@tangle-network/agent-eval 0.135.0 → 0.135.1

package/dist/{eval-campaign-CHFxPTVl.js → eval-campaign-aMG73hvX.js}

@@ -1,6 +1,6 @@
 import { f as verifyAgentProfileCell, h as hashJson, p as canonicalize, s as buildAgentProfileCell } from "./agent-profile-cell-OhuTee9n.js";
 import { t as FileSystemRawProviderSink } from "./raw-provider-sink-BQd7mzyT.js";
-import { a as assertLlmRoute } from "./llm-client-BNcP4v08.js";
+import { a as assertLlmRoute } from "./llm-client-DHx8pzyJ.js";
 import { t as TraceEmitter } from "./emitter-CPBAhxum.js";
 import { s as validateRunRecord } from "./run-record-BIwU2wdV.js";
 import { i as researchReport } from "./summary-report-BxtossFi.js";
@@ -346,4 +346,4 @@ function defaultRunId(params) {
 //#endregion
 export { runEvalCampaign as t };
 
-//# sourceMappingURL=eval-campaign-CHFxPTVl.js.map
+//# sourceMappingURL=eval-campaign-aMG73hvX.js.map

package/dist/{eval-campaign-CHFxPTVl.js.map → eval-campaign-aMG73hvX.js.map}

@@ -1 +1 @@
-{"version":3,"file":"eval-campaign-CHFxPTVl.js","names":[],"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  RunCostProvenance,\n  RunJudgeMetadata,\n  RunOutcome,\n  RunRecord,\n  RunSplitTag,\n  RunTaskFailure,\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 { 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\ninterface CampaignRunOutcomeFields {\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  /** Cost in USD, or null when the runner could not capture it. */\n  costUsd: number | null\n  /** Source of the cost amount. */\n  costProvenance: RunCostProvenance\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  /** 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\n/** Campaign result with the same task-failure invariant as `RunRecord`. */\nexport type CampaignRunOutcome = CampaignRunOutcomeFields & RunTaskFailure\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        costProvenance: outcome.costProvenance,\n        tokenUsage: outcome.tokenUsage,\n        terminalOutcome: 'succeeded',\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,MAAM,oBAA8C;CAClD,aAAa;CACb,8BAA8B;CAC9B,gBAAgB;AAClB;AAEA,MAAM,gBAAsC;CAC1C,wBAAwB;CACxB,aAAa;AACf;AAEA,eAAsB,gBACpB,MAC6B;CAE7B,eAAe,KAAK,SAAS,KAAK,qBAAqB,aAAa;CAEpE,IAAI,KAAK,SAAS,WAAW,GAC3B,MAAM,IAAI,MAAM,8CAA8C;CAEhE,IAAI,KAAK,UAAU,WAAW,GAC5B,MAAM,IAAI,MAAM,+CAA+C;CAEjE,MAAM,6BAAa,IAAI,IAAY;CACnC,KAAK,MAAM,KAAK,KAAK,UAAU;EAC7B,IAAI,WAAW,IAAI,EAAE,EAAE,GACrB,MAAM,IAAI,MAAM,0CAA0C,EAAE,GAAG,GAAG;EAEpE,WAAW,IAAI,EAAE,EAAE;CACrB;CACA,MAAM,8BAAc,IAAI,IAAY;CACpC,KAAK,MAAM,KAAK,KAAK,WAAW;EAC9B,IAAI,YAAY,IAAI,EAAE,UAAU,GAC9B,MAAM,IAAI,MAAM,0CAA0C,EAAE,WAAW,GAAG;EAE5E,YAAY,IAAI,EAAE,UAAU;CAC9B;CACA,IAAI,KAAK,QAAQ,cAAc,CAAC,WAAW,IAAI,KAAK,OAAO,UAAU,GACnE,MAAM,IAAI,MACR,uCAAuC,KAAK,OAAO,WAAW,iCAChE;CAEF,IAAI,CAAC,KAAK,WACR,MAAM,IAAI,MAAM,oEAAoE;CAGtF,MAAM,QAAQ,KAAK,SAAS;EAAC;EAAG;EAAG;CAAC;CACpC,MAAM,WAAwB,KAAK,YAAY;CAC/C,MAAM,cAAc,KAAK,IAAI,GAAG,KAAK,eAAe,CAAC;CACrD,MAAM,YAAY;EAAE,GAAG;EAAmB,GAAI,KAAK,aAAa,CAAC;CAAG;CACpE,MAAM,qBAA8C,KAAK,sBAAsB;CAC/E,MAAM,MAAM,KAAK,cAAc,KAAK,IAAI;CACxC,MAAM,WAAW,KAAK,QAAQ,WAAW,GAAA,CAAI,QAAQ,QAAQ,EAAE;CAC/D,MAAM,WAAW,KAAK,QAAQ,YAAY;CAC1C,MAAM,sBAAsB,KAAK,uBAAuB;CAExD,MAAM,iBAAiB,KAAK,kBAAkB,sBAAsB,KAAK,OAAO;CAGhF,MAAM,sBAAsB,MAAM,SAChC,aAAa;EACX,YAAY,KAAK;EACjB,UAAU,KAAK,SAAS,KAAK,MAAM,EAAE,EAAE,CAAC,CAAC,KAAK;EAC9C,WAAW,KAAK,UAAU,KAAK,MAAM,EAAE,UAAU,CAAC,CAAC,KAAK;EACxD,OAAO,CAAC,GAAG,KAAK,CAAC,CAAC,MAAM,GAAG,MAAM,IAAI,CAAC;EACtC;EACA,YAAY,KAAK,QAAQ,cAAc;EACvC;EACA;EACA;CACF,CAAC,CACH;CAIA,MAAM,QAAgB,CAAC;CACvB,KAAK,MAAM,WAAW,KAAK,UACzB,KAAK,MAAM,YAAY,KAAK,WAC1B,KAAK,MAAM,QAAQ,OACjB,MAAM,KAAK;EAAE;EAAS;EAAU;CAAK,CAAC;CAK5C,MAAM,YAAY,IAAI,KAAK,IAAI,CAAC,CAAC,CAAC,YAAY;CAC9C,MAAM,OAAoB,CAAC;CAC3B,MAAM,mBAAyC,CAAC;CAChD,MAAM,aAA0B,CAAC;CASjC,IAAI,SAAS;CACb,IAAI,WAAW;CACf,MAAM,gBAA2B,CAAC;CAKlC,MAAM,2BAAW,IAAI,IAA0B;CAE/C,eAAe,SAAwB;EACrC,OAAO,CAAC,UAAU;GAChB,MAAM,IAAI;GACV,IAAI,KAAK,MAAM,QAAQ;GACvB,MAAM,OAAO,MAAM;GACnB,IAAI;IACF,MAAM,SAAS,MAAM,WAAW,IAAI;IACpC,KAAK,KAAK,OAAO,MAAM;IACvB,iBAAiB,KAAK,OAAO,SAAS;GACxC,SAAS,KAAK;IACZ,IAAI,eAAe,oBAAoB;KACrC,WAAW,KAAK,IAAI,MAAM;KAC1B,IAAI,IAAI,WAAW,iBAAiB,KAAK,IAAI,SAAS;IACxD,OAAO;KAKL,cAAc,KAAK,GAAG;KACtB,WAAW;KACX;IACF;GACF;EACF;CACF;CAEA,eAAe,WACb,MAC+D;EAC/D,MAAM,SAAS,KAAK,SAAS,aAAA,CAAc;GACzC,YAAY,KAAK;GACjB,OAAO;GACP,WAAW,KAAK,QAAQ;GACxB,YAAY,KAAK,SAAS;GAC1B,MAAM,KAAK;EACb,CAAC;EACD,MAAM,gBAAuC;GAC3C,YAAY,KAAK;GACjB;GACA,WAAW,KAAK,QAAQ;GACxB,YAAY,KAAK,SAAS;GAC1B,MAAM,KAAK;EACb;EACA,MAAM,QAAQ,KAAK,aAAa,aAAa;EAC7C,MAAM,UAAU,eAAe,aAAa;EAE5C,MAAM,UAAU,IAAI,aAAa,OAAO;GACtC;GACA,KAAK,KAAK;GACV,eAAe,KAAK;EACtB,CAAC;EAGD,SAAS,IAAI,OAAO,OAAO;EAE3B,MAAM,UAA4B;GAChC,GAAG,KAAK;GACR;GACA,cAAc,EAAE,MAAM;EACxB;EAEA,MAAM,MAA6B;GACjC;GACA,cAAc,KAAK;GACnB,SAAS,KAAK,QAAQ;GACtB,WAAW,KAAK,QAAQ;GACxB,YAAY,KAAK,SAAS;GAC1B,cAAc,KAAK,SAAS,QAAQ,CAAC;GACrC,MAAM,KAAK;GACX;GACA;GACA;GACA;GACA;EACF;EAEA,IAAI;GACF,MAAM,YAAY,IAAI;GACtB,IAAI;GACJ,IAAI;IACF,UAAU,MAAM,KAAK,OAAO,GAAG;GACjC,SAAS,KAAK;IACZ,MAAM,UAAU,eAAe,QAAQ,IAAI,UAAU,OAAO,GAAG;IAM/D,MAAM,cAAc,SAAS,OAAO,OAAO;IAC3C,MAAM,IAAI,mBAAmB;KAC3B;KACA,WAAW,KAAK,QAAQ;KACxB,YAAY,KAAK,SAAS;KAC1B,MAAM,KAAK;KACX,QAAQ;KACR,OAAO;IACT,CAAC;GACH;GACA,MAAM,SAAS,IAAI,IAAI;GAEvB,MAAM,kBAAkB,MAAM,kBAAkB,OAAO,OAAO;IAAE,GAAG;IAAW;GAAQ,CAAC;GACvF,IAAI,CAAC,gBAAgB,IACnB,QAAQ,oBAAR;IACE,KAAK,SACH,MAAM,IAAI,kBAAkB,eAAe;IAC7C,KAAK,eACH,MAAM,IAAI,mBACR;KACE;KACA,WAAW,KAAK,QAAQ;KACxB,YAAY,KAAK,SAAS;KAC1B,MAAM,KAAK;KACX,QAAQ;KACR,OAAO,gBAAgB,OAAO,KAAK,MAAM,EAAE,IAAI,CAAC,CAAC,KAAK,IAAI;IAC5D,GACA,eACF;IACF,KAAK,OAEH;GACJ;GAGF,MAAM,gBAA4B,EAChC,KAAK,QAAQ,OAAO,CAAC,EACvB;GACA,IAAI,aAAa,WAAW,cAAc,eAAe,QAAQ;QAC5D,cAAc,cAAc,QAAQ;GACzC,IAAI,QAAQ,gBAAgB,KAAA,GAAW,cAAc,cAAc,QAAQ;GAE3E,MAAM,SAAoB;IACxB;IACA,cAAc,KAAK;IACnB,aAAa,KAAK,QAAQ;IAC1B,MAAM,KAAK;IACX,OAAO,QAAQ;IACf,YAAY,QAAQ;IACpB,YAAY,QAAQ;IACpB,WAAW,KAAK;IAChB;IACA,SAAS,QAAQ;IACjB,gBAAgB,QAAQ;IACxB,YAAY,QAAQ;IACpB,iBAAiB;IACjB,eAAe,QAAQ;IACvB,SAAS;IACT,GAAI,QAAQ,eAAe,EAAE,cAAc,QAAQ,aAAa,IAAI,CAAC;IACrE,aAAa,QAAQ;IACrB;IACA,YAAY,KAAK,SAAS;GAC5B;GACA,MAAM,gBACJ,QAAQ,iBACP,OAAO,KAAK,iBAAiB,aAC1B,MAAM,KAAK,aAAa;IACtB,YAAY,KAAK;IACjB;IACA,WAAW,KAAK,QAAQ;IACxB,YAAY,KAAK,SAAS;IAC1B,MAAM,KAAK;IACX,SAAS,KAAK,QAAQ;IACtB,cAAc,KAAK,SAAS,QAAQ,CAAC;GACvC,CAAC,IACD,KAAK;GACX,IAAI,kBAAkB,KAAA,GAAW;IAC/B,MAAM,eAAe,MAAM,wBAAwB,aAAa;IAChE,6BAA6B,cAAc,QAAQ,OAAO,QAAQ,UAAU;IAC5E,OAAO,eAAe;GACxB;GACA,OAAO;IAAE,QAAQ,kBAAkB,MAAM;IAAG,WAAW;GAAgB;EACzE,UAAU;GAIR,SAAS,OAAO,KAAK;EACvB;CACF;CAEA,MAAM,UAAU,MAAM,KAAK,EAAE,QAAQ,KAAK,IAAI,aAAa,MAAM,MAAM,EAAE,SAAS,OAAO,CAAC;CAI1F,MAAM,QAAQ,WAAW,OAAO;CAMhC,KAAK,MAAM,CAAC,OAAO,YAAY,UAC7B,MAAM,cAAc,SAAS,OAAO,kDAAkD;CAExF,SAAS,MAAM;CAEf,IAAI,cAAc,SAAS,GACzB,MAAM,cAAc,WAAW,IAC3B,cAAc,KACd,IAAI,eACF,eACA,oBAAoB,cAAc,OAAO,iDAC3C;CAIN,IAAI;CACJ,IAAI,KAAK,QAQP,SAAS,MAAM,eAAe,MAAM;EANlC,GAAG,KAAK;EACR,YAAY,KAAK,OAAO;EACxB,OAAO,aAAa,QAAQ,WAAW;EACvC,aAAa,IAAI,KAAK,IAAI,CAAC,CAAC,CAAC,YAAY;EACzC,qBAAqB,uBAAuB,KAAA;CAED,CAAC;CAGhD,MAAM,UAAU,IAAI,KAAK,IAAI,CAAC,CAAC,CAAC,YAAY;CAE5C,OAAO;EACL,YAAY,KAAK;EACjB;EACA;EACA;EACA;EACA;EACA;EACA;EACA;CACF;AACF;AAIA,IAAM,qBAAN,cAAiC,MAAM;CACrC;CACA;CACA,YAAY,QAAmB,WAAgC;EAC7D,MAAM,QAAQ,OAAO,UAAU,GAAG,OAAO,WAAW,GAAG,OAAO,KAAK,WAAW,OAAO,QAAQ;EAC7F,KAAK,SAAS;EACd,KAAK,YAAY;CACnB;AACF;;;;;;;;;;;;;;;;AAiBA,eAAsB,cACpB,SACA,OACA,QACe;CACf,MAAM,WAAW,MAAM,QAAQ,WAAW,OAAO,KAAK;CACtD,IAAI,aAAa,KAAA,GAAW;CAC5B,IAAI,SAAS,WAAW,WAAW;CACnC,MAAM,QAAQ,SAAS,MAAM;AAC/B;AAEA,SAAS,sBAAsB,SAA6B;CAC1D,QAAQ,WAAmD;EACzD,IAAI,CAAC,SACH,MAAM,IAAI,MACR,6LACF;EAEF,OAAO,IAAI,0BAA0B,EACnC,KAAK,GAAG,QAAQ,cAAc,OAAO,QACvC,CAAC;CACH;AACF;AAEA,eAAe,wBACb,OAC2B;CAC3B,IAAI,mBAAmB,KAAK,GAAG;EAC7B,IAAI,CAAE,MAAM,uBAAuB,KAAK,GACtC,MAAM,IAAI,MAAM,iEAAiE;EAEnF,OAAO;CACT;CACA,OAAO,sBAAsB,KAAK;AACpC;AAEA,SAAS,mBACP,OAC2B;CAC3B,OAAO,mBAAmB,SAAS,YAAY;AACjD;AAEA,SAAS,6BACP,SACA,OACA,YACM;CACN,IAAI,QAAQ,UAAU,KAAA,KAAa,QAAQ,UAAU,OACnD,MAAM,IAAI,MACR,wCAAwC,QAAQ,MAAM,kCAAkC,MAAM,EAChG;CAEF,IAAI,QAAQ,eAAe,KAAA,KAAa,QAAQ,eAAe,YAC7D,MAAM,IAAI,MACR,6CAA6C,QAAQ,WAAW,uCAAuC,WAAW,EACpH;AAEJ;AAEA,SAAS,aAAa,QAAuC;CAG3D,MAAM,OAAO,GAAG,OAAO,WAAW,IAAI,OAAO,UAAU,IAAI,OAAO,WAAW,IAAI,OAAO;CAExF,IAAI,KAAK;CACT,IAAI,KAAK;CACT,KAAK,IAAI,IAAI,GAAG,IAAI,KAAK,QAAQ,KAAK;EACpC,MAAM,IAAI,KAAK,WAAW,CAAC;EAC3B,KAAK,KAAK,KAAK,KAAK,GAAG,QAAU,MAAM;EACvC,KAAK,KAAK,KAAK,KAAK,GAAG,UAAU,MAAM;CACzC;CACA,OAAO,OAAO,GAAG,SAAS,EAAE,CAAC,CAAC,SAAS,GAAG,GAAG,IAAI,GAAG,SAAS,EAAE,CAAC,CAAC,SAAS,GAAG,GAAG;AAClF"}
+{"version":3,"file":"eval-campaign-aMG73hvX.js","names":[],"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  RunCostProvenance,\n  RunJudgeMetadata,\n  RunOutcome,\n  RunRecord,\n  RunSplitTag,\n  RunTaskFailure,\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 { 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\ninterface CampaignRunOutcomeFields {\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  /** Cost in USD, or null when the runner could not capture it. */\n  costUsd: number | null\n  /** Source of the cost amount. */\n  costProvenance: RunCostProvenance\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  /** 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\n/** Campaign result with the same task-failure invariant as `RunRecord`. */\nexport type CampaignRunOutcome = CampaignRunOutcomeFields & RunTaskFailure\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        costProvenance: outcome.costProvenance,\n        tokenUsage: outcome.tokenUsage,\n        terminalOutcome: 'succeeded',\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,MAAM,oBAA8C;CAClD,aAAa;CACb,8BAA8B;CAC9B,gBAAgB;AAClB;AAEA,MAAM,gBAAsC;CAC1C,wBAAwB;CACxB,aAAa;AACf;AAEA,eAAsB,gBACpB,MAC6B;CAE7B,eAAe,KAAK,SAAS,KAAK,qBAAqB,aAAa;CAEpE,IAAI,KAAK,SAAS,WAAW,GAC3B,MAAM,IAAI,MAAM,8CAA8C;CAEhE,IAAI,KAAK,UAAU,WAAW,GAC5B,MAAM,IAAI,MAAM,+CAA+C;CAEjE,MAAM,6BAAa,IAAI,IAAY;CACnC,KAAK,MAAM,KAAK,KAAK,UAAU;EAC7B,IAAI,WAAW,IAAI,EAAE,EAAE,GACrB,MAAM,IAAI,MAAM,0CAA0C,EAAE,GAAG,GAAG;EAEpE,WAAW,IAAI,EAAE,EAAE;CACrB;CACA,MAAM,8BAAc,IAAI,IAAY;CACpC,KAAK,MAAM,KAAK,KAAK,WAAW;EAC9B,IAAI,YAAY,IAAI,EAAE,UAAU,GAC9B,MAAM,IAAI,MAAM,0CAA0C,EAAE,WAAW,GAAG;EAE5E,YAAY,IAAI,EAAE,UAAU;CAC9B;CACA,IAAI,KAAK,QAAQ,cAAc,CAAC,WAAW,IAAI,KAAK,OAAO,UAAU,GACnE,MAAM,IAAI,MACR,uCAAuC,KAAK,OAAO,WAAW,iCAChE;CAEF,IAAI,CAAC,KAAK,WACR,MAAM,IAAI,MAAM,oEAAoE;CAGtF,MAAM,QAAQ,KAAK,SAAS;EAAC;EAAG;EAAG;CAAC;CACpC,MAAM,WAAwB,KAAK,YAAY;CAC/C,MAAM,cAAc,KAAK,IAAI,GAAG,KAAK,eAAe,CAAC;CACrD,MAAM,YAAY;EAAE,GAAG;EAAmB,GAAI,KAAK,aAAa,CAAC;CAAG;CACpE,MAAM,qBAA8C,KAAK,sBAAsB;CAC/E,MAAM,MAAM,KAAK,cAAc,KAAK,IAAI;CACxC,MAAM,WAAW,KAAK,QAAQ,WAAW,GAAA,CAAI,QAAQ,QAAQ,EAAE;CAC/D,MAAM,WAAW,KAAK,QAAQ,YAAY;CAC1C,MAAM,sBAAsB,KAAK,uBAAuB;CAExD,MAAM,iBAAiB,KAAK,kBAAkB,sBAAsB,KAAK,OAAO;CAGhF,MAAM,sBAAsB,MAAM,SAChC,aAAa;EACX,YAAY,KAAK;EACjB,UAAU,KAAK,SAAS,KAAK,MAAM,EAAE,EAAE,CAAC,CAAC,KAAK;EAC9C,WAAW,KAAK,UAAU,KAAK,MAAM,EAAE,UAAU,CAAC,CAAC,KAAK;EACxD,OAAO,CAAC,GAAG,KAAK,CAAC,CAAC,MAAM,GAAG,MAAM,IAAI,CAAC;EACtC;EACA,YAAY,KAAK,QAAQ,cAAc;EACvC;EACA;EACA;CACF,CAAC,CACH;CAIA,MAAM,QAAgB,CAAC;CACvB,KAAK,MAAM,WAAW,KAAK,UACzB,KAAK,MAAM,YAAY,KAAK,WAC1B,KAAK,MAAM,QAAQ,OACjB,MAAM,KAAK;EAAE;EAAS;EAAU;CAAK,CAAC;CAK5C,MAAM,YAAY,IAAI,KAAK,IAAI,CAAC,CAAC,CAAC,YAAY;CAC9C,MAAM,OAAoB,CAAC;CAC3B,MAAM,mBAAyC,CAAC;CAChD,MAAM,aAA0B,CAAC;CASjC,IAAI,SAAS;CACb,IAAI,WAAW;CACf,MAAM,gBAA2B,CAAC;CAKlC,MAAM,2BAAW,IAAI,IAA0B;CAE/C,eAAe,SAAwB;EACrC,OAAO,CAAC,UAAU;GAChB,MAAM,IAAI;GACV,IAAI,KAAK,MAAM,QAAQ;GACvB,MAAM,OAAO,MAAM;GACnB,IAAI;IACF,MAAM,SAAS,MAAM,WAAW,IAAI;IACpC,KAAK,KAAK,OAAO,MAAM;IACvB,iBAAiB,KAAK,OAAO,SAAS;GACxC,SAAS,KAAK;IACZ,IAAI,eAAe,oBAAoB;KACrC,WAAW,KAAK,IAAI,MAAM;KAC1B,IAAI,IAAI,WAAW,iBAAiB,KAAK,IAAI,SAAS;IACxD,OAAO;KAKL,cAAc,KAAK,GAAG;KACtB,WAAW;KACX;IACF;GACF