,YAAY;AACd,YAAM,IAAI,QAAc,CAAC,YAAY,WAAY,IAAI,MAAM,QAAQ,CAAC,CAAC;AAAA,IACvE;AAAA,EACF;AAEA,SAAO;AAAA,IACL,QAAQ,OAAO,OAAO,WAAW,WAAW,OAAO,SAAS,OAAO,OAAO,UAAU,EAAE;AAAA,IACtF,UAAU,MAAM,QAAQ,OAAO,QAAQ,IACnC,OAAO,SAAS,OAAO,CAAC,MAAmB,OAAO,MAAM,QAAQ,IAChE,CAAC;AAAA,IACL;AAAA,IACA,WAAW,MAAM;AAAA,IACjB,OAAO,oBAAoB,QAAQ,SAAS,CAAC;AAAA,IAC7C,SAAS,oBAAoB,QAAQ,WAAW,CAAC;AAAA,IACjD,oBAAoB;AAAA,EACtB;AACF;AAEA,SAAS,oBAAoB,OAG3B;AACA,QAAM,SAAS,SAAS,OAAO,UAAU,WAAY,QAAoC,CAAC;AAC1F,SAAO;AAAA,IACL,OAAO,qBAAqB,OAAO,KAAK;AAAA,IACxC,WAAW,qBAAqB,OAAO,SAAS;AAAA,EAClD;AACF;AAEA,SAAS,qBAAqB,OAA2C;AACvE,MAAI,CAAC,MAAM,QAAQ,KAAK,EAAG,QAAO,CAAC;AACnC,SAAO,MAAM;AAAA,IAAI,CAAC,SAChB,QAAQ,OAAO,SAAS,WAAW,EAAE,GAAI,KAAiC,IAAI,EAAE,OAAO,KAAK;AAAA,EAC9F;AACF;","names":[]}

package/dist/{chunk-OKQ2LAT7.js → chunk-4LWD6GC7.js}

@@ -12,13 +12,15 @@ import {
   FileSystemRawProviderSink
 } from "./chunk-PC4UYEBM.js";
 import {
-  buildAgentProfileCell,
-  validateRunRecord,
-  verifyAgentProfileCell
-} from "./chunk-LO6IOIJ2.js";
+  validateRunRecord
+} from "./chunk-G6S73VA7.js";
 import {
   TraceEmitter
 } from "./chunk-TVVP3ZZQ.js";
+import {
+  buildAgentProfileCell,
+  verifyAgentProfileCell
+} from "./chunk-ABOIVNXL.js";
 import {
   canonicalize,
   hashJson
@@ -352,4 +354,4 @@ function defaultRunId(params) {
 export {
   runEvalCampaign
 };
-//# sourceMappingURL=chunk-OKQ2LAT7.js.map
+//# sourceMappingURL=chunk-4LWD6GC7.js.map

package/dist/{chunk-OKQ2LAT7.js.map → chunk-4LWD6GC7.js.map}

@@ -1 +1 @@
-{"version":3,"sources":["../src/eval-campaign.ts"],"sourcesContent":["/**\n * EvalCampaign — opinionated matrix runner that wires the four\n * capture-integrity directives by construction.\n *\n * The canonical benchmark shape — matrix runner → for each\n * (variant, scenario, seed) → start a TraceEmitter → call LLMs → end the\n * run → analyze — has a bug class at the integration boundary: raw\n * events not captured, route silently wrong, integrity not asserted,\n * analyst never run. The directives in `SKILL.md § Capture integrity`\n * are the mitigations.\n *\n * `EvalCampaign` is the structural fix — consumers don't wire the\n * integrity surface themselves; the campaign owns it. Specifically:\n *\n *   - calls `assertLlmRoute` once at preflight before any work runs\n *   - constructs a per-run `TraceStore` and `RawProviderSink` via factories\n *   - constructs the `TraceEmitter` with `onRunComplete: [analyst hook]`\n *   - hands the runner an `LlmClientOptions` pre-wired with the sink and\n *     trace context — the runner can't accidentally call an LLM without\n *     capturing the raw HTTP envelope\n *   - calls `assertRunCaptured` after every `endRun` and routes failures\n *     through a configurable policy (`throw` / `mark_failed` / `log`)\n *   - assembles per-run `RunRecord`s and runs `researchReport` at the end\n *     so the campaign artifact is launch-decision-grade by default\n *   - embeds the campaign fingerprint (a SHA-256 over the canonicalised\n *     run set) and optional `preregistrationHash` in the report\n *\n * The runner contract is intentionally narrow: produce a `CampaignRunOutcome`\n * given a fully-wired `CampaignRunContext`. Everything orchestration-shaped\n * lives in the campaign. This is the inversion-of-control point — consumers\n * stop writing matrix runners and start writing scenario-runners.\n *\n * Out of scope for v1 (tracked in `docs/research-report-methodology.md`):\n *\n *   - Distributed/cluster execution (concurrency is local async)\n *   - Adaptive sampling / sequential interim looks\n *   - Resume from partial state across crashes\n *   - LLM-call retry beyond what `LlmClient` already does\n */\n\nimport {\n  type AgentProfileCell,\n  type AgentProfileCellInput,\n  buildAgentProfileCell,\n  verifyAgentProfileCell,\n} from './agent-profile-cell'\nimport { assertLlmRoute, type LlmClientOptions, type LlmRouteRequirements } from './llm-client'\nimport { canonicalize, hashJson } from './pre-registration'\nimport type {\n  JudgeScoresRecord,\n  RunJudgeMetadata,\n  RunOutcome,\n  RunRecord,\n  RunSplitTag,\n  RunTokenUsage,\n} from './run-record'\nimport { validateRunRecord } from './run-record'\nimport { type ResearchReport, type ResearchReportOptions, researchReport } from './summary-report'\nimport type { RunCompleteHook } from './trace/emitter'\nimport { TraceEmitter } from './trace/emitter'\nimport {\n  assertRunCaptured,\n  RunIntegrityError,\n  type RunIntegrityExpectations,\n  type RunIntegrityReport,\n} from './trace/integrity'\nimport { FileSystemRawProviderSink, type RawProviderSink } from './trace/raw-provider-sink'\nimport type { FailureClass } from './trace/schema'\nimport type { TraceStore } from './trace/store'\n\n// ── Public types ─────────────────────────────────────────────────────────\n\nexport interface CampaignVariant<V> {\n  id: string\n  payload: V\n}\n\nexport interface CampaignScenario {\n  scenarioId: string\n  /** Free-form metadata propagated to runs and reports. */\n  tags?: Record<string, string>\n}\n\nexport interface CampaignRunContext<V> {\n  /** Stable run id. The campaign generates this; the runner does not. */\n  runId: string\n  /** Logical experiment id (campaignId by default; overridable per-run via opts). */\n  experimentId: string\n  variant: V\n  variantId: string\n  scenarioId: string\n  scenarioTags: Record<string, string>\n  seed: number\n  splitTag: RunSplitTag\n  /**\n   * The TraceEmitter for this run, with `onRunComplete` hooks pre-wired\n   * (analyst auto-execution if configured, plus integrity check). The\n   * runner MUST call `emitter.startRun` before doing any work and either\n   * `emitter.endRun` or `emitter.abortRun` before returning.\n   */\n  emitter: TraceEmitter\n  store: TraceStore\n  rawSink: RawProviderSink\n  /**\n   * Pre-wired LLM client options — `rawSink` and `traceContext` are populated\n   * so any `callLlm(req, ctx.llmOpts)` automatically captures raw HTTP. The\n   * runner can spread additional fields if needed.\n   */\n  llmOpts: LlmClientOptions\n}\n\nexport interface CampaignRunOutcome {\n  /** Did the run pass? Mirrors `RunOutcome.pass` semantics. */\n  pass: boolean\n  /** Score for the run on its split. Maps to `searchScore` or `holdoutScore`. */\n  score: number\n  /** Mandatory cost in USD. Use 0 + raw.cost_unknown=1 only if truly unknown. */\n  costUsd: number\n  tokenUsage: RunTokenUsage\n  /** Snapshot model id (e.g. `claude-sonnet-4-6@2025-04-15`). */\n  model: string\n  /** sha256 of the effective prompt sent to the model. */\n  promptHash: string\n  /** sha256 of the effective config (model, temperature, tools, judges, splits). */\n  configHash: string\n  /** Optional extra numeric metrics to land in `outcome.raw`. */\n  raw?: Record<string, number>\n  /** Canonical cross-agent failure class from the shared `FAILURE_CLASSES`\n   *  taxonomy. Propagated to `RunRecord.failureClass` so campaign runs\n   *  aggregate failures in the same vocabulary as every other producer. */\n  failureClass?: FailureClass\n  /** Optional free-form failure detail, scoped under `failureClass`. */\n  failureMode?: string\n  /** Optional judge metadata when a judge was used. */\n  judgeMetadata?: RunJudgeMetadata\n  /**\n   * Optional per-judge / per-dim breakdown for ensemble-judged runs.\n   * Propagated to `outcome.judgeScores` on the resulting `RunRecord`.\n   * Single-judge or scalar-only runs leave this unset.\n   */\n  judgeScores?: JudgeScoresRecord\n  /**\n   * Agent profile cell observed by the runner. When supplied, it overrides\n   * `EvalCampaignOptions.agentProfile` for this run and must match the\n   * outcome's `model` and `promptHash`.\n   */\n  agentProfile?: AgentProfileCell | AgentProfileCellInput\n}\n\nexport type CampaignRunner<V> = (ctx: CampaignRunContext<V>) => Promise<CampaignRunOutcome>\n\nexport type CampaignIntegrityPolicy = 'throw' | 'mark_failed' | 'log'\n\nexport interface EvalCampaignOptions<V> {\n  /**\n   * Stable id for the campaign. Used as the default `experimentId` on\n   * every run, and folded into the campaign fingerprint.\n   */\n  campaignId: string\n  variants: CampaignVariant<V>[]\n  scenarios: CampaignScenario[]\n  /** Default `[0, 1, 2]`. */\n  seeds?: number[]\n  /** Default `'holdout'` — the split that anchors a launch decision. */\n  splitTag?: RunSplitTag\n  /** Git SHA the campaign is run against. Mandatory; `RunRecord` rejects unset. */\n  commitSha: string\n  /**\n   * LLM client config. Augmented per-run with `rawSink` and `traceContext`\n   * before being passed to the runner. The campaign asserts this config\n   * matches `routeRequirements` once at preflight.\n   */\n  llmOpts: LlmClientOptions\n  /**\n   * Default `{ requireExplicitBaseUrl: true, requireAuth: true }` — fail\n   * loud if the campaign would silently fall back to the public router or\n   * run unauthenticated. Override with an empty object to disable.\n   */\n  routeRequirements?: LlmRouteRequirements\n  /**\n   * Per-run TraceStore factory. Common shape: a fresh store per run keyed\n   * on `runId`. Implementations that share a store across the campaign\n   * are valid — the campaign only writes through `emitter`.\n   */\n  storeFactory: (params: CampaignFactoryParams) => TraceStore\n  /**\n   * Per-run RawProviderSink factory. Defaults to `FileSystemRawProviderSink`\n   * rooted at `${workDir}/raw-events/${runId}` if `workDir` is supplied;\n   * otherwise required. Forensic capture is non-negotiable in a campaign\n   * run — pass `NoopRawProviderSink` explicitly if you want to opt out.\n   */\n  rawSinkFactory?: (params: CampaignFactoryParams) => RawProviderSink\n  /**\n   * Filesystem root for default `rawSinkFactory`. Ignored if\n   * `rawSinkFactory` is supplied.\n   */\n  workDir?: string\n  /**\n   * Extra `onRunComplete` hooks the campaign appends (after its own\n   * integrity-check hook). Pass `traceAnalystOnRunComplete(...)` here.\n   */\n  onRunComplete?: RunCompleteHook[]\n  /**\n   * Per-run integrity expectations. Defaults to:\n   *   `{ llmSpansMin: 1, requireRawCoverageOfLlmSpans: true, requireOutcome: true }`.\n   * Override (e.g. `{ llmSpansMin: 0 }`) for runs that don't call LLMs.\n   */\n  integrity?: RunIntegrityExpectations\n  /** Behaviour when integrity fails. Default `'mark_failed'`. */\n  onIntegrityFailure?: CampaignIntegrityPolicy\n  /**\n   * Per-run runner. Receives a fully-wired context; produces an outcome\n   * the campaign converts into a `RunRecord`.\n   */\n  runner: CampaignRunner<V>\n  /**\n   * If set, the campaign computes `researchReport` at the end. `comparator`\n   * is a `variantId`. Other fields are forwarded verbatim.\n   */\n  report?: { comparator?: string } & Omit<\n    ResearchReportOptions,\n    'comparator' | 'preregistrationHash' | 'generatedAt'\n  >\n  /**\n   * Hash of a signed `HypothesisManifest` (see `pre-registration.ts`).\n   * Embedded in the campaign fingerprint and the research report.\n   */\n  preregistrationHash?: string\n  /** Local concurrency. Default `1` (sequential). */\n  concurrency?: number\n  /**\n   * Override the time source. Tests pass a mock to make wallMs deterministic.\n   */\n  now?: () => number\n  /** Override the runId generator. Tests pin this. */\n  runId?: (params: CampaignFactoryParams) => string\n  /**\n   * Agent profile cell for campaign runs. Static profiles can pass an object;\n   * routers or variant-specific harnesses can pass a factory. The campaign\n   * stamps the built cell onto every `RunRecord` and rejects profile/model or\n   * profile/prompt contradictions.\n   */\n  agentProfile?:\n    | AgentProfileCell\n    | AgentProfileCellInput\n    | ((\n        params: CampaignFactoryParams & {\n          variant: V\n          scenarioTags: Record<string, string>\n        },\n      ) =>\n        | AgentProfileCell\n        | AgentProfileCellInput\n        | Promise<AgentProfileCell | AgentProfileCellInput>)\n}\n\nexport interface CampaignFactoryParams {\n  campaignId: string\n  runId: string\n  variantId: string\n  scenarioId: string\n  seed: number\n}\n\nexport interface FailedRun {\n  runId: string\n  variantId: string\n  scenarioId: string\n  seed: number\n  reason: string\n  error?: string\n}\n\nexport interface EvalCampaignResult {\n  campaignId: string\n  /** SHA-256 over canonicalised `(variantIds, scenarioIds, seeds, comparator, splitTag, baseUrl, provider, preregistrationHash)`. */\n  campaignFingerprint: string\n  preregistrationHash: string | null\n  /** Successful runs only. Failed runs land in `failedRuns`. */\n  runs: RunRecord[]\n  /** Integrity reports for every successful run. */\n  integrityReports: RunIntegrityReport[]\n  failedRuns: FailedRun[]\n  /** Computed when `report` is set on options. */\n  report?: ResearchReport\n  startedAt: string\n  endedAt: string\n}\n\n// ── Implementation ───────────────────────────────────────────────────────\n\nconst DEFAULT_INTEGRITY: RunIntegrityExpectations = {\n  llmSpansMin: 1,\n  requireRawCoverageOfLlmSpans: true,\n  requireOutcome: true,\n}\n\nconst DEFAULT_ROUTE: LlmRouteRequirements = {\n  requireExplicitBaseUrl: true,\n  requireAuth: true,\n}\n\nexport async function runEvalCampaign<V>(\n  opts: EvalCampaignOptions<V>,\n): Promise<EvalCampaignResult> {\n  // ── Preflight ──────────────────────────────────────────────────────\n  assertLlmRoute(opts.llmOpts, opts.routeRequirements ?? DEFAULT_ROUTE)\n\n  if (opts.variants.length === 0) {\n    throw new Error('runEvalCampaign: variants must be non-empty.')