;EACF;CACF;CAEA,eAAe,WACb,MAC+D;EAC/D,MAAM,SAAS,KAAK,SAAS,aAAA,CAAc;GACzC,YAAY,KAAK;GACjB,OAAO;GACP,WAAW,KAAK,QAAQ;GACxB,YAAY,KAAK,SAAS;GAC1B,MAAM,KAAK;EACb,CAAC;EACD,MAAM,gBAAuC;GAC3C,YAAY,KAAK;GACjB;GACA,WAAW,KAAK,QAAQ;GACxB,YAAY,KAAK,SAAS;GAC1B,MAAM,KAAK;EACb;EACA,MAAM,QAAQ,KAAK,aAAa,aAAa;EAC7C,MAAM,UAAU,eAAe,aAAa;EAE5C,MAAM,UAAU,IAAI,aAAa,OAAO;GACtC;GACA,KAAK,KAAK;GACV,eAAe,KAAK;EACtB,CAAC;EAGD,SAAS,IAAI,OAAO,OAAO;EAE3B,MAAM,UAA4B;GAChC,GAAG,KAAK;GACR;GACA,cAAc,EAAE,MAAM;EACxB;EAEA,MAAM,MAA6B;GACjC;GACA,cAAc,KAAK;GACnB,SAAS,KAAK,QAAQ;GACtB,WAAW,KAAK,QAAQ;GACxB,YAAY,KAAK,SAAS;GAC1B,cAAc,KAAK,SAAS,QAAQ,CAAC;GACrC,MAAM,KAAK;GACX;GACA;GACA;GACA;GACA;EACF;EAEA,IAAI;GACF,MAAM,YAAY,IAAI;GACtB,IAAI;GACJ,IAAI;IACF,UAAU,MAAM,KAAK,OAAO,GAAG;GACjC,SAAS,KAAK;IACZ,MAAM,UAAU,eAAe,QAAQ,IAAI,UAAU,OAAO,GAAG;IAM/D,MAAM,cAAc,SAAS,OAAO,OAAO;IAC3C,MAAM,IAAI,mBAAmB;KAC3B;KACA,WAAW,KAAK,QAAQ;KACxB,YAAY,KAAK,SAAS;KAC1B,MAAM,KAAK;KACX,QAAQ;KACR,OAAO;IACT,CAAC;GACH;GACA,MAAM,SAAS,IAAI,IAAI;GAEvB,MAAM,kBAAkB,MAAM,kBAAkB,OAAO,OAAO;IAAE,GAAG;IAAW;GAAQ,CAAC;GACvF,IAAI,CAAC,gBAAgB,IACnB,QAAQ,oBAAR;IACE,KAAK,SACH,MAAM,IAAI,kBAAkB,eAAe;IAC7C,KAAK,eACH,MAAM,IAAI,mBACR;KACE;KACA,WAAW,KAAK,QAAQ;KACxB,YAAY,KAAK,SAAS;KAC1B,MAAM,KAAK;KACX,QAAQ;KACR,OAAO,gBAAgB,OAAO,KAAK,MAAM,EAAE,IAAI,CAAC,CAAC,KAAK,IAAI;IAC5D,GACA,eACF;IACF,KAAK,OAEH;GACJ;GAGF,MAAM,gBAA4B,EAChC,KAAK,QAAQ,OAAO,CAAC,EACvB;GACA,IAAI,aAAa,WAAW,cAAc,eAAe,QAAQ;QAC5D,cAAc,cAAc,QAAQ;GACzC,IAAI,QAAQ,gBAAgB,KAAA,GAAW,cAAc,cAAc,QAAQ;GAE3E,MAAM,SAAoB;IACxB;IACA,cAAc,KAAK;IACnB,aAAa,KAAK,QAAQ;IAC1B,MAAM,KAAK;IACX,OAAO,QAAQ;IACf,YAAY,QAAQ;IACpB,YAAY,QAAQ;IACpB,WAAW,KAAK;IAChB;IACA,SAAS,QAAQ;IACjB,gBAAgB,QAAQ;IACxB,YAAY,QAAQ;IACpB,iBAAiB;IACjB,eAAe,QAAQ;IACvB,SAAS;IACT,GAAI,QAAQ,eAAe,EAAE,cAAc,QAAQ,aAAa,IAAI,CAAC;IACrE,aAAa,QAAQ;IACrB;IACA,YAAY,KAAK,SAAS;GAC5B;GACA,MAAM,gBACJ,QAAQ,iBACP,OAAO,KAAK,iBAAiB,aAC1B,MAAM,KAAK,aAAa;IACtB,YAAY,KAAK;IACjB;IACA,WAAW,KAAK,QAAQ;IACxB,YAAY,KAAK,SAAS;IAC1B,MAAM,KAAK;IACX,SAAS,KAAK,QAAQ;IACtB,cAAc,KAAK,SAAS,QAAQ,CAAC;GACvC,CAAC,IACD,KAAK;GACX,IAAI,kBAAkB,KAAA,GAAW;IAC/B,MAAM,eAAe,MAAM,wBAAwB,aAAa;IAChE,6BAA6B,cAAc,QAAQ,OAAO,QAAQ,UAAU;IAC5E,OAAO,eAAe;GACxB;GACA,OAAO;IAAE,QAAQ,kBAAkB,MAAM;IAAG,WAAW;GAAgB;EACzE,UAAU;GAIR,SAAS,OAAO,KAAK;EACvB;CACF;CAEA,MAAM,UAAU,MAAM,KAAK,EAAE,QAAQ,KAAK,IAAI,aAAa,MAAM,MAAM,EAAE,SAAS,OAAO,CAAC;CAI1F,MAAM,QAAQ,WAAW,OAAO;CAMhC,KAAK,MAAM,CAAC,OAAO,YAAY,UAC7B,MAAM,cAAc,SAAS,OAAO,kDAAkD;CAExF,SAAS,MAAM;CAEf,IAAI,cAAc,SAAS,GACzB,MAAM,cAAc,WAAW,IAC3B,cAAc,KACd,IAAI,eACF,eACA,oBAAoB,cAAc,OAAO,iDAC3C;CAIN,IAAI;CACJ,IAAI,KAAK,QAQP,SAAS,MAAM,eAAe,MAAM;EANlC,GAAG,KAAK;EACR,YAAY,KAAK,OAAO;EACxB,OAAO,aAAa,QAAQ,WAAW;EACvC,aAAa,IAAI,KAAK,IAAI,CAAC,CAAC,CAAC,YAAY;EACzC,qBAAqB,uBAAuB,KAAA;CAED,CAAC;CAGhD,MAAM,UAAU,IAAI,KAAK,IAAI,CAAC,CAAC,CAAC,YAAY;CAE5C,OAAO;EACL,YAAY,KAAK;EACjB;EACA;EACA;EACA;EACA;EACA;EACA;EACA;CACF;AACF;AAIA,IAAM,qBAAN,cAAiC,MAAM;CACrC;CACA;CACA,YAAY,QAAmB,WAAgC;EAC7D,MAAM,QAAQ,OAAO,UAAU,GAAG,OAAO,WAAW,GAAG,OAAO,KAAK,WAAW,OAAO,QAAQ;EAC7F,KAAK,SAAS;EACd,KAAK,YAAY;CACnB;AACF;;;;;;;;;;;;;;;;AAiBA,eAAsB,cACpB,SACA,OACA,QACe;CACf,MAAM,WAAW,MAAM,QAAQ,WAAW,OAAO,KAAK;CACtD,IAAI,aAAa,KAAA,GAAW;CAC5B,IAAI,SAAS,WAAW,WAAW;CACnC,MAAM,QAAQ,SAAS,MAAM;AAC/B;AAEA,SAAS,sBAAsB,SAA6B;CAC1D,QAAQ,WAAmD;EACzD,IAAI,CAAC,SACH,MAAM,IAAI,MACR,6LACF;EAEF,OAAO,IAAI,0BAA0B,EACnC,KAAK,GAAG,QAAQ,cAAc,OAAO,QACvC,CAAC;CACH;AACF;AAEA,eAAe,wBACb,OAC2B;CAC3B,IAAI,mBAAmB,KAAK,GAAG;EAC7B,IAAI,CAAE,MAAM,uBAAuB,KAAK,GACtC,MAAM,IAAI,MAAM,iEAAiE;EAEnF,OAAO;CACT;CACA,OAAO,sBAAsB,KAAK;AACpC;AAEA,SAAS,mBACP,OAC2B;CAC3B,OAAO,mBAAmB,SAAS,YAAY;AACjD;AAEA,SAAS,6BACP,SACA,OACA,YACM;CACN,IAAI,QAAQ,UAAU,KAAA,KAAa,QAAQ,UAAU,OACnD,MAAM,IAAI,MACR,wCAAwC,QAAQ,MAAM,kCAAkC,MAAM,EAChG;CAEF,IAAI,QAAQ,eAAe,KAAA,KAAa,QAAQ,eAAe,YAC7D,MAAM,IAAI,MACR,6CAA6C,QAAQ,WAAW,uCAAuC,WAAW,EACpH;AAEJ;AAEA,SAAS,aAAa,QAAuC;CAG3D,MAAM,OAAO,GAAG,OAAO,WAAW,IAAI,OAAO,UAAU,IAAI,OAAO,WAAW,IAAI,OAAO;CAExF,IAAI,KAAK;CACT,IAAI,KAAK;CACT,KAAK,IAAI,IAAI,GAAG,IAAI,KAAK,QAAQ,KAAK;EACpC,MAAM,IAAI,KAAK,WAAW,CAAC;EAC3B,KAAK,KAAK,KAAK,KAAK,GAAG,QAAU,MAAM;EACvC,KAAK,KAAK,KAAK,KAAK,GAAG,UAAU,MAAM;CACzC;CACA,OAAO,OAAO,GAAG,SAAS,EAAE,CAAC,CAAC,SAAS,GAAG,GAAG,IAAI,GAAG,SAAS,EAAE,CAAC,CAAC,SAAS,GAAG,GAAG;AAClF"}

package/dist/fuzz.js

@@ -1,5 +1,5 @@
 import { s as ValidationError } from "./errors-8YnH8WlF.js";
-import { i as CostLedger, o as CostReceiptCaptureError } from "./cost-ledger-ZAa_P4r0.js";
+import { i as CostLedger, o as CostReceiptCaptureError } from "./cost-ledger-DHAjwNj7.js";
 import { r as varianceBasedCurriculum } from "./active-curriculum-C4mk67HP.js";
 //#region src/fuzz/cube.ts
 /** Enumerate every input cell (cartesian product of the axes), in stable order. */

package/dist/index.js

@@ -2,17 +2,17 @@ import { t as __exportAll } from "./rolldown-runtime-8H4AJuhK.js";
 import { _ as observeAll, a as jsonlReviewStore, c as scoreFromEvals, d as runAgentControlLoop, f as stopOnNoProgress, g as noProgressDetector, h as errorStreakDetector, i as inMemoryReviewStore, l as allCriticalPassed, m as subjectiveEval, n as runProposeReviewAsControlLoop, o as runProposeReview, p as stopOnRepeatedAction, r as createLlmReviewer, s as controlRunToRunRecord, t as controlFailureClassFromVerification, u as objectiveEval, v as repeatedActionDetector, y as evaluateActionPolicy } from "./propose-review-control-SQ-n9-We.js";
 import { a as NotFoundError, c as VerificationError, i as JudgeError, n as CaptureIntegrityError, o as ReplayError, r as ConfigError, s as ValidationError, t as AgentEvalError } from "./errors-8YnH8WlF.js";
 import { _ as verifyManifest, a as assertRunAgentProfileCell, c as groupRunsByAgentProfileCell, d as validateAgentProfileCell, f as verifyAgentProfileCell, g as signManifest, h as hashJson, i as agentProfileCellKey, l as requireAgentProfileCell, m as evaluateHypothesis, n as AgentProfileCellValidationError, o as buildAgentInterfaceProfileCell, p as canonicalize, r as agentProfileCellHashMaterial, s as buildAgentProfileCell, t as AGENT_PROFILE_KINDS, u as toAgentProfileJson } from "./agent-profile-cell-OhuTee9n.js";
-import { F as createChatClient, I as createAnalystAi, P as computeTraceMetrics, a as KNOWLEDGE_GAP_KIND_SPEC, d as renderUpstreamFindings, i as KNOWLEDGE_POISONING_KIND_SPEC, l as createTraceAnalystKind, n as AnalystRegistry, o as IMPROVEMENT_KIND_SPEC, r as DEFAULT_TRACE_ANALYST_KINDS, s as FAILURE_MODE_KIND_SPEC, t as buildDefaultAnalystRegistry, u as renderPriorFindings } from "./default-registry-CFUZyNeZ.js";
+import { F as createChatClient, I as createAnalystAi, P as computeTraceMetrics, a as KNOWLEDGE_GAP_KIND_SPEC, d as renderUpstreamFindings, i as KNOWLEDGE_POISONING_KIND_SPEC, l as createTraceAnalystKind, n as AnalystRegistry, o as IMPROVEMENT_KIND_SPEC, r as DEFAULT_TRACE_ANALYST_KINDS, s as FAILURE_MODE_KIND_SPEC, t as buildDefaultAnalystRegistry, u as renderPriorFindings } from "./default-registry-BAhV-lbE.js";
 import { a as estimateTokens, i as estimateCost, n as MetricsCollector, o as isModelPriced, r as TokenCounter, s as resolveModelPricing, t as MODEL_PRICING } from "./metrics-C9YY1OcL.js";
-import { a as CostLedgerPersistenceError, c as costForTokenPricing, i as CostLedger, l as costForUsage, n as CostCallConflictError, o as CostReceiptCaptureError, r as CostCeilingReachedError, s as CostReservationExceededError, t as CostAccountingIncompleteError, u as modelPriceKey } from "./cost-ledger-ZAa_P4r0.js";
+import { a as CostLedgerPersistenceError, c as costForTokenPricing, i as CostLedger, l as costForUsage, n as CostCallConflictError, o as CostReceiptCaptureError, r as CostCeilingReachedError, s as CostReservationExceededError, t as CostAccountingIncompleteError, u as modelPriceKey } from "./cost-ledger-DHAjwNj7.js";