\n  }\n  if (opts.scenarios.length === 0) {\n    throw new Error('runEvalCampaign: scenarios must be non-empty.')\n  }\n  const variantIds = new Set<string>()\n  for (const v of opts.variants) {\n    if (variantIds.has(v.id)) {\n      throw new Error(`runEvalCampaign: duplicate variant id \"${v.id}\".`)\n    }\n    variantIds.add(v.id)\n  }\n  const scenarioIds = new Set<string>()\n  for (const s of opts.scenarios) {\n    if (scenarioIds.has(s.scenarioId)) {\n      throw new Error(`runEvalCampaign: duplicate scenarioId \"${s.scenarioId}\".`)\n    }\n    scenarioIds.add(s.scenarioId)\n  }\n  if (opts.report?.comparator && !variantIds.has(opts.report.comparator)) {\n    throw new Error(\n      `runEvalCampaign: report.comparator \"${opts.report.comparator}\" is not a configured variantId.`,\n    )\n  }\n  if (!opts.commitSha) {\n    throw new Error('runEvalCampaign: commitSha is required (every RunRecord needs it).')\n  }\n\n  const seeds = opts.seeds ?? [0, 1, 2]\n  const splitTag: RunSplitTag = opts.splitTag ?? 'holdout'\n  const concurrency = Math.max(1, opts.concurrency ?? 1)\n  const integrity = { ...DEFAULT_INTEGRITY, ...(opts.integrity ?? {}) }\n  const onIntegrityFailure: CampaignIntegrityPolicy = opts.onIntegrityFailure ?? 'mark_failed'\n  const now = opts.now ?? (() => Date.now())\n  const baseUrl = (opts.llmOpts.baseUrl ?? '').replace(/\\/+$/, '')\n  const provider = opts.llmOpts.provider ?? null\n  const preregistrationHash = opts.preregistrationHash ?? null\n\n  const rawSinkFactory = opts.rawSinkFactory ?? defaultRawSinkFactory(opts.workDir)\n\n  // ── Fingerprint ────────────────────────────────────────────────────\n  const campaignFingerprint = await hashJson(\n    canonicalize({\n      campaignId: opts.campaignId,\n      variants: opts.variants.map((v) => v.id).sort(),\n      scenarios: opts.scenarios.map((s) => s.scenarioId).sort(),\n      seeds: [...seeds].sort((a, b) => a - b),\n      splitTag,\n      comparator: opts.report?.comparator ?? null,\n      baseUrl,\n      provider,\n      preregistrationHash,\n    }),\n  )\n\n  // ── Plan the matrix ────────────────────────────────────────────────\n  type Cell = { variant: CampaignVariant<V>; scenario: CampaignScenario; seed: number }\n  const cells: Cell[] = []\n  for (const variant of opts.variants) {\n    for (const scenario of opts.scenarios) {\n      for (const seed of seeds) {\n        cells.push({ variant, scenario, seed })\n      }\n    }\n  }\n\n  const startedAt = new Date(now()).toISOString()\n  const runs: RunRecord[] = []\n  const integrityReports: RunIntegrityReport[] = []\n  const failedRuns: FailedRun[] = []\n\n  // ── Execute (bounded-concurrency worker pool) ──────────────────────\n  // A genuine (non-CellExecutionError) error from any worker is a bug, not a\n  // run-level failure. We must NOT let `Promise.all` reject mid-flight and\n  // orphan the other workers' in-progress runs (emitters never finalized,\n  // sinks/handles leak, partial work silently discarded). Instead: capture the\n  // first genuine error, stop dispatching new cells so in-flight workers wind\n  // down, finalize every still-open run, then re-throw the aggregated error.\n  let cursor = 0\n  let aborting = false\n  const genuineErrors: unknown[] = []\n  // Emitters for runs that are currently mid-flight, keyed by runId. A worker\n  // registers its emitter before invoking the runner and removes it once the\n  // run is finalized (via endRun/abortRun, success or failure). Anything left\n  // here after the pool settles is an orphan we must abort.\n  const openRuns = new Map<string, TraceEmitter>()\n\n  async function worker(): Promise<void> {\n    while (!aborting) {\n      const i = cursor++\n      if (i >= cells.length) return\n      const cell = cells[i]!\n      try {\n        const result = await runOneCell(cell)\n        runs.push(result.record)\n        integrityReports.push(result.integrity)\n      } catch (err) {\n        if (err instanceof CellExecutionError) {\n          failedRuns.push(err.failed)\n          if (err.integrity) integrityReports.push(err.integrity)\n        } else {\n          // Genuine bug — not a runner failure, not an integrity failure.\n          // Capture it and stop dispatching so peers wind down gracefully;\n          // re-thrown after the pool settles. Do not surface here — that would\n          // reject Promise.all and orphan the other workers' runs.\n          genuineErrors.push(err)\n          aborting = true\n          return\n        }\n      }\n    }\n  }\n\n  async function runOneCell(\n    cell: Cell,\n  ): Promise<{ record: RunRecord; integrity: RunIntegrityReport }> {\n    const runId = (opts.runId ?? defaultRunId)({\n      campaignId: opts.campaignId,\n      runId: '', // unused by default generator\n      variantId: cell.variant.id,\n      scenarioId: cell.scenario.scenarioId,\n      seed: cell.seed,\n    })\n    const factoryParams: CampaignFactoryParams = {\n      campaignId: opts.campaignId,\n      runId,\n      variantId: cell.variant.id,\n      scenarioId: cell.scenario.scenarioId,\n      seed: cell.seed,\n    }\n    const store = opts.storeFactory(factoryParams)\n    const rawSink = rawSinkFactory(factoryParams)\n\n    const emitter = new TraceEmitter(store, {\n      runId,\n      now: opts.now,\n      onRunComplete: opts.onRunComplete,\n    })\n    // Track this run as open so a genuine error elsewhere in the pool can\n    // finalize it instead of orphaning it. Removed in the finally below.\n    openRuns.set(runId, emitter)\n\n    const llmOpts: LlmClientOptions = {\n      ...opts.llmOpts,\n      rawSink,\n      traceContext: { runId },\n    }\n\n    const ctx: CampaignRunContext<V> = {\n      runId,\n      experimentId: opts.campaignId,\n      variant: cell.variant.payload,\n      variantId: cell.variant.id,\n      scenarioId: cell.scenario.scenarioId,\n      scenarioTags: cell.scenario.tags ?? {},\n      seed: cell.seed,\n      splitTag,\n      emitter,\n      store,\n      rawSink,\n      llmOpts,\n    }\n\n    try {\n      const wallStart = now()\n      let outcome: CampaignRunOutcome\n      try {\n        outcome = await opts.runner(ctx)\n      } catch (err) {\n        const message = err instanceof Error ? err.message : String(err)\n        // The runner threw mid-execution. Abort the run so the emitter\n        // finalizes. The only benign abortRun failure is \"the runner never\n        // started the run\" (nothing to finalize). A store-write failure (disk\n        // full, FS error) is a genuine diagnostic — surface it rather than\n        // masking it as a plain runner failure.\n        await finalizeAbort(emitter, runId, message)\n        throw new CellExecutionError({\n          runId,\n          variantId: cell.variant.id,\n          scenarioId: cell.scenario.scenarioId,\n          seed: cell.seed,\n          reason: 'runner_threw',\n          error: message,\n        })\n      }\n      const wallMs = now() - wallStart\n\n      const integrityReport = await assertRunCaptured(store, runId, { ...integrity, rawSink })\n      if (!integrityReport.ok) {\n        switch (onIntegrityFailure) {\n          case 'throw':\n            throw new RunIntegrityError(integrityReport)\n          case 'mark_failed':\n            throw new CellExecutionError(\n              {\n                runId,\n                variantId: cell.variant.id,\n                scenarioId: cell.scenario.scenarioId,\n                seed: cell.seed,\n                reason: 'integrity_failed',\n                error: integrityReport.issues.map((i) => i.code).join(', '),\n              },\n              integrityReport,\n            )\n          case 'log':\n            // Caller wants the run admitted with a flagged report; fall through.\n            break\n        }\n      }\n\n      const recordOutcome: RunOutcome = {\n        raw: outcome.raw ?? {},\n      }\n      if (splitTag === 'holdout') recordOutcome.holdoutScore = outcome.score\n      else recordOutcome.searchScore = outcome.score\n      if (outcome.judgeScores !== undefined) recordOutcome.judgeScores = outcome.judgeScores\n\n      const record: RunRecord = {\n        runId,\n        experimentId: opts.campaignId,\n        candidateId: cell.variant.id,\n        seed: cell.seed,\n        model: outcome.model,\n        promptHash: outcome.promptHash,\n        configHash: outcome.configHash,\n        commitSha: opts.commitSha,\n        wallMs,\n        costUsd: outcome.costUsd,\n        tokenUsage: outcome.tokenUsage,\n        judgeMetadata: outcome.judgeMetadata,\n        outcome: recordOutcome,\n        ...(outcome.failureClass ? { failureClass: outcome.failureClass } : {}),\n        failureMode: outcome.failureMode,\n        splitTag,\n        scenarioId: cell.scenario.scenarioId,\n      }\n      const profileSource =\n        outcome.agentProfile ??\n        (typeof opts.agentProfile === 'function'\n          ? await opts.agentProfile({\n              campaignId: opts.campaignId,\n              runId,\n              variantId: cell.variant.id,\n              scenarioId: cell.scenario.scenarioId,\n              seed: cell.seed,\n              variant: cell.variant.payload,\n              scenarioTags: cell.scenario.tags ?? {},\n            })\n          : opts.agentProfile)\n      if (profileSource !== undefined) {\n        const agentProfile = await resolveAgentProfileCell(profileSource)\n        assertAgentProfileMatchesRun(agentProfile, outcome.model, outcome.promptHash)\n        record.agentProfile = agentProfile\n      }\n      return { record: validateRunRecord(record), integrity: integrityReport }\n    } finally {\n      // This run's worker has finished with it (success, run-level failure, or\n      // genuine error). It is no longer the pool's job to finalize — drop it so\n      // the post-settle sweep doesn't double-abort a finalized run.\n      openRuns.delete(runId)\n    }\n  }\n\n  const workers = Array.from({ length: Math.min(concurrency, cells.length) }, () => worker())\n  // allSettled (not all): a genuine error in one worker must not reject the\n  // pool mid-flight and orphan the others. Each worker captures its own\n  // genuine error into `genuineErrors` and returns; we re-throw below.\n  await Promise.allSettled(workers)\n\n  // Finalize any run still open after the pool wound down. With the\n  // stop-dispatch flag these are the runs that were mid-flight in peer workers\n  // when the first genuine error fired — abort them so their emitters finalize\n  // (hooks fire, store records a terminal status) instead of leaking.\n  for (const [runId, emitter] of openRuns) {\n    await finalizeAbort(emitter, runId, 'campaign aborted: genuine error in a sibling run')\n  }\n  openRuns.clear()\n\n  if (genuineErrors.length > 0) {\n    throw genuineErrors.length === 1\n      ? genuineErrors[0]\n      : new AggregateError(\n          genuineErrors,\n          `runEvalCampaign: ${genuineErrors.length} runs failed with genuine (non-run-level) errors`,\n        )\n  }\n\n  // ── Optional research report ───────────────────────────────────────\n  let report: ResearchReport | undefined\n  if (opts.report) {\n    const reportOpts: ResearchReportOptions = {\n      ...opts.report,\n      comparator: opts.report.comparator,\n      split: splitTag === 'dev' ? 'search' : splitTag,\n      generatedAt: new Date(now()).toISOString(),\n      preregistrationHash: preregistrationHash ?? undefined,\n    }\n    report = await researchReport(runs, reportOpts)\n  }\n\n  const endedAt = new Date(now()).toISOString()\n\n  return {\n    campaignId: opts.campaignId,\n    campaignFingerprint,\n    preregistrationHash,\n    runs,\n    integrityReports,\n    failedRuns,\n    report,\n    startedAt,\n    endedAt,\n  }\n}\n\n// ── Internal ─────────────────────────────────────────────────────────────\n\nclass CellExecutionError extends Error {\n  readonly failed: FailedRun\n  readonly integrity?: RunIntegrityReport\n  constructor(failed: FailedRun, integrity?: RunIntegrityReport) {\n    super(`cell ${failed.variantId}/${failed.scenarioId}@${failed.seed} failed: ${failed.reason}`)\n    this.failed = failed\n    this.integrity = integrity\n  }\n}\n\n/**\n * Abort a run whose owning work threw or was orphaned by a sibling's genuine\n * error, finalizing the emitter so hooks fire and the store records a terminal\n * status. Safe to call unconditionally: it only aborts runs that are still\n * `running`. Two no-op cases are intentional and benign:\n *\n *   - the run was never started (absent from the store) — nothing to finalize\n *   - the run is already terminal (`completed` / `failed` / `aborted`) —\n *     finalizing again would overwrite the real outcome (e.g. flip a passed run\n *     to `aborted` with `{ pass: false, notes: reason }`), so we leave it alone\n *\n * Any OTHER failure of the store read or the abort write (disk full, FS fault,\n * backend down) is a genuine diagnostic and propagates rather than being\n * swallowed.