 import { a as providerFromBaseUrl, i as defaultProviderRedactor, n as InMemoryRawProviderSink, r as NoopRawProviderSink, t as FileSystemRawProviderSink } from "./raw-provider-sink-BQd7mzyT.js";
-import { a as assertLlmRoute, c as callLlmJson, d as isTransientLlmError, f as maximumChargeForLlmRequest, i as LlmRouteAssertionError, l as costReceiptFromLlm, m as stripFencedJson, n as LlmClient, o as backoffMs, p as probeLlm, r as LlmResponseError, s as callLlm, t as LlmCallError, u as costReceiptFromLlmError } from "./llm-client-BNcP4v08.js";
+import { a as assertLlmRoute, c as callLlmJson, d as isTransientLlmError, f as maximumChargeForLlmRequest, i as LlmRouteAssertionError, l as costReceiptFromLlm, m as stripFencedJson, n as LlmClient, o as backoffMs, p as probeLlm, r as LlmResponseError, s as callLlm, t as LlmCallError, u as costReceiptFromLlmError } from "./llm-client-DHx8pzyJ.js";
 import { INPUT_VALUE, LLM_CACHED_TOKENS, LLM_CACHED_TOKEN_ATTR_KEYS, LLM_CACHE_WRITE_TOKENS, LLM_CACHE_WRITE_TOKEN_ATTR_KEYS, LLM_CONTEXT_TOKENS, LLM_COST_ATTR_KEYS, LLM_COST_USD, LLM_INPUT_TOKENS, LLM_INPUT_TOKEN_ATTR_KEYS, LLM_MODEL_ATTR_KEYS, LLM_MODEL_NAME, LLM_OUTPUT_TOKENS, LLM_OUTPUT_TOKEN_ATTR_KEYS, LLM_REASONING_TOKENS, LLM_REASONING_TOKEN_ATTR_KEYS, OPENINFERENCE_SPAN_KIND, OUTPUT_VALUE, RUN_COST_ATTR_KEYS, SPAN_KIND_ATTR_KEYS, TOOL_ARGS_CAPTURED, TOOL_LATENCY_MS, TOOL_NAME, TOOL_NAME_ATTR_KEYS, applyLlmSpanOtlpAttributes, asNumber, contextInputTokens, firstNumberAttr } from "./trace-attributes.js";
 import { S as traceSpanKindToOpenInferenceKind, a as SpanNotFoundError, b as classifyOtlpSpanRole, c as DEFAULT_TRACE_ANALYST_BUDGETS, f as extractOtlpAttributes, g as readOtlpStatus, h as projectOtlpFlatLine, i as OtlpFileTraceStore, l as TRACE_ANALYST_TRUNCATION_MARKER_PREFIX, m as inferOtlpKind, n as traceAnalystFunctionGroup, o as TraceFileMissingError, p as firstStringAttr, s as TraceNotFoundError, t as buildTraceAnalystTools, u as asString, v as stringField, x as isOtlpModelCall, y as applyToolSpanOtlpAttributes } from "./tools-BmuN627J.js";
 import { a as clamp01, c as makeFinding, i as aggregateRunScore, l as makeProposalFinding, r as DEFAULT_RUN_SCORE_WEIGHTS, s as computeFindingId } from "./proposal-findings-DCawte-y.js";
 import { n as mapConcurrent, t as Mutex } from "./concurrency-MUjT7VjM.js";
-import { a as RunCritic, d as defaultIsMaterial, f as diffFindings, i as runSemanticConceptJudge, n as SEMANTIC_CONCEPT_JUDGE_VERSION, o as SKILL_USAGE_ANALYST, p as LockedJsonlAppender, r as createSemanticConceptJudge, s as SkillUsageAnalyst, t as DEFAULT_COMPLEXITY_WEIGHTS, u as FindingsStore } from "./semantic-concept-judge-Ca--u10C.js";
-import { At as dominates, B as surfaceContentHash, Bt as summarizeAgentReceiptIntegrity, Ct as llmJudge, D as runCanaries, Dt as fileVerdictCache, Et as contentHash, Ft as minimumPairsForPairedDeltaTest, G as DEFAULT_MUTATION_PRIMITIVES, Ht as JudgeParseError, It as pairedDeltaTest, K as buildReflectionPrompt, Lt as BackendIntegrityError, Mt as paretoFrontierWithCrowding, Nt as scalarScore, Ot as inMemoryVerdictCache, Rt as assertRealAgentReceipts, St as hashScenarios, Tt as canonicalJson, Vt as summarizeBackendIntegrity, _t as redTeamReport, bt as Dataset, dt as REFERENCE_EQUIVALENCE_INPUT_LIMITS, ft as REFERENCE_EQUIVALENCE_JUDGE_VERSION, gt as redTeamDataset, ht as DEFAULT_RED_TEAM_CORPUS, jt as paretoFrontier, kt as crowdingDistance, mt as runReferenceEquivalenceJudge, pt as createReferenceEquivalenceJudge, q as parseReflectionResponse, vt as scoreRedTeamOutput, wt as cachedJudge, xt as HoldoutLockedError, yt as toolNamesForRun, zt as assertRealBackend } from "./skillopt-optimization-method-Cl4XPkLC.js";
+import { a as RunCritic, d as defaultIsMaterial, f as diffFindings, i as runSemanticConceptJudge, n as SEMANTIC_CONCEPT_JUDGE_VERSION, o as SKILL_USAGE_ANALYST, p as LockedJsonlAppender, r as createSemanticConceptJudge, s as SkillUsageAnalyst, t as DEFAULT_COMPLEXITY_WEIGHTS, u as FindingsStore } from "./semantic-concept-judge-Btozx3Vc.js";
+import { At as dominates, B as surfaceContentHash, Bt as summarizeAgentReceiptIntegrity, Ct as llmJudge, D as runCanaries, Dt as fileVerdictCache, Et as contentHash, Ft as minimumPairsForPairedDeltaTest, G as DEFAULT_MUTATION_PRIMITIVES, Ht as JudgeParseError, It as pairedDeltaTest, K as buildReflectionPrompt, Lt as BackendIntegrityError, Mt as paretoFrontierWithCrowding, Nt as scalarScore, Ot as inMemoryVerdictCache, Rt as assertRealAgentReceipts, St as hashScenarios, Tt as canonicalJson, Vt as summarizeBackendIntegrity, _t as redTeamReport, bt as Dataset, dt as REFERENCE_EQUIVALENCE_INPUT_LIMITS, ft as REFERENCE_EQUIVALENCE_JUDGE_VERSION, gt as redTeamDataset, ht as DEFAULT_RED_TEAM_CORPUS, jt as paretoFrontier, kt as crowdingDistance, mt as runReferenceEquivalenceJudge, pt as createReferenceEquivalenceJudge, q as parseReflectionResponse, vt as scoreRedTeamOutput, wt as cachedJudge, xt as HoldoutLockedError, yt as toolNamesForRun, zt as assertRealBackend } from "./skillopt-optimization-method-C9yGg-4C.js";
 import { A as passAtK, B as studentTCdf, C as pairedBootstrap, D as pairedSignTest, E as pairedRiskDifference, F as spearmanR, G as continuousAgreement, H as normalCdf, I as weightedComposite, J as verbosityBias, K as positionalBias, L as weightedMean, M as ranks, N as requiredPairedSampleSize, O as pairedTTest, P as requiredSampleSize, R as wilcoxonSignedRank, S as normalizeScores, T as pairedMde, U as calibrateJudge, W as calibrateJudgeContinuous, _ as mannWhitneyU, a as WILCOXON_EXACT_MAX_N, b as mcnemarRequiredN, c as cliffsDelta, d as corpusInterRaterAgreement, f as corpusInterRaterAgreementFromJudgeScores, g as interpretCliffs, h as interRaterReliability, i as MANN_WHITNEY_EXACT_MAX_WORK, j as pearsonR, k as partialCredit, l as cohensD, m as holm, n as DEFAULT_PERMUTATIONS, o as benjaminiHochberg, p as eProcess, q as selfPreference, r as MANN_WHITNEY_EXACT_MAX_STATES, s as bonferroni, t as BOOTSTRAP_GATE_MIN_N, u as confidenceInterval, v as mcnemar, w as pairedCohensDz, x as mulberry32, y as mcnemarPower, z as wilson } from "./statistics-RwRNu2__.js";
 import { n as pairArms, r as pairRunRecords, t as comparePairedArms } from "./paired-arms-CA_8pN01.js";
 import { n as llmSpanFromProvider, t as TraceEmitter } from "./emitter-CPBAhxum.js";
@@ -29,7 +29,7 @@ import { D as showMeasured, E as isUnavailable, S as rollupSupervisorRuns, T as
 import { n as TRACE_ANALYST_ACTOR_DESCRIPTION, r as TRACE_ANALYST_ACTOR_DESCRIPTION_VERSION, t as analyzeTraces } from "./analyst-LsnNpSkm.js";
 import { C as planTraceInsightQuestions, E as traceAnalystOnRunComplete, S as inferDomainKeywords, T as tokenizeDomainWords, _ as buildTraceInsightContext, a as convertTraceStoresToOtlp, b as describeTraceInsightScope, c as otelRunCompleteHook, d as captureFetchToRawSink, f as otlpRowsToRunRecords, g as flattenOtlpExportToNdjson, h as otlpToTraceRunRecords, i as iterateRawCalls, l as OTEL_AGENT_EVAL_SCOPE, m as otlpToRunRecords, n as ReplayCacheMissError, o as createOtelExporter, p as otlpRowsToTraceRunRecords, r as createReplayFetch, s as createOtelTracingStore, t as ReplayCache, u as exportRunAsOtlp, v as buildTraceInsightPrompt, w as scoreTraceInsightReadiness, x as domainEvidencePattern, y as defaultTraceInsightPanel } from "./replay-CJfGLdx4.js";
 import { n as extractUsageFromResponse, r as extractUsageFromSse, t as extractUsage } from "./extract-usage-2j25whHw.js";
-import { B as harnessAxisOf, F as HARNESS_NATIVE_MODEL, G as parseCorrectnessResponse, H as completionVerdict, I as agentProfileHash, K as verifyCompletion, L as agentProfileId, P as CODING_HARNESSES, R as agentProfileModelId, U as createLlmCorrectnessChecker, V as extractProducedState, W as createTokenRecallChecker, z as expandProfileAxes } from "./campaign-B1c1T0kv.js";