\n */\nexport async function finalizeAbort(\n  emitter: TraceEmitter,\n  runId: string,\n  reason: string,\n): Promise<void> {\n  const existing = await emitter.traceStore.getRun(runId)\n  if (existing === undefined) return // run never started; nothing to abort\n  if (existing.status !== 'running') return // already finalized; never overwrite a real outcome\n  await emitter.abortRun(reason)\n}\n\nfunction defaultRawSinkFactory(workDir: string | undefined) {\n  return (params: CampaignFactoryParams): RawProviderSink => {\n    if (!workDir) {\n      throw new Error(\n        'runEvalCampaign: rawSinkFactory not supplied and workDir not set. Pass either to enable raw provider capture, or pass `new NoopRawProviderSink()` via rawSinkFactory to opt out explicitly.',\n      )\n    }\n    return new FileSystemRawProviderSink({\n      dir: `${workDir}/raw-events/${params.runId}`,\n    })\n  }\n}\n\nasync function resolveAgentProfileCell(\n  input: AgentProfileCell | AgentProfileCellInput,\n): Promise<AgentProfileCell> {\n  if (isAgentProfileCell(input)) {\n    if (!(await verifyAgentProfileCell(input))) {\n      throw new Error(`runEvalCampaign: agentProfile.cellId does not match its content`)\n    }\n    return input\n  }\n  return buildAgentProfileCell(input)\n}\n\nfunction isAgentProfileCell(\n  input: AgentProfileCell | AgentProfileCellInput,\n): input is AgentProfileCell {\n  return 'schemaVersion' in input && 'cellId' in input\n}\n\nfunction assertAgentProfileMatchesRun(\n  profile: AgentProfileCell,\n  model: string,\n  promptHash: string,\n): void {\n  if (profile.model !== undefined && profile.model !== model) {\n    throw new Error(\n      `runEvalCampaign: agentProfile.model \"${profile.model}\" does not match outcome.model \"${model}\"`,\n    )\n  }\n  if (profile.promptHash !== undefined && profile.promptHash !== promptHash) {\n    throw new Error(\n      `runEvalCampaign: agentProfile.promptHash \"${profile.promptHash}\" does not match outcome.promptHash \"${promptHash}\"`,\n    )\n  }\n}\n\nfunction defaultRunId(params: CampaignFactoryParams): string {\n  // Stable across re-runs: fingerprint of (campaignId, variantId, scenarioId, seed).\n  // Caller can override via opts.runId for non-deterministic IDs.\n  const base = `${params.campaignId}::${params.variantId}::${params.scenarioId}::${params.seed}`\n  // Lightweight hex: we don't need crypto-grade here, just stability + uniqueness.\n  let h1 = 0x811c9dc5\n  let h2 = 0x12345678\n  for (let i = 0; i < base.length; i++) {\n    const c = base.charCodeAt(i)\n    h1 = Math.imul(h1 ^ c, 0x01000193) >>> 0\n    h2 = Math.imul(h2 ^ c, 0x9e3779b1) >>> 0\n  }\n  return `run-${h1.toString(16).padStart(8, '0')}${h2.toString(16).padStart(8, '0')}`\n}\n"],"mappings":";;;;;;;;;;;;;;;;;;;;;;;;;;;AAmSA,IAAM,oBAA8C;AAAA,EAClD,aAAa;AAAA,EACb,8BAA8B;AAAA,EAC9B,gBAAgB;AAClB;AAEA,IAAM,gBAAsC;AAAA,EAC1C,wBAAwB;AAAA,EACxB,aAAa;AACf;AAEA,eAAsB,gBACpB,MAC6B;AAE7B,iBAAe,KAAK,SAAS,KAAK,qBAAqB,aAAa;AAEpE,MAAI,KAAK,SAAS,WAAW,GAAG;AAC9B,UAAM,IAAI,MAAM,8CAA8C;AAAA,EAChE;AACA,MAAI,KAAK,UAAU,WAAW,GAAG;AAC/B,UAAM,IAAI,MAAM,+CAA+C;AAAA,EACjE;AACA,QAAM,aAAa,oBAAI,IAAY;AACnC,aAAW,KAAK,KAAK,UAAU;AAC7B,QAAI,WAAW,IAAI,EAAE,EAAE,GAAG;AACxB,YAAM,IAAI,MAAM,0CAA0C,EAAE,EAAE,IAAI;AAAA,IACpE;AACA,eAAW,IAAI,EAAE,EAAE;AAAA,EACrB;AACA,QAAM,cAAc,oBAAI,IAAY;AACpC,aAAW,KAAK,KAAK,WAAW;AAC9B,QAAI,YAAY,IAAI,EAAE,UAAU,GAAG;AACjC,YAAM,IAAI,MAAM,0CAA0C,EAAE,UAAU,IAAI;AAAA,IAC5E;AACA,gBAAY,IAAI,EAAE,UAAU;AAAA,EAC9B;AACA,MAAI,KAAK,QAAQ,cAAc,CAAC,WAAW,IAAI,KAAK,OAAO,UAAU,GAAG;AACtE,UAAM,IAAI;AAAA,MACR,uCAAuC,KAAK,OAAO,UAAU;AAAA,IAC/D;AAAA,EACF;AACA,MAAI,CAAC,KAAK,WAAW;AACnB,UAAM,IAAI,MAAM,oEAAoE;AAAA,EACtF;AAEA,QAAM,QAAQ,KAAK,SAAS,CAAC,GAAG,GAAG,CAAC;AACpC,QAAM,WAAwB,KAAK,YAAY;AAC/C,QAAM,cAAc,KAAK,IAAI,GAAG,KAAK,eAAe,CAAC;AACrD,QAAM,YAAY,EAAE,GAAG,mBAAmB,GAAI,KAAK,aAAa,CAAC,EAAG;AACpE,QAAM,qBAA8C,KAAK,sBAAsB;AAC/E,QAAM,MAAM,KAAK,QAAQ,MAAM,KAAK,IAAI;AACxC,QAAM,WAAW,KAAK,QAAQ,WAAW,IAAI,QAAQ,QAAQ,EAAE;AAC/D,QAAM,WAAW,KAAK,QAAQ,YAAY;AAC1C,QAAM,sBAAsB,KAAK,uBAAuB;AAExD,QAAM,iBAAiB,KAAK,kBAAkB,sBAAsB,KAAK,OAAO;AAGhF,QAAM,sBAAsB,MAAM;AAAA,IAChC,aAAa;AAAA,MACX,YAAY,KAAK;AAAA,MACjB,UAAU,KAAK,SAAS,IAAI,CAAC,MAAM,EAAE,EAAE,EAAE,KAAK;AAAA,MAC9C,WAAW,KAAK,UAAU,IAAI,CAAC,MAAM,EAAE,UAAU,EAAE,KAAK;AAAA,MACxD,OAAO,CAAC,GAAG,KAAK,EAAE,KAAK,CAAC,GAAG,MAAM,IAAI,CAAC;AAAA,MACtC;AAAA,MACA,YAAY,KAAK,QAAQ,cAAc;AAAA,MACvC;AAAA,MACA;AAAA,MACA;AAAA,IACF,CAAC;AAAA,EACH;AAIA,QAAM,QAAgB,CAAC;AACvB,aAAW,WAAW,KAAK,UAAU;AACnC,eAAW,YAAY,KAAK,WAAW;AACrC,iBAAW,QAAQ,OAAO;AACxB,cAAM,KAAK,EAAE,SAAS,UAAU,KAAK,CAAC;AAAA,MACxC;AAAA,IACF;AAAA,EACF;AAEA,QAAM,YAAY,IAAI,KAAK,IAAI,CAAC,EAAE,YAAY;AAC9C,QAAM,OAAoB,CAAC;AAC3B,QAAM,mBAAyC,CAAC;AAChD,QAAM,aAA0B,CAAC;AASjC,MAAI,SAAS;AACb,MAAI,WAAW;AACf,QAAM,gBAA2B,CAAC;AAKlC,QAAM,WAAW,oBAAI,IAA0B;AAE/C,iBAAe,SAAwB;AACrC,WAAO,CAAC,UAAU;AAChB,YAAM,IAAI;AACV,UAAI,KAAK,MAAM,OAAQ;AACvB,YAAM,OAAO,MAAM,CAAC;AACpB,UAAI;AACF,cAAM,SAAS,MAAM,WAAW,IAAI;AACpC,aAAK,KAAK,OAAO,MAAM;AACvB,yBAAiB,KAAK,OAAO,SAAS;AAAA,MACxC,SAAS,KAAK;AACZ,YAAI,eAAe,oBAAoB;AACrC,qBAAW,KAAK,IAAI,MAAM;AAC1B,cAAI,IAAI,UAAW,kBAAiB,KAAK,IAAI,SAAS;AAAA,QACxD,OAAO;AAKL,wBAAc,KAAK,GAAG;AACtB,qBAAW;AACX;AAAA,QACF;AAAA,MACF;AAAA,IACF;AAAA,EACF;AAEA,iBAAe,WACb,MAC+D;AAC/D,UAAM,SAAS,KAAK,SAAS,cAAc;AAAA,MACzC,YAAY,KAAK;AAAA,MACjB,OAAO;AAAA;AAAA,MACP,WAAW,KAAK,QAAQ;AAAA,MACxB,YAAY,KAAK,SAAS;AAAA,MAC1B,MAAM,KAAK;AAAA,IACb,CAAC;AACD,UAAM,gBAAuC;AAAA,MAC3C,YAAY,KAAK;AAAA,MACjB;AAAA,MACA,WAAW,KAAK,QAAQ;AAAA,MACxB,YAAY,KAAK,SAAS;AAAA,MAC1B,MAAM,KAAK;AAAA,IACb;AACA,UAAM,QAAQ,KAAK,aAAa,aAAa;AAC7C,UAAM,UAAU,eAAe,aAAa;AAE5C,UAAM,UAAU,IAAI,aAAa,OAAO;AAAA,MACtC;AAAA,MACA,KAAK,KAAK;AAAA,MACV,eAAe,KAAK;AAAA,IACtB,CAAC;AAGD,aAAS,IAAI,OAAO,OAAO;AAE3B,UAAM,UAA4B;AAAA,MAChC,GAAG,KAAK;AAAA,MACR;AAAA,MACA,cAAc,EAAE,MAAM;AAAA,IACxB;AAEA,UAAM,MAA6B;AAAA,MACjC;AAAA,MACA,cAAc,KAAK;AAAA,MACnB,SAAS,KAAK,QAAQ;AAAA,MACtB,WAAW,KAAK,QAAQ;AAAA,MACxB,YAAY,KAAK,SAAS;AAAA,MAC1B,cAAc,KAAK,SAAS,QAAQ,CAAC;AAAA,MACrC,MAAM,KAAK;AAAA,MACX;AAAA,MACA;AAAA,MACA;AAAA,MACA;AAAA,MACA;AAAA,IACF;AAEA,QAAI;AACF,YAAM,YAAY,IAAI;AACtB,UAAI;AACJ,UAAI;AACF,kBAAU,MAAM,KAAK,OAAO,GAAG;AAAA,MACjC,SAAS,KAAK;AACZ,cAAM,UAAU,eAAe,QAAQ,IAAI,UAAU,OAAO,GAAG;AAM/D,cAAM,cAAc,SAAS,OAAO,OAAO;AAC3C,cAAM,IAAI,mBAAmB;AAAA,UAC3B;AAAA,UACA,WAAW,KAAK,QAAQ;AAAA,UACxB,YAAY,KAAK,SAAS;AAAA,UAC1B,MAAM,KAAK;AAAA,UACX,QAAQ;AAAA,UACR,OAAO;AAAA,QACT,CAAC;AAAA,MACH;AACA,YAAM,SAAS,IAAI,IAAI;AAEvB,YAAM,kBAAkB,MAAM,kBAAkB,OAAO,OAAO,EAAE,GAAG,WAAW,QAAQ,CAAC;AACvF,UAAI,CAAC,gBAAgB,IAAI;AACvB,gBAAQ,oBAAoB;AAAA,UAC1B,KAAK;AACH,kBAAM,IAAI,kBAAkB,eAAe;AAAA,UAC7C,KAAK;AACH,kBAAM,IAAI;AAAA,cACR;AAAA,gBACE;AAAA,gBACA,WAAW,KAAK,QAAQ;AAAA,gBACxB,YAAY,KAAK,SAAS;AAAA,gBAC1B,MAAM,KAAK;AAAA,gBACX,QAAQ;AAAA,gBACR,OAAO,gBAAgB,OAAO,IAAI,CAAC,MAAM,EAAE,IAAI,EAAE,KAAK,IAAI;AAAA,cAC5D;AAAA,cACA;AAAA,YACF;AAAA,UACF,KAAK;AAEH;AAAA,QACJ;AAAA,MACF;AAEA,YAAM,gBAA4B;AAAA,QAChC,KAAK,QAAQ,OAAO,CAAC;AAAA,MACvB;AACA,UAAI,aAAa,UAAW,eAAc,eAAe,QAAQ;AAAA,UAC5D,eAAc,cAAc,QAAQ;AACzC,UAAI,QAAQ,gBAAgB,OAAW,eAAc,cAAc,QAAQ;AAE3E,YAAM,SAAoB;AAAA,QACxB;AAAA,QACA,cAAc,KAAK;AAAA,QACnB,aAAa,KAAK,QAAQ;AAAA,QAC1B,MAAM,KAAK;AAAA,QACX,OAAO,QAAQ;AAAA,QACf,YAAY,QAAQ;AAAA,QACpB,YAAY,QAAQ;AAAA,QACpB,WAAW,KAAK;AAAA,QAChB;AAAA,QACA,SAAS,QAAQ;AAAA,QACjB,YAAY,QAAQ;AAAA,QACpB,eAAe,QAAQ;AAAA,QACvB,SAAS;AAAA,QACT,GAAI,QAAQ,eAAe,EAAE,cAAc,QAAQ,aAAa,IAAI,CAAC;AAAA,QACrE,aAAa,QAAQ;AAAA,QACrB;AAAA,QACA,YAAY,KAAK,SAAS;AAAA,MAC5B;AACA,YAAM,gBACJ,QAAQ,iBACP,OAAO,KAAK,iBAAiB,aAC1B,MAAM,KAAK,aAAa;AAAA,QACtB,YAAY,KAAK;AAAA,QACjB;AAAA,QACA,WAAW,KAAK,QAAQ;AAAA,QACxB,YAAY,KAAK,SAAS;AAAA,QAC1B,MAAM,KAAK;AAAA,QACX,SAAS,KAAK,QAAQ;AAAA,QACtB,cAAc,KAAK,SAAS,QAAQ,CAAC;AAAA,MACvC,CAAC,IACD,KAAK;AACX,UAAI,kBAAkB,QAAW;AAC/B,cAAM,eAAe,MAAM,wBAAwB,aAAa;AAChE,qCAA6B,cAAc,QAAQ,OAAO,QAAQ,UAAU;AAC5E,eAAO,eAAe;AAAA,MACxB;AACA,aAAO,EAAE,QAAQ,kBAAkB,MAAM,GAAG,WAAW,gBAAgB;AAAA,IACzE,UAAE;AAIA,eAAS,OAAO,KAAK;AAAA,IACvB;AAAA,EACF;AAEA,QAAM,UAAU,MAAM,KAAK,EAAE,QAAQ,KAAK,IAAI,aAAa,MAAM,MAAM,EAAE,GAAG,MAAM,OAAO,CAAC;AAI1F,QAAM,QAAQ,WAAW,OAAO;AAMhC,aAAW,CAAC,OAAO,OAAO,KAAK,UAAU;AACvC,UAAM,cAAc,SAAS,OAAO,kDAAkD;AAAA,EACxF;AACA,WAAS,MAAM;AAEf,MAAI,cAAc,SAAS,GAAG;AAC5B,UAAM,cAAc,WAAW,IAC3B,cAAc,CAAC,IACf,IAAI;AAAA,MACF;AAAA,MACA,oBAAoB,cAAc,MAAM;AAAA,IAC1C;AAAA,EACN;AAGA,MAAI;AACJ,MAAI,KAAK,QAAQ;AACf,UAAM,aAAoC;AAAA,MACxC,GAAG,KAAK;AAAA,MACR,YAAY,KAAK,OAAO;AAAA,MACxB,OAAO,aAAa,QAAQ,WAAW;AAAA,MACvC,aAAa,IAAI,KAAK,IAAI,CAAC,EAAE,YAAY;AAAA,MACzC,qBAAqB,uBAAuB;AAAA,IAC9C;AACA,aAAS,MAAM,eAAe,MAAM,UAAU;AAAA,EAChD;AAEA,QAAM,UAAU,IAAI,KAAK,IAAI,CAAC,EAAE,YAAY;AAE5C,SAAO;AAAA,IACL,YAAY,KAAK;AAAA,IACjB;AAAA,IACA;AAAA,IACA;AAAA,IACA;AAAA,IACA;AAAA,IACA;AAAA,IACA;AAAA,IACA;AAAA,EACF;AACF;AAIA,IAAM,qBAAN,cAAiC,MAAM;AAAA,EAC5B;AAAA,EACA;AAAA,EACT,YAAY,QAAmB,WAAgC;AAC7D,UAAM,QAAQ,OAAO,SAAS,IAAI,OAAO,UAAU,IAAI,OAAO,IAAI,YAAY,OAAO,MAAM,EAAE;AAC7F,SAAK,SAAS;AACd,SAAK,YAAY;AAAA,EACnB;AACF;AAiBA,eAAsB,cACpB,SACA,OACA,QACe;AACf,QAAM,WAAW,MAAM,QAAQ,WAAW,OAAO,KAAK;AACtD,MAAI,aAAa,OAAW;AAC5B,MAAI,SAAS,WAAW,UAAW;AACnC,QAAM,QAAQ,SAAS,MAAM;AAC/B;AAEA,SAAS,sBAAsB,SAA6B;AAC1D,SAAO,CAAC,WAAmD;AACzD,QAAI,CAAC,SAAS;AACZ,YAAM,IAAI;AAAA,QACR;AAAA,MACF;AAAA,IACF;AACA,WAAO,IAAI,0BAA0B;AAAA,MACnC,KAAK,GAAG,OAAO,eAAe,OAAO,KAAK;AAAA,IAC5C,CAAC;AAAA,EACH;AACF;AAEA,eAAe,wBACb,OAC2B;AAC3B,MAAI,mBAAmB,KAAK,GAAG;AAC7B,QAAI,CAAE,MAAM,uBAAuB,KAAK,GAAI;AAC1C,YAAM,IAAI,MAAM,iEAAiE;AAAA,IACnF;AACA,WAAO;AAAA,EACT;AACA,SAAO,sBAAsB,KAAK;AACpC;AAEA,SAAS,mBACP,OAC2B;AAC3B,SAAO,mBAAmB,SAAS,YAAY;AACjD;AAEA,SAAS,6BACP,SACA,OACA,YACM;AACN,MAAI,QAAQ,UAAU,UAAa,QAAQ,UAAU,OAAO;AAC1D,UAAM,IAAI;AAAA,MACR,wCAAwC,QAAQ,KAAK,mCAAmC,KAAK;AAAA,IAC/F;AAAA,EACF;AACA,MAAI,QAAQ,eAAe,UAAa,QAAQ,eAAe,YAAY;AACzE,UAAM,IAAI;AAAA,MACR,6CAA6C,QAAQ,UAAU,wCAAwC,UAAU;AAAA,IACnH;AAAA,EACF;AACF;AAEA,SAAS,aAAa,QAAuC;AAG3D,QAAM,OAAO,GAAG,OAAO,UAAU,KAAK,OAAO,SAAS,KAAK,OAAO,UAAU,KAAK,OAAO,IAAI;AAE5F,MAAI,KAAK;AACT,MAAI,KAAK;AACT,WAAS,IAAI,GAAG,IAAI,KAAK,QAAQ,KAAK;AACpC,UAAM,IAAI,KAAK,WAAW,CAAC;AAC3B,SAAK,KAAK,KAAK,KAAK,GAAG,QAAU,MAAM;AACvC,SAAK,KAAK,KAAK,KAAK,GAAG,UAAU,MAAM;AAAA,EACzC;AACA,SAAO,OAAO,GAAG,SAAS,EAAE,EAAE,SAAS,GAAG,GAAG,CAAC,GAAG,GAAG,SAAS,EAAE,EAAE,SAAS,GAAG,GAAG,CAAC;AACnF;","names":[]}
+{"version":3,"sources":["../src/eval-campaign.ts"],"sourcesContent":["/**\n * EvalCampaign — opinionated matrix runner that wires the four\n * capture-integrity directives by construction.\n *\n * The canonical benchmark shape — matrix runner → for each\n * (variant, scenario, seed) → start a TraceEmitter → call LLMs → end the\n * run → analyze — has a bug class at the integration boundary: raw\n * events not captured, route silently wrong, integrity not asserted,\n * analyst never run. The directives in `SKILL.md § Capture integrity`\n * are the mitigations.\n *\n * `EvalCampaign` is the structural fix — consumers don't wire the\n * integrity surface themselves; the campaign owns it. Specifically:\n *\n *   - calls `assertLlmRoute` once at preflight before any work runs\n *   - constructs a per-run `TraceStore` and `RawProviderSink` via factories\n *   - constructs the `TraceEmitter` with `onRunComplete: [analyst hook]`\n *   - hands the runner an `LlmClientOptions` pre-wired with the sink and\n *     trace context — the runner can't accidentally call an LLM without\n *     capturing the raw HTTP envelope\n *   - calls `assertRunCaptured` after every `endRun` and routes failures\n *     through a configurable policy (`throw` / `mark_failed` / `log`)\n *   - assembles per-run `RunRecord`s and runs `researchReport` at the end\n *     so the campaign artifact is launch-decision-grade by default\n *   - embeds the campaign fingerprint (a SHA-256 over the canonicalised\n *     run set) and optional `preregistrationHash` in the report\n *\n * The runner contract is intentionally narrow: produce a `CampaignRunOutcome`\n * given a fully-wired `CampaignRunContext`. Everything orchestration-shaped\n * lives in the campaign. This is the inversion-of-control point — consumers\n * stop writing matrix runners and start writing scenario-runners.\n *\n * Out of scope for v1 (tracked in `docs/research-report-methodology.md`):\n *\n *   - Distributed/cluster execution (concurrency is local async)\n *   - Adaptive sampling / sequential interim looks\n *   - Resume from partial state across crashes\n *   - LLM-call retry beyond what `LlmClient` already does\n */\n\nimport {\n  type AgentProfileCell,\n  type AgentProfileCellInput,\n  buildAgentProfileCell,\n  verifyAgentProfileCell,\n} from './agent-profile-cell'\nimport { assertLlmRoute, type LlmClientOptions, type LlmRouteRequirements } from './llm-client'\nimport { canonicalize, hashJson } from './pre-registration'\nimport type {\n  JudgeScoresRecord,\n  RunJudgeMetadata,\n  RunOutcome,\n  RunRecord,\n  RunSplitTag,\n  RunTokenUsage,\n} from './run-record'\nimport { validateRunRecord } from './run-record'\nimport { type ResearchReport, type ResearchReportOptions, researchReport } from './summary-report'\nimport type { RunCompleteHook } from './trace/emitter'\nimport { TraceEmitter } from './trace/emitter'\nimport {\n  assertRunCaptured,\n  RunIntegrityError,\n  type RunIntegrityExpectations,\n  type RunIntegrityReport,\n} from './trace/integrity'\nimport { FileSystemRawProviderSink, type RawProviderSink } from './trace/raw-provider-sink'\nimport type { FailureClass } from './trace/schema'\nimport type { TraceStore } from './trace/store'\n\n// ── Public types ─────────────────────────────────────────────────────────\n\nexport interface CampaignVariant<V> {\n  id: string\n  payload: V\n}\n\nexport interface CampaignScenario {\n  scenarioId: string\n  /** Free-form metadata propagated to runs and reports. */\n  tags?: Record<string, string>\n}\n\nexport interface CampaignRunContext<V> {\n  /** Stable run id. The campaign generates this; the runner does not. */\n  runId: string\n  /** Logical experiment id (campaignId by default; overridable per-run via opts). */\n  experimentId: string\n  variant: V\n  variantId: string\n  scenarioId: string\n  scenarioTags: Record<string, string>\n  seed: number\n  splitTag: RunSplitTag\n  /**\n   * The TraceEmitter for this run, with `onRunComplete` hooks pre-wired\n   * (analyst auto-execution if configured, plus integrity check). The\n   * runner MUST call `emitter.startRun` before doing any work and either\n   * `emitter.endRun` or `emitter.abortRun` before returning.\n   */\n  emitter: TraceEmitter\n  store: TraceStore\n  rawSink: RawProviderSink\n  /**\n   * Pre-wired LLM client options — `rawSink` and `traceContext` are populated\n   * so any `callLlm(req, ctx.llmOpts)` automatically captures raw HTTP. The\n   * runner can spread additional fields if needed.\n   */\n  llmOpts: LlmClientOptions\n}\n\nexport interface CampaignRunOutcome {\n  /** Did the run pass? Mirrors `RunOutcome.pass` semantics. */\n  pass: boolean\n  /** Score for the run on its split. Maps to `searchScore` or `holdoutScore`. */\n  score: number\n  /** Mandatory cost in USD. Use 0 + raw.cost_unknown=1 only if truly unknown. */\n  costUsd: number\n  tokenUsage: RunTokenUsage\n  /** Snapshot model id (e.g. `claude-sonnet-4-6@2025-04-15`). */\n  model: string\n  /** sha256 of the effective prompt sent to the model. */\n  promptHash: string\n  /** sha256 of the effective config (model, temperature, tools, judges, splits). */\n  configHash: string\n  /** Optional extra numeric metrics to land in `outcome.raw`. */\n  raw?: Record<string, number>\n  /** Canonical cross-agent failure class from the shared `FAILURE_CLASSES`\n   *  taxonomy. Propagated to `RunRecord.failureClass` so campaign runs\n   *  aggregate failures in the same vocabulary as every other producer. */\n  failureClass?: FailureClass\n  /** Optional free-form failure detail, scoped under `failureClass`. */\n  failureMode?: string\n  /** Optional judge metadata when a judge was used. */\n  judgeMetadata?: RunJudgeMetadata\n  /**\n   * Optional per-judge / per-dim breakdown for ensemble-judged runs.\n   * Propagated to `outcome.judgeScores` on the resulting `RunRecord`.\n   * Single-judge or scalar-only runs leave this unset.\n   */\n  judgeScores?: JudgeScoresRecord\n  /**\n   * Agent profile cell observed by the runner. When supplied, it overrides\n   * `EvalCampaignOptions.agentProfile` for this run and must match the\n   * outcome's `model` and `promptHash`.\n   */\n  agentProfile?: AgentProfileCell | AgentProfileCellInput\n}\n\nexport type CampaignRunner<V> = (ctx: CampaignRunContext<V>) => Promise<CampaignRunOutcome>\n\nexport type CampaignIntegrityPolicy = 'throw' | 'mark_failed' | 'log'\n\nexport interface EvalCampaignOptions<V> {\n  /**\n   * Stable id for the campaign. Used as the default `experimentId` on\n   * every run, and folded into the campaign fingerprint.\n   */\n  campaignId: string\n  variants: CampaignVariant<V>[]\n  scenarios: CampaignScenario[]\n  /** Default `[0, 1, 2]`. */\n  seeds?: number[]\n  /** Default `'holdout'` — the split that anchors a launch decision. */\n  splitTag?: RunSplitTag\n  /** Git SHA the campaign is run against. Mandatory; `RunRecord` rejects unset. */\n  commitSha: string\n  /**\n   * LLM client config. Augmented per-run with `rawSink` and `traceContext`\n   * before being passed to the runner. The campaign asserts this config\n   * matches `routeRequirements` once at preflight.\n   */\n  llmOpts: LlmClientOptions\n  /**\n   * Default `{ requireExplicitBaseUrl: true, requireAuth: true }` — fail\n   * loud if the campaign would silently fall back to the public router or\n   * run unauthenticated. Override with an empty object to disable.\n   */\n  routeRequirements?: LlmRouteRequirements\n  /**\n   * Per-run TraceStore factory. Common shape: a fresh store per run keyed\n   * on `runId`. Implementations that share a store across the campaign\n   * are valid — the campaign only writes through `emitter`.\n   */\n  storeFactory: (params: CampaignFactoryParams) => TraceStore\n  /**\n   * Per-run RawProviderSink factory. Defaults to `FileSystemRawProviderSink`\n   * rooted at `${workDir}/raw-events/${runId}` if `workDir` is supplied;\n   * otherwise required. Forensic capture is non-negotiable in a campaign\n   * run — pass `NoopRawProviderSink` explicitly if you want to opt out.\n   */\n  rawSinkFactory?: (params: CampaignFactoryParams) => RawProviderSink\n  /**\n   * Filesystem root for default `rawSinkFactory`. Ignored if\n   * `rawSinkFactory` is supplied.\n   */\n  workDir?: string\n  /**\n   * Extra `onRunComplete` hooks the campaign appends (after its own\n   * integrity-check hook). Pass `traceAnalystOnRunComplete(...)` here.\n   */\n  onRunComplete?: RunCompleteHook[]\n  /**\n   * Per-run integrity expectations. Defaults to:\n   *   `{ llmSpansMin: 1, requireRawCoverageOfLlmSpans: true, requireOutcome: true }`.\n   * Override (e.g. `{ llmSpansMin: 0 }`) for runs that don't call LLMs.\n   */\n  integrity?: RunIntegrityExpectations\n  /** Behaviour when integrity fails. Default `'mark_failed'`. */\n  onIntegrityFailure?: CampaignIntegrityPolicy\n  /**\n   * Per-run runner. Receives a fully-wired context; produces an outcome\n   * the campaign converts into a `RunRecord`.\n   */\n  runner: CampaignRunner<V>\n  /**\n   * If set, the campaign computes `researchReport` at the end. `comparator`\n   * is a `variantId`. Other fields are forwarded verbatim.\n   */\n  report?: { comparator?: string } & Omit<\n    ResearchReportOptions,\n    'comparator' | 'preregistrationHash' | 'generatedAt'\n  >\n  /**\n   * Hash of a signed `HypothesisManifest` (see `pre-registration.ts`).\n   * Embedded in the campaign fingerprint and the research report.\n   */\n  preregistrationHash?: string\n  /** Local concurrency. Default `1` (sequential). */\n  concurrency?: number\n  /**\n   * Override the time source. Tests pass a mock to make wallMs deterministic.\n   */\n  now?: () => number\n  /** Override the runId generator. Tests pin this. */\n  runId?: (params: CampaignFactoryParams) => string\n  /**\n   * Agent profile cell for campaign runs. Static profiles can pass an object;\n   * routers or variant-specific harnesses can pass a factory. The campaign\n   * stamps the built cell onto every `RunRecord` and rejects profile/model or\n   * profile/prompt contradictions.\n   */\n  agentProfile?:\n    | AgentProfileCell\n    | AgentProfileCellInput\n    | ((\n        params: CampaignFactoryParams & {\n          variant: V\n          scenarioTags: Record<string, string>\n        },\n      ) =>\n        | AgentProfileCell\n        | AgentProfileCellInput\n        | Promise<AgentProfileCell | AgentProfileCellInput>)\n}\n\nexport interface CampaignFactoryParams {\n  campaignId: string\n  runId: string\n  variantId: string\n  scenarioId: string\n  seed: number\n}\n\nexport interface FailedRun {\n  runId: string\n  variantId: string\n  scenarioId: string\n  seed: number\n  reason: string\n  error?: string\n}\n\nexport interface EvalCampaignResult {\n  campaignId: string\n  /** SHA-256 over canonicalised `(variantIds, scenarioIds, seeds, comparator, splitTag, baseUrl, provider, preregistrationHash)`. */\n  campaignFingerprint: string\n  preregistrationHash: string | null\n  /** Successful runs only. Failed runs land in `failedRuns`. */\n  runs: RunRecord[]\n  /** Integrity reports for every successful run. */\n  integrityReports: RunIntegrityReport[]\n  failedRuns: FailedRun[]\n  /** Computed when `report` is set on options. */\n  report?: ResearchReport\n  startedAt: string\n  endedAt: string\n}\n\n// ── Implementation ───────────────────────────────────────────────────────\n\nconst DEFAULT_INTEGRITY: RunIntegrityExpectations = {\n  llmSpansMin: 1,\n  requireRawCoverageOfLlmSpans: true,\n  requireOutcome: true,\n}\n\nconst DEFAULT_ROUTE: LlmRouteRequirements = {\n  requireExplicitBaseUrl: true,\n  requireAuth: true,\n}\n\nexport async function runEvalCampaign<V>(\n  opts: EvalCampaignOptions<V>,\n): Promise<EvalCampaignResult> {\n  // ── Preflight ──────────────────────────────────────────────────────\n  assertLlmRoute(opts.llmOpts, opts.routeRequirements ?? DEFAULT_ROUTE)\n\n  if (opts.variants.length === 0) {\n    throw new Error('runEvalCampaign: variants must be non-empty.')\n  }\n  if (opts.scenarios.length === 0) {\n    throw new Error('runEvalCampaign: scenarios must be non-empty.')\n  }\n  const variantIds = new Set<string>()\n  for (const v of opts.variants) {\n    if (variantIds.has(v.id)) {\n      throw new Error(`runEvalCampaign: duplicate variant id \"${v.id}\".`)\n    }\n    variantIds.add(v.id)\n  }\n  const scenarioIds = new Set<string>()\n  for (const s of opts.scenarios) {\n    if (scenarioIds.has(s.scenarioId)) {\n      throw new Error(`runEvalCampaign: duplicate scenarioId \"${s.scenarioId}\".`)\n    }\n    scenarioIds.add(s.scenarioId)\n  }\n  if (opts.report?.comparator && !variantIds.has(opts.report.comparator)) {\n    throw new Error(\n      `runEvalCampaign: report.comparator \"${opts.report.comparator}\" is not a configured variantId.`,\n    )\n  }\n  if (!opts.commitSha) {\n    throw new Error('runEvalCampaign: commitSha is required (every RunRecord needs it).')\n  }\n\n  const seeds = opts.seeds ?? [0, 1, 2]\n  const splitTag: RunSplitTag = opts.splitTag ?? 'holdout'\n  const concurrency = Math.max(1, opts.concurrency ?? 1)\n  const integrity = { ...DEFAULT_INTEGRITY, ...(opts.integrity ?? {}) }\n  const onIntegrityFailure: CampaignIntegrityPolicy = opts.onIntegrityFailure ?? 'mark_failed'\n  const now = opts.now ?? (() => Date.now())\n  const baseUrl = (opts.llmOpts.baseUrl ?? '').replace(/\\/+$/, '')\n  const provider = opts.llmOpts.provider ?? null\n  const preregistrationHash = opts.preregistrationHash ?? null\n\n  const rawSinkFactory = opts.rawSinkFactory ?? defaultRawSinkFactory(opts.workDir)\n\n  // ── Fingerprint ────────────────────────────────────────────────────\n  const campaignFingerprint = await hashJson(\n    canonicalize({\n      campaignId: opts.campaignId,\n      variants: opts.variants.map((v) => v.id).sort(),\n      scenarios: opts.scenarios.map((s) => s.scenarioId).sort(),\n      seeds: [...seeds].sort((a, b) => a - b),\n      splitTag,\n      comparator: opts.report?.comparator ?? null,\n      baseUrl,\n      provider,\n      preregistrationHash,\n    }),\n  )\n\n  // ── Plan the matrix ────────────────────────────────────────────────\n  type Cell = { variant: CampaignVariant<V>; scenario: CampaignScenario; seed: number }\n  const cells: Cell[] = []\n  for (const variant of opts.variants) {\n    for (const scenario of opts.scenarios) {\n      for (const seed of seeds) {\n        cells.push({ variant, scenario, seed })\n      }\n    }\n  }\n\n  const startedAt = new Date(now()).toISOString()\n  const runs: RunRecord[] = []\n  const integrityReports: RunIntegrityReport[] = []\n  const failedRuns: FailedRun[] = []\n\n  // ── Execute (bounded-concurrency worker pool) ──────────────────────\n  // A genuine (non-CellExecutionError) error from any worker is a bug, not a\n  // run-level failure. We must NOT let `Promise.all` reject mid-flight and\n  // orphan the other workers' in-progress runs (emitters never finalized,\n  // sinks/handles leak, partial work silently discarded). Instead: capture the\n  // first genuine error, stop dispatching new cells so in-flight workers wind\n  // down, finalize every still-open run, then re-throw the aggregated error.\n  let cursor = 0\n  let aborting = false\n  const genuineErrors: unknown[] = []\n  // Emitters for runs that are currently mid-flight, keyed by runId. A worker\n  // registers its emitter before invoking the runner and removes it once the\n  // run is finalized (via endRun/abortRun, success or failure). Anything left\n  // here after the pool settles is an orphan we must abort.\n  const openRuns = new Map<string, TraceEmitter>()\n\n  async function worker(): Promise<void> {\n    while (!aborting) {\n      const i = cursor++\n      if (i >= cells.length) return\n      const cell = cells[i]!\n      try {\n        const result = await runOneCell(cell)\n        runs.push(result.record)\n        integrityReports.push(result.integrity)\n      } catch (err) {\n        if (err instanceof CellExecutionError) {\n          failedRuns.push(err.failed)\n          if (err.integrity) integrityReports.push(err.integrity)\n        } else {\n          // Genuine bug — not a runner failure, not an integrity failure.\n          // Capture it and stop dispatching so peers wind down gracefully;\n          // re-thrown after the pool settles. Do not surface here — that would\n          // reject Promise.all and orphan the other workers' runs.\n          genuineErrors.push(err)\n          aborting = true\n          return\n        }\n      }\n    }\n  }\n\n  async function runOneCell(\n    cell: Cell,\n  ): Promise<{ record: RunRecord; integrity: RunIntegrityReport }> {\n    const runId = (opts.runId ?? defaultRunId)({\n      campaignId: opts.campaignId,\n      runId: '', // unused by default generator\n      variantId: cell.variant.id,\n      scenarioId: cell.scenario.scenarioId,\n      seed: cell.seed,\n    })\n    const factoryParams: CampaignFactoryParams = {\n      campaignId: opts.campaignId,\n      runId,\n      variantId: cell.variant.id,\n      scenarioId: cell.scenario.scenarioId,\n      seed: cell.seed,\n    }\n    const store = opts.storeFactory(factoryParams)\n    const rawSink = rawSinkFactory(factoryParams)\n\n    const emitter = new TraceEmitter(store, {\n      runId,\n      now: opts.now,\n      onRunComplete: opts.onRunComplete,\n    })\n    // Track this run as open so a genuine error elsewhere in the pool can\n    // finalize it instead of orphaning it. Removed in the finally below.\n    openRuns.set(runId, emitter)\n\n    const llmOpts: LlmClientOptions = {\n      ...opts.llmOpts,\n      rawSink,\n      traceContext: { runId },\n    }\n\n    const ctx: CampaignRunContext<V> = {\n      runId,\n      experimentId: opts.campaignId,\n      variant: cell.variant.payload,\n      variantId: cell.variant.id,\n      scenarioId: cell.scenario.scenarioId,\n      scenarioTags: cell.scenario.tags ?? {},\n      seed: cell.seed,\n      splitTag,\n      emitter,\n      store,\n      rawSink,\n      llmOpts,\n    }\n\n    try {\n      const wallStart = now()\n      let outcome: CampaignRunOutcome\n      try {\n        outcome = await opts.runner(ctx)\n      } catch (err) {\n        const message = err instanceof Error ? err.message : String(err)\n        // The runner threw mid-execution. Abort the run so the emitter\n        // finalizes. The only benign abortRun failure is \"the runner never\n        // started the run\" (nothing to finalize). A store-write failure (disk\n        // full, FS error) is a genuine diagnostic — surface it rather than\n        // masking it as a plain runner failure.\n        await finalizeAbort(emitter, runId, message)\n        throw new CellExecutionError({\n          runId,\n          variantId: cell.variant.id,\n          scenarioId: cell.scenario.scenarioId,\n          seed: cell.seed,\n          reason: 'runner_threw',\n          error: message,\n        })\n      }\n      const wallMs = now() - wallStart\n\n      const integrityReport = await assertRunCaptured(store, runId, { ...integrity, rawSink })\n      if (!integrityReport.ok) {\n        switch (onIntegrityFailure) {\n          case 'throw':\n            throw new RunIntegrityError(integrityReport)\n          case 'mark_failed':\n            throw new CellExecutionError(\n              {\n                runId,\n                variantId: cell.variant.id,\n                scenarioId: cell.scenario.scenarioId,\n                seed: cell.seed,\n                reason: 'integrity_failed',\n                error: integrityReport.issues.map((i) => i.code).join(', '),\n              },\n              integrityReport,\n            )\n          case 'log':\n            // Caller wants the run admitted with a flagged report; fall through.\n            break\n        }\n      }\n\n      const recordOutcome: RunOutcome = {\n        raw: outcome.raw ?? {},\n      }\n      if (splitTag === 'holdout') recordOutcome.holdoutScore = outcome.score\n      else recordOutcome.searchScore = outcome.score\n      if (outcome.judgeScores !== undefined) recordOutcome.judgeScores = outcome.judgeScores\n\n      const record: RunRecord = {\n        runId,\n        experimentId: opts.campaignId,\n        candidateId: cell.variant.id,\n        seed: cell.seed,\n        model: outcome.model,\n        promptHash: outcome.promptHash,\n        configHash: outcome.configHash,\n        commitSha: opts.commitSha,\n        wallMs,\n        costUsd: outcome.costUsd,\n        tokenUsage: outcome.tokenUsage,\n        judgeMetadata: outcome.judgeMetadata,\n        outcome: recordOutcome,\n        ...(outcome.failureClass ? { failureClass: outcome.failureClass } : {}),\n        failureMode: outcome.failureMode,\n        splitTag,\n        scenarioId: cell.scenario.scenarioId,\n      }\n      const profileSource =\n        outcome.agentProfile ??\n        (typeof opts.agentProfile === 'function'\n          ? await opts.agentProfile({\n              campaignId: opts.campaignId,\n              runId,\n              variantId: cell.variant.id,\n              scenarioId: cell.scenario.scenarioId,\n              seed: cell.seed,\n              variant: cell.variant.payload,\n              scenarioTags: cell.scenario.tags ?? {},\n            })\n          : opts.agentProfile)\n      if (profileSource !== undefined) {\n        const agentProfile = await resolveAgentProfileCell(profileSource)\n        assertAgentProfileMatchesRun(agentProfile, outcome.model, outcome.promptHash)\n        record.agentProfile = agentProfile\n      }\n      return { record: validateRunRecord(record), integrity: integrityReport }\n    } finally {\n      // This run's worker has finished with it (success, run-level failure, or\n      // genuine error). It is no longer the pool's job to finalize — drop it so\n      // the post-settle sweep doesn't double-abort a finalized run.\n      openRuns.delete(runId)\n    }\n  }\n\n  const workers = Array.from({ length: Math.min(concurrency, cells.length) }, () => worker())\n  // allSettled (not all): a genuine error in one worker must not reject the\n  // pool mid-flight and orphan the others. Each worker captures its own\n  // genuine error into `genuineErrors` and returns; we re-throw below.\n  await Promise.allSettled(workers)\n\n  // Finalize any run still open after the pool wound down. With the\n  // stop-dispatch flag these are the runs that were mid-flight in peer workers\n  // when the first genuine error fired — abort them so their emitters finalize\n  // (hooks fire, store records a terminal status) instead of leaking.\n  for (const [runId, emitter] of openRuns) {\n    await finalizeAbort(emitter, runId, 'campaign aborted: genuine error in a sibling run')\n  }\n  openRuns.clear()\n\n  if (genuineErrors.length > 0) {\n    throw genuineErrors.length === 1\n      ? genuineErrors[0]\n      : new AggregateError(\n          genuineErrors,\n          `runEvalCampaign: ${genuineErrors.length} runs failed with genuine (non-run-level) errors`,\n        )\n  }\n\n  // ── Optional research report ───────────────────────────────────────\n  let report: ResearchReport | undefined\n  if (opts.report) {\n    const reportOpts: ResearchReportOptions = {\n      ...opts.report,\n      comparator: opts.report.comparator,\n      split: splitTag === 'dev' ? 'search' : splitTag,\n      generatedAt: new Date(now()).toISOString(),\n      preregistrationHash: preregistrationHash ?? undefined,\n    }\n    report = await researchReport(runs, reportOpts)\n  }\n\n  const endedAt = new Date(now()).toISOString()\n\n  return {\n    campaignId: opts.campaignId,\n    campaignFingerprint,\n    preregistrationHash,\n    runs,\n    integrityReports,\n    failedRuns,\n    report,\n    startedAt,\n    endedAt,\n  }\n}\n\n// ── Internal ─────────────────────────────────────────────────────────────\n\nclass CellExecutionError extends Error {\n  readonly failed: FailedRun\n  readonly integrity?: RunIntegrityReport\n  constructor(failed: FailedRun, integrity?: RunIntegrityReport) {\n    super(`cell ${failed.variantId}/${failed.scenarioId}@${failed.seed} failed: ${failed.reason}`)\n    this.failed = failed\n    this.integrity = integrity\n  }\n}\n\n/**\n * Abort a run whose owning work threw or was orphaned by a sibling's genuine\n * error, finalizing the emitter so hooks fire and the store records a terminal\n * status. Safe to call unconditionally: it only aborts runs that are still\n * `running`. Two no-op cases are intentional and benign:\n *\n *   - the run was never started (absent from the store) — nothing to finalize\n *   - the run is already terminal (`completed` / `failed` / `aborted`) —\n *     finalizing again would overwrite the real outcome (e.g. flip a passed run\n *     to `aborted` with `{ pass: false, notes: reason }`), so we leave it alone\n *\n * Any OTHER failure of the store read or the abort write (disk full, FS fault,\n * backend down) is a genuine diagnostic and propagates rather than being\n * swallowed.\n */\nexport async function finalizeAbort(\n  emitter: TraceEmitter,\n  runId: string,\n  reason: string,\n): Promise<void> {\n  const existing = await emitter.traceStore.getRun(runId)\n  if (existing === undefined) return // run never started; nothing to abort\n  if (existing.status !== 'running') return // already finalized; never overwrite a real outcome\n  await emitter.abortRun(reason)\n}\n\nfunction defaultRawSinkFactory(workDir: string | undefined) {\n  return (params: CampaignFactoryParams): RawProviderSink => {\n    if (!workDir) {\n      throw new Error(\n        'runEvalCampaign: rawSinkFactory not supplied and workDir not set. Pass either to enable raw provider capture, or pass `new NoopRawProviderSink()` via rawSinkFactory to opt out explicitly.',\n      )\n    }\n    return new FileSystemRawProviderSink({\n      dir: `${workDir}/raw-events/${params.runId}`,\n    })\n  }\n}\n\nasync function resolveAgentProfileCell(\n  input: AgentProfileCell | AgentProfileCellInput,\n): Promise<AgentProfileCell> {\n  if (isAgentProfileCell(input)) {\n    if (!(await verifyAgentProfileCell(input))) {\n      throw new Error(`runEvalCampaign: agentProfile.cellId does not match its content`)\n    }\n    return input\n  }\n  return buildAgentProfileCell(input)\n}\n\nfunction isAgentProfileCell(\n  input: AgentProfileCell | AgentProfileCellInput,\n): input is AgentProfileCell {\n  return 'schemaVersion' in input && 'cellId' in input\n}\n\nfunction assertAgentProfileMatchesRun(\n  profile: AgentProfileCell,\n  model: string,\n  promptHash: string,\n): void {\n  if (profile.model !== undefined && profile.model !== model) {\n    throw new Error(\n      `runEvalCampaign: agentProfile.model \"${profile.model}\" does not match outcome.model \"${model}\"`,\n    )\n  }\n  if (profile.promptHash !== undefined && profile.promptHash !== promptHash) {\n    throw new Error(\n      `runEvalCampaign: agentProfile.promptHash \"${profile.promptHash}\" does not match outcome.promptHash \"${promptHash}\"`,\n    )\n  }\n}\n\nfunction defaultRunId(params: CampaignFactoryParams): string {\n  // Stable across re-runs: fingerprint of (campaignId, variantId, scenarioId, seed).\n  // Caller can override via opts.runId for non-deterministic IDs.\n  const base = `${params.campaignId}::${params.variantId}::${params.scenarioId}::${params.seed}`\n  // Lightweight hex: we don't need crypto-grade here, just stability + uniqueness.\n  let h1 = 0x811c9dc5\n  let h2 = 0x12345678\n  for (let i = 0; i < base.length; i++) {\n    const c = base.charCodeAt(i)\n    h1 = Math.imul(h1 ^ c, 0x01000193) >>> 0\n    h2 = Math.imul(h2 ^ c, 0x9e3779b1) >>> 0\n  }\n  return `run-${h1.toString(16).padStart(8, '0')}${h2.toString(16).padStart(8, '0')}`\n}\n"],"mappings":";;;;;;;;;;;;;;;;;;;;;;;;;;;;;AAmSA,IAAM,oBAA8C;AAAA,EAClD,aAAa;AAAA,EACb,8BAA8B;AAAA,EAC9B,gBAAgB;AAClB;AAEA,IAAM,gBAAsC;AAAA,EAC1C,wBAAwB;AAAA,EACxB,aAAa;AACf;AAEA,eAAsB,gBACpB,MAC6B;AAE7B,iBAAe,KAAK,SAAS,KAAK,qBAAqB,aAAa;AAEpE,MAAI,KAAK,SAAS,WAAW,GAAG;AAC9B,UAAM,IAAI,MAAM,8CAA8C;AAAA,EAChE;AACA,MAAI,KAAK,UAAU,WAAW,GAAG;AAC/B,UAAM,IAAI,MAAM,+CAA+C;AAAA,EACjE;AACA,QAAM,aAAa,oBAAI,IAAY;AACnC,aAAW,KAAK,KAAK,UAAU;AAC7B,QAAI,WAAW,IAAI,EAAE,EAAE,GAAG;AACxB,YAAM,IAAI,MAAM,0CAA0C,EAAE,EAAE,IAAI;AAAA,IACpE;AACA,eAAW,IAAI,EAAE,EAAE;AAAA,EACrB;AACA,QAAM,cAAc,oBAAI,IAAY;AACpC,aAAW,KAAK,KAAK,WAAW;AAC9B,QAAI,YAAY,IAAI,EAAE,UAAU,GAAG;AACjC,YAAM,IAAI,MAAM,0CAA0C,EAAE,UAAU,IAAI;AAAA,IAC5E;AACA,gBAAY,IAAI,EAAE,UAAU;AAAA,EAC9B;AACA,MAAI,KAAK,QAAQ,cAAc,CAAC,WAAW,IAAI,KAAK,OAAO,UAAU,GAAG;AACtE,UAAM,IAAI;AAAA,MACR,uCAAuC,KAAK,OAAO,UAAU;AAAA,IAC/D;AAAA,EACF;AACA,MAAI,CAAC,KAAK,WAAW;AACnB,UAAM,IAAI,MAAM,oEAAoE;AAAA,EACtF;AAEA,QAAM,QAAQ,KAAK,SAAS,CAAC,GAAG,GAAG,CAAC;AACpC,QAAM,WAAwB,KAAK,YAAY;AAC/C,QAAM,cAAc,KAAK,IAAI,GAAG,KAAK,eAAe,CAAC;AACrD,QAAM,YAAY,EAAE,GAAG,mBAAmB,GAAI,KAAK,aAAa,CAAC,EAAG;AACpE,QAAM,qBAA8C,KAAK,sBAAsB;AAC/E,QAAM,MAAM,KAAK,QAAQ,MAAM,KAAK,IAAI;AACxC,QAAM,WAAW,KAAK,QAAQ,WAAW,IAAI,QAAQ,QAAQ,EAAE;AAC/D,QAAM,WAAW,KAAK,QAAQ,YAAY;AAC1C,QAAM,sBAAsB,KAAK,uBAAuB;AAExD,QAAM,iBAAiB,KAAK,kBAAkB,sBAAsB,KAAK,OAAO;AAGhF,QAAM,sBAAsB,MAAM;AAAA,IAChC,aAAa;AAAA,MACX,YAAY,KAAK;AAAA,MACjB,UAAU,KAAK,SAAS,IAAI,CAAC,MAAM,EAAE,EAAE,EAAE,KAAK;AAAA,MAC9C,WAAW,KAAK,UAAU,IAAI,CAAC,MAAM,EAAE,UAAU,EAAE,KAAK;AAAA,MACxD,OAAO,CAAC,GAAG,KAAK,EAAE,KAAK,CAAC,GAAG,MAAM,IAAI,CAAC;AAAA,MACtC;AAAA,MACA,YAAY,KAAK,QAAQ,cAAc;AAAA,MACvC;AAAA,MACA;AAAA,MACA;AAAA,IACF,CAAC;AAAA,EACH;AAIA,QAAM,QAAgB,CAAC;AACvB,aAAW,WAAW,KAAK,UAAU;AACnC,eAAW,YAAY,KAAK,WAAW;AACrC,iBAAW,QAAQ,OAAO;AACxB,cAAM,KAAK,EAAE,SAAS,UAAU,KAAK,CAAC;AAAA,MACxC;AAAA,IACF;AAAA,EACF;AAEA,QAAM,YAAY,IAAI,KAAK,IAAI,CAAC,EAAE,YAAY;AAC9C,QAAM,OAAoB,CAAC;AAC3B,QAAM,mBAAyC,CAAC;AAChD,QAAM,aAA0B,CAAC;AASjC,MAAI,SAAS;AACb,MAAI,WAAW;AACf,QAAM,gBAA2B,CAAC;AAKlC,QAAM,WAAW,oBAAI,IAA0B;AAE/C,iBAAe,SAAwB;AACrC,WAAO,CAAC,UAAU;AAChB,YAAM,IAAI;AACV,UAAI,KAAK,MAAM,OAAQ;AACvB,YAAM,OAAO,MAAM,CAAC;AACpB,UAAI;AACF,cAAM,SAAS,MAAM,WAAW,IAAI;AACpC,aAAK,KAAK,OAAO,MAAM;AACvB,yBAAiB,KAAK,OAAO,SAAS;AAAA,MACxC,SAAS,KAAK;AACZ,YAAI,eAAe,oBAAoB;AACrC,qBAAW,KAAK,IAAI,MAAM;AAC1B,cAAI,IAAI,UAAW,kBAAiB,KAAK,IAAI,SAAS;AAAA,QACxD,OAAO;AAKL,wBAAc,KAAK,GAAG;AACtB,qBAAW;AACX;AAAA,QACF;AAAA,MACF;AAAA,IACF;AAAA,EACF;AAEA,iBAAe,WACb,MAC+D;AAC/D,UAAM,SAAS,KAAK,SAAS,cAAc;AAAA,MACzC,YAAY,KAAK;AAAA,MACjB,OAAO;AAAA;AAAA,MACP,WAAW,KAAK,QAAQ;AAAA,MACxB,YAAY,KAAK,SAAS;AAAA,MAC1B,MAAM,KAAK;AAAA,IACb,CAAC;AACD,UAAM,gBAAuC;AAAA,MAC3C,YAAY,KAAK;AAAA,MACjB;AAAA,MACA,WAAW,KAAK,QAAQ;AAAA,MACxB,YAAY,KAAK,SAAS;AAAA,MAC1B,MAAM,KAAK;AAAA,IACb;AACA,UAAM,QAAQ,KAAK,aAAa,aAAa;AAC7C,UAAM,UAAU,eAAe,aAAa;AAE5C,UAAM,UAAU,IAAI,aAAa,OAAO;AAAA,MACtC;AAAA,MACA,KAAK,KAAK;AAAA,MACV,eAAe,KAAK;AAAA,IACtB,CAAC;AAGD,aAAS,IAAI,OAAO,OAAO;AAE3B,UAAM,UAA4B;AAAA,MAChC,GAAG,KAAK;AAAA,MACR;AAAA,MACA,cAAc,EAAE,MAAM;AAAA,IACxB;AAEA,UAAM,MAA6B;AAAA,MACjC;AAAA,MACA,cAAc,KAAK;AAAA,MACnB,SAAS,KAAK,QAAQ;AAAA,MACtB,WAAW,KAAK,QAAQ;AAAA,MACxB,YAAY,KAAK,SAAS;AAAA,MAC1B,cAAc,KAAK,SAAS,QAAQ,CAAC;AAAA,MACrC,MAAM,KAAK;AAAA,MACX;AAAA,MACA;AAAA,MACA;AAAA,MACA;AAAA,MACA;AAAA,IACF;AAEA,QAAI;AACF,YAAM,YAAY,IAAI;AACtB,UAAI;AACJ,UAAI;AACF,kBAAU,MAAM,KAAK,OAAO,GAAG;AAAA,MACjC,SAAS,KAAK;AACZ,cAAM,UAAU,eAAe,QAAQ,IAAI,UAAU,OAAO,GAAG;AAM/D,cAAM,cAAc,SAAS,OAAO,OAAO;AAC3C,cAAM,IAAI,mBAAmB;AAAA,UAC3B;AAAA,UACA,WAAW,KAAK,QAAQ;AAAA,UACxB,YAAY,KAAK,SAAS;AAAA,UAC1B,MAAM,KAAK;AAAA,UACX,QAAQ;AAAA,UACR,OAAO;AAAA,QACT,CAAC;AAAA,MACH;AACA,YAAM,SAAS,IAAI,IAAI;AAEvB,YAAM,kBAAkB,MAAM,kBAAkB,OAAO,OAAO,EAAE,GAAG,WAAW,QAAQ,CAAC;AACvF,UAAI,CAAC,gBAAgB,IAAI;AACvB,gBAAQ,oBAAoB;AAAA,UAC1B,KAAK;AACH,kBAAM,IAAI,kBAAkB,eAAe;AAAA,UAC7C,KAAK;AACH,kBAAM,IAAI;AAAA,cACR;AAAA,gBACE;AAAA,gBACA,WAAW,KAAK,QAAQ;AAAA,gBACxB,YAAY,KAAK,SAAS;AAAA,gBAC1B,MAAM,KAAK;AAAA,gBACX,QAAQ;AAAA,gBACR,OAAO,gBAAgB,OAAO,IAAI,CAAC,MAAM,EAAE,IAAI,EAAE,KAAK,IAAI;AAAA,cAC5D;AAAA,cACA;AAAA,YACF;AAAA,UACF,KAAK;AAEH;AAAA,QACJ;AAAA,MACF;AAEA,YAAM,gBAA4B;AAAA,QAChC,KAAK,QAAQ,OAAO,CAAC;AAAA,MACvB;AACA,UAAI,aAAa,UAAW,eAAc,eAAe,QAAQ;AAAA,UAC5D,eAAc,cAAc,QAAQ;AACzC,UAAI,QAAQ,gBAAgB,OAAW,eAAc,cAAc,QAAQ;AAE3E,YAAM,SAAoB;AAAA,QACxB;AAAA,QACA,cAAc,KAAK;AAAA,QACnB,aAAa,KAAK,QAAQ;AAAA,QAC1B,MAAM,KAAK;AAAA,QACX,OAAO,QAAQ;AAAA,QACf,YAAY,QAAQ;AAAA,QACpB,YAAY,QAAQ;AAAA,QACpB,WAAW,KAAK;AAAA,QAChB;AAAA,QACA,SAAS,QAAQ;AAAA,QACjB,YAAY,QAAQ;AAAA,QACpB,eAAe,QAAQ;AAAA,QACvB,SAAS;AAAA,QACT,GAAI,QAAQ,eAAe,EAAE,cAAc,QAAQ,aAAa,IAAI,CAAC;AAAA,QACrE,aAAa,QAAQ;AAAA,QACrB;AAAA,QACA,YAAY,KAAK,SAAS;AAAA,MAC5B;AACA,YAAM,gBACJ,QAAQ,iBACP,OAAO,KAAK,iBAAiB,aAC1B,MAAM,KAAK,aAAa;AAAA,QACtB,YAAY,KAAK;AAAA,QACjB;AAAA,QACA,WAAW,KAAK,QAAQ;AAAA,QACxB,YAAY,KAAK,SAAS;AAAA,QAC1B,MAAM,KAAK;AAAA,QACX,SAAS,KAAK,QAAQ;AAAA,QACtB,cAAc,KAAK,SAAS,QAAQ,CAAC;AAAA,MACvC,CAAC,IACD,KAAK;AACX,UAAI,kBAAkB,QAAW;AAC/B,cAAM,eAAe,MAAM,wBAAwB,aAAa;AAChE,qCAA6B,cAAc,QAAQ,OAAO,QAAQ,UAAU;AAC5E,eAAO,eAAe;AAAA,MACxB;AACA,aAAO,EAAE,QAAQ,kBAAkB,MAAM,GAAG,WAAW,gBAAgB;AAAA,IACzE,UAAE;AAIA,eAAS,OAAO,KAAK;AAAA,IACvB;AAAA,EACF;AAEA,QAAM,UAAU,MAAM,KAAK,EAAE,QAAQ,KAAK,IAAI,aAAa,MAAM,MAAM,EAAE,GAAG,MAAM,OAAO,CAAC;AAI1F,QAAM,QAAQ,WAAW,OAAO;AAMhC,aAAW,CAAC,OAAO,OAAO,KAAK,UAAU;AACvC,UAAM,cAAc,SAAS,OAAO,kDAAkD;AAAA,EACxF;AACA,WAAS,MAAM;AAEf,MAAI,cAAc,SAAS,GAAG;AAC5B,UAAM,cAAc,WAAW,IAC3B,cAAc,CAAC,IACf,IAAI;AAAA,MACF;AAAA,MACA,oBAAoB,cAAc,MAAM;AAAA,IAC1C;AAAA,EACN;AAGA,MAAI;AACJ,MAAI,KAAK,QAAQ;AACf,UAAM,aAAoC;AAAA,MACxC,GAAG,KAAK;AAAA,MACR,YAAY,KAAK,OAAO;AAAA,MACxB,OAAO,aAAa,QAAQ,WAAW;AAAA,MACvC,aAAa,IAAI,KAAK,IAAI,CAAC,EAAE,YAAY;AAAA,MACzC,qBAAqB,uBAAuB;AAAA,IAC9C;AACA,aAAS,MAAM,eAAe,MAAM,UAAU;AAAA,EAChD;AAEA,QAAM,UAAU,IAAI,KAAK,IAAI,CAAC,EAAE,YAAY;AAE5C,SAAO;AAAA,IACL,YAAY,KAAK;AAAA,IACjB;AAAA,IACA;AAAA,IACA;AAAA,IACA;AAAA,IACA;AAAA,IACA;AAAA,IACA;AAAA,IACA;AAAA,EACF;AACF;AAIA,IAAM,qBAAN,cAAiC,MAAM;AAAA,EAC5B;AAAA,EACA;AAAA,EACT,YAAY,QAAmB,WAAgC;AAC7D,UAAM,QAAQ,OAAO,SAAS,IAAI,OAAO,UAAU,IAAI,OAAO,IAAI,YAAY,OAAO,MAAM,EAAE;AAC7F,SAAK,SAAS;AACd,SAAK,YAAY;AAAA,EACnB;AACF;AAiBA,eAAsB,cACpB,SACA,OACA,QACe;AACf,QAAM,WAAW,MAAM,QAAQ,WAAW,OAAO,KAAK;AACtD,MAAI,aAAa,OAAW;AAC5B,MAAI,SAAS,WAAW,UAAW;AACnC,QAAM,QAAQ,SAAS,MAAM;AAC/B;AAEA,SAAS,sBAAsB,SAA6B;AAC1D,SAAO,CAAC,WAAmD;AACzD,QAAI,CAAC,SAAS;AACZ,YAAM,IAAI;AAAA,QACR;AAAA,MACF;AAAA,IACF;AACA,WAAO,IAAI,0BAA0B;AAAA,MACnC,KAAK,GAAG,OAAO,eAAe,OAAO,KAAK;AAAA,IAC5C,CAAC;AAAA,EACH;AACF;AAEA,eAAe,wBACb,OAC2B;AAC3B,MAAI,mBAAmB,KAAK,GAAG;AAC7B,QAAI,CAAE,MAAM,uBAAuB,KAAK,GAAI;AAC1C,YAAM,IAAI,MAAM,iEAAiE;AAAA,IACnF;AACA,WAAO;AAAA,EACT;AACA,SAAO,sBAAsB,KAAK;AACpC;AAEA,SAAS,mBACP,OAC2B;AAC3B,SAAO,mBAAmB,SAAS,YAAY;AACjD;AAEA,SAAS,6BACP,SACA,OACA,YACM;AACN,MAAI,QAAQ,UAAU,UAAa,QAAQ,UAAU,OAAO;AAC1D,UAAM,IAAI;AAAA,MACR,wCAAwC,QAAQ,KAAK,mCAAmC,KAAK;AAAA,IAC/F;AAAA,EACF;AACA,MAAI,QAAQ,eAAe,UAAa,QAAQ,eAAe,YAAY;AACzE,UAAM,IAAI;AAAA,MACR,6CAA6C,QAAQ,UAAU,wCAAwC,UAAU;AAAA,IACnH;AAAA,EACF;AACF;AAEA,SAAS,aAAa,QAAuC;AAG3D,QAAM,OAAO,GAAG,OAAO,UAAU,KAAK,OAAO,SAAS,KAAK,OAAO,UAAU,KAAK,OAAO,IAAI;AAE5F,MAAI,KAAK;AACT,MAAI,KAAK;AACT,WAAS,IAAI,GAAG,IAAI,KAAK,QAAQ,KAAK;AACpC,UAAM,IAAI,KAAK,WAAW,CAAC;AAC3B,SAAK,KAAK,KAAK,KAAK,GAAG,QAAU,MAAM;AACvC,SAAK,KAAK,KAAK,KAAK,GAAG,UAAU,MAAM;AAAA,EACzC;AACA,SAAO,OAAO,GAAG,SAAS,EAAE,EAAE,SAAS,GAAG,GAAG,CAAC,GAAG,GAAG,SAAS,EAAE,EAAE,SAAS,GAAG,GAAG,CAAC;AACnF;","names":[]}

package/dist/{chunk-LO6IOIJ2.js → chunk-ABOIVNXL.js}

@@ -300,239 +300,6 @@ async function buildAgentInterfaceProfileCell(profile, input) {
   });
 }
 
-// src/run-record.ts
-var MANDATORY_TOP_LEVEL = [
-  "runId",
-  "experimentId",
-  "candidateId",
-  "seed",
-  "model",
-  "promptHash",
-  "configHash",
-  "commitSha",
-  "wallMs",
-  "costUsd",
-  "tokenUsage",
-  "outcome",
-  "splitTag"
-];
-var SPLIT_TAGS = ["search", "dev", "holdout"];
-var RunRecordValidationError = class extends ValidationError {
-  path;
-  constructor(message, path = "") {
-    super(path ? `${message} (at ${path})` : message);
-    this.path = path;
-  }
-};
-function validateRunRecord(input) {
-  if (input === null || typeof input !== "object") {
-    throw new RunRecordValidationError("expected object");
-  }
-  const obj = input;
-  for (const key of MANDATORY_TOP_LEVEL) {
-    if (!(key in obj)) {
-      throw new RunRecordValidationError(`missing mandatory field "${key}"`);
-    }
-  }
-  expectString2(obj.runId, "runId");
-  expectString2(obj.experimentId, "experimentId");
-  expectString2(obj.candidateId, "candidateId");
-  expectFiniteNumber(obj.seed, "seed");
-  expectString2(obj.model, "model");
-  expectString2(obj.promptHash, "promptHash");
-  expectString2(obj.configHash, "configHash");
-  expectString2(obj.commitSha, "commitSha");
-  expectFiniteNumber(obj.wallMs, "wallMs");
-  if (obj.queueMs !== void 0) expectFiniteNumber(obj.queueMs, "queueMs");
-  expectFiniteNumber(obj.costUsd, "costUsd");
-  if (!modelHasSnapshot(obj.model)) {
-    throw new RunRecordValidationError(
-      `model "${obj.model}" lacks a snapshot version (use 'name@YYYY-MM-DD' or 'name-YYYYMMDD')`,
-      "model"
-    );
-  }
-  const tu = obj.tokenUsage;
-  if (tu === null || typeof tu !== "object") {
-    throw new RunRecordValidationError("tokenUsage must be an object", "tokenUsage");
-  }
-  const tuRec = tu;
-  expectFiniteNumber(tuRec.input, "tokenUsage.input");
-  expectFiniteNumber(tuRec.output, "tokenUsage.output");
-  if (tuRec.cached !== void 0) expectFiniteNumber(tuRec.cached, "tokenUsage.cached");
-  if (obj.judgeMetadata !== void 0) {
-    const jm = obj.judgeMetadata;
-    if (jm === null || typeof jm !== "object") {
-      throw new RunRecordValidationError("judgeMetadata must be an object", "judgeMetadata");
-    }
-    const jmRec = jm;
-    expectString2(jmRec.model, "judgeMetadata.model");
-    expectString2(jmRec.promptVersion, "judgeMetadata.promptVersion");
-    expectFiniteNumber(jmRec.confidence, "judgeMetadata.confidence");
-    if (typeof jmRec.fallback !== "boolean") {
-      throw new RunRecordValidationError(
-        "judgeMetadata.fallback must be boolean",
-        "judgeMetadata.fallback"
-      );
-    }
-  }
-  const out = obj.outcome;
-  if (out === null || typeof out !== "object") {
-    throw new RunRecordValidationError("outcome must be an object", "outcome");
-  }
-  const outRec = out;
-  if (outRec.searchScore !== void 0)
-    expectFiniteNumber(outRec.searchScore, "outcome.searchScore");
-  if (outRec.holdoutScore !== void 0)
-    expectFiniteNumber(outRec.holdoutScore, "outcome.holdoutScore");
-  if (outRec.searchScore === void 0 && outRec.holdoutScore === void 0) {
-    throw new RunRecordValidationError(
-      "outcome must define searchScore or holdoutScore (or both)",
-      "outcome"
-    );
-  }
-  const raw = outRec.raw;
-  if (raw === null || typeof raw !== "object") {
-    throw new RunRecordValidationError("outcome.raw must be an object", "outcome.raw");
-  }
-  for (const [k, v] of Object.entries(raw)) {
-    expectFiniteNumber(v, `outcome.raw.${k}`);
-  }
-  if (outRec.realness !== void 0) {
-    const r = outRec.realness;
-    if (r === null || typeof r !== "object") {
-      throw new RunRecordValidationError("outcome.realness must be an object", "outcome.realness");
-    }
-    const rr = r;
-    expectFiniteNumber(rr.score, "outcome.realness.score");
-    if (typeof rr.gated !== "boolean") {
-      throw new RunRecordValidationError(
-        "outcome.realness.gated must be a boolean",
-        "outcome.realness.gated"
-      );
-    }
-  }
-  if (outRec.judgeScores !== void 0) {
-    validateJudgeScores(outRec.judgeScores, "outcome.judgeScores");
-  }
-  if (obj.failureMode !== void 0) expectString2(obj.failureMode, "failureMode");
-  if (obj.agentProfile !== void 0) {
-    try {
-      const profile = validateAgentProfileCell(obj.agentProfile);
-      if (profile.model !== void 0 && profile.model !== obj.model) {
-        throw new RunRecordValidationError(
-          `agentProfile.model "${profile.model}" does not match model "${obj.model}"`,
-          "agentProfile.model"
-        );
-      }
-      if (profile.promptHash !== void 0 && profile.promptHash !== obj.promptHash) {
-        throw new RunRecordValidationError(
-          `agentProfile.promptHash "${profile.promptHash}" does not match promptHash "${obj.promptHash}"`,
-          "agentProfile.promptHash"
-        );
-      }
-    } catch (error) {
-      if (error instanceof RunRecordValidationError) throw error;
-      if (error instanceof Error) {
-        throw new RunRecordValidationError(error.message, "agentProfile");
-      }
-      throw error;
-    }
-  }
-  if (typeof obj.splitTag !== "string" || !SPLIT_TAGS.includes(obj.splitTag)) {
-    throw new RunRecordValidationError(
-      `splitTag must be one of ${SPLIT_TAGS.join(", ")}, got ${String(obj.splitTag)}`,
-      "splitTag"
-    );
-  }
-  return input;
-}
-function isRunRecord(input) {
-  try {
-    validateRunRecord(input);
-    return true;
-  } catch {
-    return false;
-  }
-}
-function parseRunRecordSafe(input) {
-  try {
-    return { ok: true, value: validateRunRecord(input) };
-  } catch (e) {
-    if (e instanceof RunRecordValidationError) return { ok: false, error: e };
-    throw e;
-  }
-}
-function roundTripRunRecord(record) {
-  const json = JSON.stringify(record);
-  return validateRunRecord(JSON.parse(json));
-}
-function expectString2(value, path) {
-  if (typeof value !== "string" || value.length === 0) {
-    throw new RunRecordValidationError(`expected non-empty string`, path);
-  }
-}
-function expectFiniteNumber(value, path) {
-  if (typeof value !== "number" || !Number.isFinite(value)) {
-    throw new RunRecordValidationError(`expected finite number`, path);
-  }
-}
-function validateJudgeScores(value, path) {
-  if (value === null || typeof value !== "object") {
-    throw new RunRecordValidationError("judgeScores must be an object", path);
-  }
-  const rec = value;
-  const perJudge = rec.perJudge;
-  if (perJudge === null || typeof perJudge !== "object") {
-    throw new RunRecordValidationError("perJudge must be an object", `${path}.perJudge`);
-  }
-  for (const [judgeId, dims] of Object.entries(perJudge)) {
-    if (dims === null || typeof dims !== "object") {
-      throw new RunRecordValidationError(
-        "per-judge entry must be an object of dimension scores",
-        `${path}.perJudge.${judgeId}`
-      );
-    }
-    for (const [dim, score] of Object.entries(dims)) {
-      expectFiniteNumber(score, `${path}.perJudge.${judgeId}.${dim}`);
-    }
-  }
-  const perDimMean = rec.perDimMean;
-  if (perDimMean === null || typeof perDimMean !== "object") {
-    throw new RunRecordValidationError("perDimMean must be an object", `${path}.perDimMean`);
-  }
-  for (const [dim, mean] of Object.entries(perDimMean)) {
-    expectFiniteNumber(mean, `${path}.perDimMean.${dim}`);
-  }
-  expectFiniteNumber(rec.composite, `${path}.composite`);
-  if (rec.failedJudges !== void 0) {
-    if (!Array.isArray(rec.failedJudges)) {
-      throw new RunRecordValidationError(
-        "failedJudges must be an array of strings",
-        `${path}.failedJudges`
-      );
-    }
-    for (let i = 0; i < rec.failedJudges.length; i++) {
-      const id = rec.failedJudges[i];
-      if (typeof id !== "string" || id.length === 0) {
-        throw new RunRecordValidationError(
-          "failedJudges entry must be a non-empty string",
-          `${path}.failedJudges[${i}]`
-        );
-      }
-    }
-  }
-  if (rec.notes !== void 0 && typeof rec.notes !== "string") {
-    throw new RunRecordValidationError("notes must be a string", `${path}.notes`);
-  }
-}
-function modelHasSnapshot(model) {
-  if (model.includes("@")) return true;
-  if (/-\d{8}$/.test(model)) return true;
-  if (/-\d{4}-\d{2}-\d{2}$/.test(model)) return true;
-  if (/:date-/.test(model)) return true;
-  return false;
-}
-
 export {
   AgentProfileCellValidationError,
   buildAgentProfileCell,
@@ -545,11 +312,6 @@ export {
   groupRunsByAgentProfileCell,
   AGENT_PROFILE_KINDS,
   toAgentProfileJson,
-  buildAgentInterfaceProfileCell,
-  RunRecordValidationError,
-  validateRunRecord,
-  isRunRecord,
-  parseRunRecordSafe,
-  roundTripRunRecord
+  buildAgentInterfaceProfileCell
 };
-//# sourceMappingURL=chunk-LO6IOIJ2.js.map
+//# sourceMappingURL=chunk-ABOIVNXL.js.map

package/dist/chunk-ABOIVNXL.js.map

@@ -0,0 +1 @@
+{"version":3,"sources":["../src/agent-profile-cell.ts"],"sourcesContent":["import type { AgentProfile } from '@tangle-network/agent-interface'\nimport { ValidationError } from './errors'\nimport { hashJson } from './pre-registration'\n\nexport type AgentProfileCellSchemaVersion = 'agent-profile-cell/v1'\n\nexport type AgentProfileJson =\n  | string\n  | number\n  | boolean\n  | null\n  | AgentProfileJson[]\n  | { [key: string]: AgentProfileJson }\n\nexport type AgentProfileDimensionValue = string | number | boolean | null\n\nexport interface AgentProfileSource {\n  /** Runtime/profile contract being fingerprinted, e.g. `agent-interface-profile`. */\n  kind: string\n  /** sha256 over the canonical source profile object. */\n  hash: string\n}\n\nexport interface AgentProfileSourceInput {\n  kind: string\n  /** Precomputed sha256 for callers that already sign their profile artifact. */\n  hash?: string\n  /** Full canonical runtime profile; hashed and then discarded from the cell. */\n  profile?: AgentProfileJson\n}\n\nexport interface AgentProfileHarness {\n  id: string\n  version?: string\n  hash?: string\n}\n\nexport interface AgentProfileCellInput {\n  profileId: string\n  sourceProfile: AgentProfileSourceInput\n  harness?: AgentProfileHarness\n  model?: string\n  promptHash?: string\n  dimensions?: Record<string, AgentProfileDimensionValue>\n}\n\nexport interface AgentProfileCell {\n  schemaVersion: AgentProfileCellSchemaVersion\n  cellId: string\n  profileId: string\n  sourceProfile: AgentProfileSource\n  harness?: AgentProfileHarness\n  model?: string\n  promptHash?: string\n  dimensions?: Record<string, AgentProfileDimensionValue>\n}\n\nexport class AgentProfileCellValidationError extends ValidationError {\n  readonly path: string\n  constructor(message: string, path = '') {\n    super(path ? `${message} (at ${path})` : message)\n    this.path = path\n  }\n}\n\nconst SHA256_HEX = /^[0-9a-f]{64}$/\nconst CELL_ID = /^agent-profile-cell:sha256:[0-9a-f]{64}$/\n\nexport async function buildAgentProfileCell(\n  input: AgentProfileCellInput,\n): Promise<AgentProfileCell> {\n  const material = await normalizeAgentProfileCellInput(input)\n  const cellId = `agent-profile-cell:sha256:${await hashJson(material)}`\n  return { ...material, cellId }\n}\n\nexport function agentProfileCellHashMaterial(\n  cell: AgentProfileCell,\n): Omit<AgentProfileCell, 'cellId'> {\n  const { cellId: _cellId, ...material } = cell\n  void _cellId\n  return normalizeAgentProfileCell(material)\n}\n\nexport async function verifyAgentProfileCell(cell: AgentProfileCell): Promise<boolean> {\n  validateAgentProfileCell(cell)\n  return (\n    cell.cellId ===\n    `agent-profile-cell:sha256:${await hashJson(agentProfileCellHashMaterial(cell))}`\n  )\n}\n\nexport function validateAgentProfileCell(input: unknown): AgentProfileCell {\n  if (input === null || typeof input !== 'object') {\n    throw new AgentProfileCellValidationError('expected object')\n  }\n  const obj = input as Record<string, unknown>\n  expectLiteral(obj.schemaVersion, 'agent-profile-cell/v1', 'schemaVersion')\n  if (typeof obj.cellId !== 'string' || !CELL_ID.test(obj.cellId)) {\n    throw new AgentProfileCellValidationError(\n      'cellId must match agent-profile-cell:sha256:<64 lowercase hex chars>',\n      'cellId',\n    )\n  }\n  expectString(obj.profileId, 'profileId')\n  validateSource(obj.sourceProfile, 'sourceProfile')\n  if (obj.harness !== undefined) validateHarness(obj.harness, 'harness')\n  if (obj.model !== undefined) expectString(obj.model, 'model')\n  if (obj.promptHash !== undefined) expectString(obj.promptHash, 'promptHash')\n  if (obj.dimensions !== undefined) validateDimensions(obj.dimensions, 'dimensions')\n  return input as AgentProfileCell\n}\n\nexport function requireAgentProfileCell(record: {\n  runId: string\n  agentProfile?: AgentProfileCell\n}): AgentProfileCell {\n  if (!record.agentProfile) {\n    throw new AgentProfileCellValidationError(\n      `run \"${record.runId}\" is missing agentProfile; profile-cell grouping requires explicit profile identity`,\n      'agentProfile',\n    )\n  }\n  return validateAgentProfileCell(record.agentProfile)\n}\n\nexport function agentProfileCellKey(record: {\n  runId: string\n  agentProfile?: AgentProfileCell\n}): string {\n  return requireAgentProfileCell(record).cellId\n}\n\nexport async function assertRunAgentProfileCell(record: {\n  runId: string\n  model: string\n  promptHash: string\n  agentProfile?: AgentProfileCell\n}): Promise<AgentProfileCell> {\n  const profile = requireAgentProfileCell(record)\n  if (!(await verifyAgentProfileCell(profile))) {\n    throw new AgentProfileCellValidationError(\n      `run \"${record.runId}\" has an agentProfile.cellId that does not match its content`,\n      'agentProfile.cellId',\n    )\n  }\n  if (profile.model !== undefined && profile.model !== record.model) {\n    throw new AgentProfileCellValidationError(\n      `run \"${record.runId}\" agentProfile.model \"${profile.model}\" does not match model \"${record.model}\"`,\n      'agentProfile.model',\n    )\n  }\n  if (profile.promptHash !== undefined && profile.promptHash !== record.promptHash) {\n    throw new AgentProfileCellValidationError(\n      `run \"${record.runId}\" agentProfile.promptHash \"${profile.promptHash}\" does not match promptHash \"${record.promptHash}\"`,\n      'agentProfile.promptHash',\n    )\n  }\n  return profile\n}\n\nexport function groupRunsByAgentProfileCell<\n  T extends { runId: string; agentProfile?: AgentProfileCell },\n>(records: readonly T[]): Map<string, T[]> {\n  const groups = new Map<string, T[]>()\n  for (const record of records) {\n    const key = agentProfileCellKey(record)\n    const bucket = groups.get(key)\n    if (bucket) bucket.push(record)\n    else groups.set(key, [record])\n  }\n  return groups\n}\n\nasync function normalizeAgentProfileCellInput(\n  input: AgentProfileCellInput,\n): Promise<Omit<AgentProfileCell, 'cellId'>> {\n  return normalizeAgentProfileCell({\n    schemaVersion: 'agent-profile-cell/v1',\n    profileId: input.profileId,\n    sourceProfile: await normalizeSourceInput(input.sourceProfile),\n    harness: input.harness,\n    model: input.model,\n    promptHash: input.promptHash,\n    dimensions: input.dimensions,\n  })\n}\n\nfunction normalizeAgentProfileCell(\n  input: Omit<AgentProfileCell, 'cellId'>,\n): Omit<AgentProfileCell, 'cellId'> {\n  return compactObject({\n    schemaVersion: 'agent-profile-cell/v1' as const,\n    profileId: requireNonEmpty(input.profileId, 'profileId'),\n    sourceProfile: normalizeSource(input.sourceProfile),\n    harness: input.harness ? normalizeHarness(input.harness, 'harness') : undefined,\n    model: optionalNonEmpty(input.model, 'model'),\n    promptHash: optionalNonEmpty(input.promptHash, 'promptHash'),\n    dimensions: input.dimensions\n      ? nonEmptyRecord(normalizeDimensions(input.dimensions))\n      : undefined,\n  })\n}\n\nasync function normalizeSourceInput(input: AgentProfileSourceInput): Promise<AgentProfileSource> {\n  const kind = requireNonEmpty(input.kind, 'sourceProfile.kind')\n  if (input.hash !== undefined && input.profile !== undefined) {\n    throw new AgentProfileCellValidationError(\n      'sourceProfile must provide either hash or profile, not both',\n      'sourceProfile',\n    )\n  }\n  if (input.hash !== undefined) {\n    return { kind, hash: requireSha256Hex(input.hash, 'sourceProfile.hash') }\n  }\n  if (input.profile === undefined) {\n    throw new AgentProfileCellValidationError(\n      'sourceProfile must provide hash or profile',\n      'sourceProfile',\n    )\n  }\n  assertJson(input.profile, 'sourceProfile.profile')\n  return { kind, hash: await hashJson(input.profile) }\n}\n\nfunction normalizeSource(input: AgentProfileSource): AgentProfileSource {\n  return {\n    kind: requireNonEmpty(input.kind, 'sourceProfile.kind'),\n    hash: requireSha256Hex(input.hash, 'sourceProfile.hash'),\n  }\n}\n\nfunction normalizeHarness(input: AgentProfileHarness, path: string): AgentProfileHarness {\n  return compactObject({\n    id: requireNonEmpty(input.id, `${path}.id`),\n    version: optionalNonEmpty(input.version, `${path}.version`),\n    hash: optionalNonEmpty(input.hash, `${path}.hash`),\n  })\n}\n\nfunction normalizeDimensions(\n  input: Record<string, AgentProfileDimensionValue>,\n): Record<string, AgentProfileDimensionValue> {\n  const out: Record<string, AgentProfileDimensionValue> = {}\n  for (const key of Object.keys(input).sort()) {\n    const value = input[key]\n    requireNonEmpty(key, 'dimensions.<key>')\n    if (\n      value !== null &&\n      typeof value !== 'string' &&\n      typeof value !== 'number' &&\n      typeof value !== 'boolean'\n    ) {\n      throw new AgentProfileCellValidationError(\n        'expected primitive dimension value',\n        `dimensions.${key}`,\n      )\n    }\n    if (typeof value === 'number' && !Number.isFinite(value)) {\n      throw new AgentProfileCellValidationError('expected finite number', `dimensions.${key}`)\n    }\n    out[key] = value\n  }\n  return out\n}\n\nfunction compactObject<T extends Record<string, unknown>>(input: T): T {\n  const out: Record<string, unknown> = {}\n  for (const [key, value] of Object.entries(input)) {\n    if (value !== undefined) out[key] = value\n  }\n  return out as T\n}\n\nfunction nonEmptyRecord<T extends Record<string, unknown>>(input: T): T | undefined {\n  return Object.keys(input).length > 0 ? input : undefined\n}\n\nfunction validateSource(value: unknown, path: string): void {\n  if (value === null || typeof value !== 'object' || Array.isArray(value)) {\n    throw new AgentProfileCellValidationError('expected object', path)\n  }\n  const rec = value as Record<string, unknown>\n  expectString(rec.kind, `${path}.kind`)\n  requireSha256Hex(rec.hash, `${path}.hash`)\n}\n\nfunction validateHarness(value: unknown, path: string): void {\n  if (value === null || typeof value !== 'object' || Array.isArray(value)) {\n    throw new AgentProfileCellValidationError('expected object', path)\n  }\n  const rec = value as Record<string, unknown>\n  expectString(rec.id, `${path}.id`)\n  if (rec.version !== undefined) expectString(rec.version, `${path}.version`)\n  if (rec.hash !== undefined) expectString(rec.hash, `${path}.hash`)\n}\n\nfunction validateDimensions(value: unknown, path: string): void {\n  if (value === null || typeof value !== 'object' || Array.isArray(value)) {\n    throw new AgentProfileCellValidationError('expected object', path)\n  }\n  normalizeDimensions(value as Record<string, AgentProfileDimensionValue>)\n}\n\nfunction assertJson(value: AgentProfileJson, path: string): void {\n  if (value === null) return\n  const type = typeof value\n  if (type === 'string' || type === 'boolean') return\n  if (type === 'number') {\n    if (!Number.isFinite(value)) {\n      throw new AgentProfileCellValidationError('expected finite number', path)\n    }\n    return\n  }\n  if (Array.isArray(value)) {\n    value.forEach((item, index) => {\n      assertJson(item, `${path}[${index}]`)\n    })\n    return\n  }\n  if (type === 'object') {\n    for (const [key, nested] of Object.entries(value)) {\n      requireNonEmpty(key, `${path}.<key>`)\n      assertJson(nested, `${path}.${key}`)\n    }\n    return\n  }\n  throw new AgentProfileCellValidationError('expected JSON-compatible value', path)\n}\n\nfunction expectLiteral(value: unknown, expected: string, path: string): void {\n  if (value !== expected) {\n    throw new AgentProfileCellValidationError(`expected ${expected}`, path)\n  }\n}\n\nfunction expectString(value: unknown, path: string): void {\n  if (typeof value !== 'string' || value.length === 0) {\n    throw new AgentProfileCellValidationError('expected non-empty string', path)\n  }\n}\n\nfunction requireNonEmpty(value: string, path: string): string {\n  if (typeof value !== 'string' || value.length === 0) {\n    throw new AgentProfileCellValidationError('expected non-empty string', path)\n  }\n  return value\n}\n\nfunction optionalNonEmpty(value: string | undefined, path: string): string | undefined {\n  if (value === undefined) return undefined\n  return requireNonEmpty(value, path)\n}\n\nfunction requireSha256Hex(value: unknown, path: string): string {\n  if (typeof value !== 'string' || !SHA256_HEX.test(value)) {\n    throw new AgentProfileCellValidationError('expected 64 lowercase sha256 hex chars', path)\n  }\n  return value\n}\n\n// ── Consumer helpers ─────────────────────────────────────────────────\n//\n// Boilerplate every product consuming `buildAgentProfileCell` used to duplicate:\n//\n//   1. A `JSON.parse(JSON.stringify(value))` helper that canonicalizes an\n//      arbitrary `@tangle-network/agent-interface` `AgentProfile` into the recursive\n//      `AgentProfileJson` shape, with a fail-loud error when the profile\n//      is not JSON-serializable.\n//\n//   2. The magic string `'agent-interface-profile'` for `sourceProfile.kind`.\n//\n// Both belong here so the cross-product cell join (same canonical profile\n// hashes to the same `sourceProfile.hash` across products) is enforced by\n// the type system, not by every consumer remembering to do it right.\n// See blueprint-agent issue tangle-network/agent-eval#82.\n\n/** Canonical `sourceProfile.kind` values. Two products fingerprinting the\n *  same canonical profile MUST use the same kind for their cells to share\n *  `sourceProfile.hash`. Extend rather than create new strings — adding a\n *  new kind is a deliberate cross-product schema change. */\nexport const AGENT_PROFILE_KINDS = {\n  /** A profile declared via `defineAgentProfile(...)` from\n   *  `@tangle-network/agent-interface`. The default kind for router-backed\n   *  and sandbox-backed products. */\n  AGENT_INTERFACE_PROFILE: 'agent-interface-profile',\n} as const\n\nexport type AgentProfileKind = (typeof AGENT_PROFILE_KINDS)[keyof typeof AGENT_PROFILE_KINDS]\n\n/** Canonicalize an arbitrary value into `AgentProfileJson` by JSON\n *  round-trip. Throws when the value contains anything not representable\n *  as JSON (functions, BigInt, cycles) — non-portable profiles fail loud\n *  rather than silently dropping fields. */\nexport function toAgentProfileJson(value: unknown): AgentProfileJson {\n  let serialized: string | undefined\n  try {\n    serialized = JSON.stringify(value)\n  } catch (err) {\n    throw new AgentProfileCellValidationError(\n      `agent profile must be JSON-serializable: ${err instanceof Error ? err.message : String(err)}`,\n      'sourceProfile.profile',\n    )\n  }\n  if (serialized === undefined) {\n    throw new AgentProfileCellValidationError(\n      'agent profile must be JSON-serializable (got undefined after JSON.stringify)',\n      'sourceProfile.profile',\n    )\n  }\n  return JSON.parse(serialized) as AgentProfileJson\n}\n\n/** Canonical AgentProfile shape required when deriving a stable cell id. */\nexport type AgentInterfaceProfileLike = AgentProfile & { name: string; version: string }\n\n/** Higher-level helper that hard-codes the canonical\n *  `agent-interface-profile` kind plus the JSON canonicalization. Equivalent\n *  to calling `buildAgentProfileCell` with `profileId = \\`${name}@${version}\\``\n *  and `sourceProfile = { kind: AGENT_INTERFACE_PROFILE, profile: <round-tripped> }`.\n *\n *  Use this from any product consuming an agent-interface `AgentProfile`; the\n *  manual `buildAgentProfileCell` call is reserved for advanced cases\n *  (custom kinds, pre-computed source hashes, alternate profileId\n *  conventions). */\nexport async function buildAgentInterfaceProfileCell(\n  profile: AgentInterfaceProfileLike,\n  input: Omit<AgentProfileCellInput, 'profileId' | 'sourceProfile'>,\n): Promise<AgentProfileCell> {\n  if (!profile || typeof profile !== 'object') {\n    throw new AgentProfileCellValidationError('AgentProfile must be an object', 'profile')\n  }\n  if (typeof profile.name !== 'string' || profile.name.length === 0) {\n    throw new AgentProfileCellValidationError(\n      'AgentProfile must have a non-empty `name`',\n      'profile.name',\n    )\n  }\n  if (typeof profile.version !== 'string' || profile.version.length === 0) {\n    throw new AgentProfileCellValidationError(\n      'AgentProfile must have a non-empty `version`',\n      'profile.version',\n    )\n  }\n  return buildAgentProfileCell({\n    ...input,\n    profileId: `${profile.name}@${profile.version}`,\n    sourceProfile: {\n      kind: AGENT_PROFILE_KINDS.AGENT_INTERFACE_PROFILE,\n      profile: toAgentProfileJson(profile),\n    },\n  })\n}\n"],"mappings":";;;;;;;;AAyDO,IAAM,kCAAN,cAA8C,gBAAgB;AAAA,EAC1D;AAAA,EACT,YAAY,SAAiB,OAAO,IAAI;AACtC,UAAM,OAAO,GAAG,OAAO,QAAQ,IAAI,MAAM,OAAO;AAChD,SAAK,OAAO;AAAA,EACd;AACF;AAEA,IAAM,aAAa;AACnB,IAAM,UAAU;AAEhB,eAAsB,sBACpB,OAC2B;AAC3B,QAAM,WAAW,MAAM,+BAA+B,KAAK;AAC3D,QAAM,SAAS,6BAA6B,MAAM,SAAS,QAAQ,CAAC;AACpE,SAAO,EAAE,GAAG,UAAU,OAAO;AAC/B;AAEO,SAAS,6BACd,MACkC;AAClC,QAAM,EAAE,QAAQ,SAAS,GAAG,SAAS,IAAI;AACzC,OAAK;AACL,SAAO,0BAA0B,QAAQ;AAC3C;AAEA,eAAsB,uBAAuB,MAA0C;AACrF,2BAAyB,IAAI;AAC7B,SACE,KAAK,WACL,6BAA6B,MAAM,SAAS,6BAA6B,IAAI,CAAC,CAAC;AAEnF;AAEO,SAAS,yBAAyB,OAAkC;AACzE,MAAI,UAAU,QAAQ,OAAO,UAAU,UAAU;AAC/C,UAAM,IAAI,gCAAgC,iBAAiB;AAAA,EAC7D;AACA,QAAM,MAAM;AACZ,gBAAc,IAAI,eAAe,yBAAyB,eAAe;AACzE,MAAI,OAAO,IAAI,WAAW,YAAY,CAAC,QAAQ,KAAK,IAAI,MAAM,GAAG;AAC/D,UAAM,IAAI;AAAA,MACR;AAAA,MACA;AAAA,IACF;AAAA,EACF;AACA,eAAa,IAAI,WAAW,WAAW;AACvC,iBAAe,IAAI,eAAe,eAAe;AACjD,MAAI,IAAI,YAAY,OAAW,iBAAgB,IAAI,SAAS,SAAS;AACrE,MAAI,IAAI,UAAU,OAAW,cAAa,IAAI,OAAO,OAAO;AAC5D,MAAI,IAAI,eAAe,OAAW,cAAa,IAAI,YAAY,YAAY;AAC3E,MAAI,IAAI,eAAe,OAAW,oBAAmB,IAAI,YAAY,YAAY;AACjF,SAAO;AACT;AAEO,SAAS,wBAAwB,QAGnB;AACnB,MAAI,CAAC,OAAO,cAAc;AACxB,UAAM,IAAI;AAAA,MACR,QAAQ,OAAO,KAAK;AAAA,MACpB;AAAA,IACF;AAAA,EACF;AACA,SAAO,yBAAyB,OAAO,YAAY;AACrD;AAEO,SAAS,oBAAoB,QAGzB;AACT,SAAO,wBAAwB,MAAM,EAAE;AACzC;AAEA,eAAsB,0BAA0B,QAKlB;AAC5B,QAAM,UAAU,wBAAwB,MAAM;AAC9C,MAAI,CAAE,MAAM,uBAAuB,OAAO,GAAI;AAC5C,UAAM,IAAI;AAAA,MACR,QAAQ,OAAO,KAAK;AAAA,MACpB;AAAA,IACF;AAAA,EACF;AACA,MAAI,QAAQ,UAAU,UAAa,QAAQ,UAAU,OAAO,OAAO;AACjE,UAAM,IAAI;AAAA,MACR,QAAQ,OAAO,KAAK,yBAAyB,QAAQ,KAAK,2BAA2B,OAAO,KAAK;AAAA,MACjG;AAAA,IACF;AAAA,EACF;AACA,MAAI,QAAQ,eAAe,UAAa,QAAQ,eAAe,OAAO,YAAY;AAChF,UAAM,IAAI;AAAA,MACR,QAAQ,OAAO,KAAK,8BAA8B,QAAQ,UAAU,gCAAgC,OAAO,UAAU;AAAA,MACrH;AAAA,IACF;AAAA,EACF;AACA,SAAO;AACT;AAEO,SAAS,4BAEd,SAAyC;AACzC,QAAM,SAAS,oBAAI,IAAiB;AACpC,aAAW,UAAU,SAAS;AAC5B,UAAM,MAAM,oBAAoB,MAAM;AACtC,UAAM,SAAS,OAAO,IAAI,GAAG;AAC7B,QAAI,OAAQ,QAAO,KAAK,MAAM;AAAA,QACzB,QAAO,IAAI,KAAK,CAAC,MAAM,CAAC;AAAA,EAC/B;AACA,SAAO;AACT;AAEA,eAAe,+BACb,OAC2C;AAC3C,SAAO,0BAA0B;AAAA,IAC/B,eAAe;AAAA,IACf,WAAW,MAAM;AAAA,IACjB,eAAe,MAAM,qBAAqB,MAAM,aAAa;AAAA,IAC7D,SAAS,MAAM;AAAA,IACf,OAAO,MAAM;AAAA,IACb,YAAY,MAAM;AAAA,IAClB,YAAY,MAAM;AAAA,EACpB,CAAC;AACH;AAEA,SAAS,0BACP,OACkC;AAClC,SAAO,cAAc;AAAA,IACnB,eAAe;AAAA,IACf,WAAW,gBAAgB,MAAM,WAAW,WAAW;AAAA,IACvD,eAAe,gBAAgB,MAAM,aAAa;AAAA,IAClD,SAAS,MAAM,UAAU,iBAAiB,MAAM,SAAS,SAAS,IAAI;AAAA,IACtE,OAAO,iBAAiB,MAAM,OAAO,OAAO;AAAA,IAC5C,YAAY,iBAAiB,MAAM,YAAY,YAAY;AAAA,IAC3D,YAAY,MAAM,aACd,eAAe,oBAAoB,MAAM,UAAU,CAAC,IACpD;AAAA,EACN,CAAC;AACH;AAEA,eAAe,qBAAqB,OAA6D;AAC/F,QAAM,OAAO,gBAAgB,MAAM,MAAM,oBAAoB;AAC7D,MAAI,MAAM,SAAS,UAAa,MAAM,YAAY,QAAW;AAC3D,UAAM,IAAI;AAAA,MACR;AAAA,MACA;AAAA,IACF;AAAA,EACF;AACA,MAAI,MAAM,SAAS,QAAW;AAC5B,WAAO,EAAE,MAAM,MAAM,iBAAiB,MAAM,MAAM,oBAAoB,EAAE;AAAA,EAC1E;AACA,MAAI,MAAM,YAAY,QAAW;AAC/B,UAAM,IAAI;AAAA,MACR;AAAA,MACA;AAAA,IACF;AAAA,EACF;AACA,aAAW,MAAM,SAAS,uBAAuB;AACjD,SAAO,EAAE,MAAM,MAAM,MAAM,SAAS,MAAM,OAAO,EAAE;AACrD;AAEA,SAAS,gBAAgB,OAA+C;AACtE,SAAO;AAAA,IACL,MAAM,gBAAgB,MAAM,MAAM,oBAAoB;AAAA,IACtD,MAAM,iBAAiB,MAAM,MAAM,oBAAoB;AAAA,EACzD;AACF;AAEA,SAAS,iBAAiB,OAA4B,MAAmC;AACvF,SAAO,cAAc;AAAA,IACnB,IAAI,gBAAgB,MAAM,IAAI,GAAG,IAAI,KAAK;AAAA,IAC1C,SAAS,iBAAiB,MAAM,SAAS,GAAG,IAAI,UAAU;AAAA,IAC1D,MAAM,iBAAiB,MAAM,MAAM,GAAG,IAAI,OAAO;AAAA,EACnD,CAAC;AACH;AAEA,SAAS,oBACP,OAC4C;AAC5C,QAAM,MAAkD,CAAC;AACzD,aAAW,OAAO,OAAO,KAAK,KAAK,EAAE,KAAK,GAAG;AAC3C,UAAM,QAAQ,MAAM,GAAG;AACvB,oBAAgB,KAAK,kBAAkB;AACvC,QACE,UAAU,QACV,OAAO,UAAU,YACjB,OAAO,UAAU,YACjB,OAAO,UAAU,WACjB;AACA,YAAM,IAAI;AAAA,QACR;AAAA,QACA,cAAc,GAAG;AAAA,MACnB;AAAA,IACF;AACA,QAAI,OAAO,UAAU,YAAY,CAAC,OAAO,SAAS,KAAK,GAAG;AACxD,YAAM,IAAI,gCAAgC,0BAA0B,cAAc,GAAG,EAAE;AAAA,IACzF;AACA,QAAI,GAAG,IAAI;AAAA,EACb;AACA,SAAO;AACT;AAEA,SAAS,cAAiD,OAAa;AACrE,QAAM,MAA+B,CAAC;AACtC,aAAW,CAAC,KAAK,KAAK,KAAK,OAAO,QAAQ,KAAK,GAAG;AAChD,QAAI,UAAU,OAAW,KAAI,GAAG,IAAI;AAAA,EACtC;AACA,SAAO;AACT;AAEA,SAAS,eAAkD,OAAyB;AAClF,SAAO,OAAO,KAAK,KAAK,EAAE,SAAS,IAAI,QAAQ;AACjD;AAEA,SAAS,eAAe,OAAgB,MAAoB;AAC1D,MAAI,UAAU,QAAQ,OAAO,UAAU,YAAY,MAAM,QAAQ,KAAK,GAAG;AACvE,UAAM,IAAI,gCAAgC,mBAAmB,IAAI;AAAA,EACnE;AACA,QAAM,MAAM;AACZ,eAAa,IAAI,MAAM,GAAG,IAAI,OAAO;AACrC,mBAAiB,IAAI,MAAM,GAAG,IAAI,OAAO;AAC3C;AAEA,SAAS,gBAAgB,OAAgB,MAAoB;AAC3D,MAAI,UAAU,QAAQ,OAAO,UAAU,YAAY,MAAM,QAAQ,KAAK,GAAG;AACvE,UAAM,IAAI,gCAAgC,mBAAmB,IAAI;AAAA,EACnE;AACA,QAAM,MAAM;AACZ,eAAa,IAAI,IAAI,GAAG,IAAI,KAAK;AACjC,MAAI,IAAI,YAAY,OAAW,cAAa,IAAI,SAAS,GAAG,IAAI,UAAU;AAC1E,MAAI,IAAI,SAAS,OAAW,cAAa,IAAI,MAAM,GAAG,IAAI,OAAO;AACnE;AAEA,SAAS,mBAAmB,OAAgB,MAAoB;AAC9D,MAAI,UAAU,QAAQ,OAAO,UAAU,YAAY,MAAM,QAAQ,KAAK,GAAG;AACvE,UAAM,IAAI,gCAAgC,mBAAmB,IAAI;AAAA,EACnE;AACA,sBAAoB,KAAmD;AACzE;AAEA,SAAS,WAAW,OAAyB,MAAoB;AAC/D,MAAI,UAAU,KAAM;AACpB,QAAM,OAAO,OAAO;AACpB,MAAI,SAAS,YAAY,SAAS,UAAW;AAC7C,MAAI,SAAS,UAAU;AACrB,QAAI,CAAC,OAAO,SAAS,KAAK,GAAG;AAC3B,YAAM,IAAI,gCAAgC,0BAA0B,IAAI;AAAA,IAC1E;AACA;AAAA,EACF;AACA,MAAI,MAAM,QAAQ,KAAK,GAAG;AACxB,UAAM,QAAQ,CAAC,MAAM,UAAU;AAC7B,iBAAW,MAAM,GAAG,IAAI,IAAI,KAAK,GAAG;AAAA,IACtC,CAAC;AACD;AAAA,EACF;AACA,MAAI,SAAS,UAAU;AACrB,eAAW,CAAC,KAAK,MAAM,KAAK,OAAO,QAAQ,KAAK,GAAG;AACjD,sBAAgB,KAAK,GAAG,IAAI,QAAQ;AACpC,iBAAW,QAAQ,GAAG,IAAI,IAAI,GAAG,EAAE;AAAA,IACrC;AACA;AAAA,EACF;AACA,QAAM,IAAI,gCAAgC,kCAAkC,IAAI;AAClF;AAEA,SAAS,cAAc,OAAgB,UAAkB,MAAoB;AAC3E,MAAI,UAAU,UAAU;AACtB,UAAM,IAAI,gCAAgC,YAAY,QAAQ,IAAI,IAAI;AAAA,EACxE;AACF;AAEA,SAAS,aAAa,OAAgB,MAAoB;AACxD,MAAI,OAAO,UAAU,YAAY,MAAM,WAAW,GAAG;AACnD,UAAM,IAAI,gCAAgC,6BAA6B,IAAI;AAAA,EAC7E;AACF;AAEA,SAAS,gBAAgB,OAAe,MAAsB;AAC5D,MAAI,OAAO,UAAU,YAAY,MAAM,WAAW,GAAG;AACnD,UAAM,IAAI,gCAAgC,6BAA6B,IAAI;AAAA,EAC7E;AACA,SAAO;AACT;AAEA,SAAS,iBAAiB,OAA2B,MAAkC;AACrF,MAAI,UAAU,OAAW,QAAO;AAChC,SAAO,gBAAgB,OAAO,IAAI;AACpC;AAEA,SAAS,iBAAiB,OAAgB,MAAsB;AAC9D,MAAI,OAAO,UAAU,YAAY,CAAC,WAAW,KAAK,KAAK,GAAG;AACxD,UAAM,IAAI,gCAAgC,0CAA0C,IAAI;AAAA,EAC1F;AACA,SAAO;AACT;AAsBO,IAAM,sBAAsB;AAAA;AAAA;AAAA;AAAA,EAIjC,yBAAyB;AAC3B;AAQO,SAAS,mBAAmB,OAAkC;AACnE,MAAI;AACJ,MAAI;AACF,iBAAa,KAAK,UAAU,KAAK;AAAA,EACnC,SAAS,KAAK;AACZ,UAAM,IAAI;AAAA,MACR,4CAA4C,eAAe,QAAQ,IAAI,UAAU,OAAO,GAAG,CAAC;AAAA,MAC5F;AAAA,IACF;AAAA,EACF;AACA,MAAI,eAAe,QAAW;AAC5B,UAAM,IAAI;AAAA,MACR;AAAA,MACA;AAAA,IACF;AAAA,EACF;AACA,SAAO,KAAK,MAAM,UAAU;AAC9B;AAcA,eAAsB,+BACpB,SACA,OAC2B;AAC3B,MAAI,CAAC,WAAW,OAAO,YAAY,UAAU;AAC3C,UAAM,IAAI,gCAAgC,kCAAkC,SAAS;AAAA,EACvF;AACA,MAAI,OAAO,QAAQ,SAAS,YAAY,QAAQ,KAAK,WAAW,GAAG;AACjE,UAAM,IAAI;AAAA,MACR;AAAA,MACA;AAAA,IACF;AAAA,EACF;AACA,MAAI,OAAO,QAAQ,YAAY,YAAY,QAAQ,QAAQ,WAAW,GAAG;AACvE,UAAM,IAAI;AAAA,MACR;AAAA,MACA;AAAA,IACF;AAAA,EACF;AACA,SAAO,sBAAsB;AAAA,IAC3B,GAAG;AAAA,IACH,WAAW,GAAG,QAAQ,IAAI,IAAI,QAAQ,OAAO;AAAA,IAC7C,eAAe;AAAA,MACb,MAAM,oBAAoB;AAAA,MAC1B,SAAS,mBAAmB,OAAO;AAAA,IACrC;AAAA,EACF,CAAC;AACH;","names":[]}

package/dist/{chunk-NZEQVRH5.js → chunk-BOETF6BU.js}

@@ -4,7 +4,7 @@ import {
 } from "./chunk-YEHAEDUD.js";
 import {
   validateRunRecord
-} from "./chunk-LO6IOIJ2.js";
+} from "./chunk-G6S73VA7.js";
 import {
   TraceEmitter
 } from "./chunk-TVVP3ZZQ.js";
@@ -610,4 +610,4 @@ export {
   runProposeReviewAsControlLoop,
   controlFailureClassFromVerification
 };
-//# sourceMappingURL=chunk-NZEQVRH5.js.map
+//# sourceMappingURL=chunk-BOETF6BU.js.map