+import { B as harnessAxisOf, F as HARNESS_NATIVE_MODEL, G as parseCorrectnessResponse, H as completionVerdict, I as agentProfileHash, K as verifyCompletion, L as agentProfileId, P as CODING_HARNESSES, R as agentProfileModelId, U as createLlmCorrectnessChecker, V as extractProducedState, W as createTokenRecallChecker, z as expandProfileAxes } from "./campaign-B0oyngTs.js";
 import { n as iqr, r as welchsTTest, t as compareToBaseline } from "./baseline-BaPxoROc.js";
 import { a as judgeSpans, c as runsForScenario, i as hasCapturedToolArgs, l as toolSpans, n as argHash, o as llmSpans, r as groupBy, s as runFailureClass, t as aggregateLlm } from "./query-Di7eEQ79.js";
 import { a as checkBehavioralCanary, i as canaryLeakView, o as checkCanaries, r as HoldoutAuditor, s as runBehavioralCanaries, t as analyzeRuns } from "./analyze-runs-qk8op0tN.js";
@@ -39,8 +39,8 @@ import { a as InMemoryTraceStore, i as FileSystemTraceStore, n as assertRunCaptu
 import { i as redactValue, n as REDACTION_VERSION, r as redactString, t as DEFAULT_REDACTION_RULES } from "./redact-7Aq1ukl-.js";
 import { n as DEFAULT_RULES, r as classifyFailure, t as computeToolUseMetrics } from "./tool-use-metrics-DEGMKycK.js";
 import { t as analyzeSeries } from "./series-convergence-CjO2QdRW.js";
-import { _ as deterministicSplit, g as BENCHMARK_SPLIT_SEED, t as benchmarks_exports } from "./benchmarks-Dw1Wv_JQ.js";
-import { t as runEvalCampaign } from "./eval-campaign-CHFxPTVl.js";
+import { _ as deterministicSplit, g as BENCHMARK_SPLIT_SEED, t as benchmarks_exports } from "./benchmarks-DwFrCrCS.js";
+import { t as runEvalCampaign } from "./eval-campaign-aMG73hvX.js";
 import { n as pairedEvalueSequence, t as evaluateInterimReleaseConfidence } from "./sequential-Br0mAPHA.js";
 import { AxGEPA, AxSignature, ax } from "@ax-llm/ax";
 import { accessSync, appendFileSync, constants, cpSync, existsSync, mkdirSync, promises, readFileSync, readdirSync, statSync, writeFileSync } from "node:fs";

package/dist/{llm-client-BNcP4v08.js → llm-client-DHx8pzyJ.js}

@@ -1,5 +1,5 @@
 import { n as CaptureIntegrityError, t as AgentEvalError } from "./errors-8YnH8WlF.js";
-import { c as costForTokenPricing } from "./cost-ledger-ZAa_P4r0.js";
+import { c as costForTokenPricing } from "./cost-ledger-DHAjwNj7.js";
 import { a as providerFromBaseUrl, i as defaultProviderRedactor } from "./raw-provider-sink-BQd7mzyT.js";
 //#region src/llm-client.ts
 /**
@@ -684,4 +684,4 @@ var LlmClient = class {
 //#endregion
 export { assertLlmRoute as a, callLlmJson as c, isTransientLlmError as d, maximumChargeForLlmRequest as f, LlmRouteAssertionError as i, costReceiptFromLlm as l, stripFencedJson as m, LlmClient as n, backoffMs as o, probeLlm as p, LlmResponseError as r, callLlm as s, LlmCallError as t, costReceiptFromLlmError as u };
 
-//# sourceMappingURL=llm-client-BNcP4v08.js.map
+//# sourceMappingURL=llm-client-DHx8pzyJ.js.map

package/dist/{llm-client-BNcP4v08.js.map → llm-client-DHx8pzyJ.js.map}

@@ -1 +1 @@
-{"version":3,"file":"llm-client-BNcP4v08.js","names":[],"sources":["../src/llm-client.ts"],"sourcesContent":["/**\n * LLM client with graceful degrade.\n *\n * OpenAI-compatible `/v1/chat/completions` client with:\n *   - Exponential-backoff retry on 429 + 5xx gateway errors (502/503/504).\n *   - Retry on transient network errors (fetch failed, AbortError, ECONNRESET).\n *   - One retry at temperature 1 when a model explicitly requires it.\n *   - Graceful json_schema → json_object degrade on 400 with schema-reject body.\n *   - Fenced-JSON stripping (```json ... ```) for models that wrap structured output.\n *   - Configurable base URL + api key / bearer, works with LiteLLM proxies, OpenAI\n *     directly, cli-bridge subscriptions, and any router that speaks the spec.\n *\n * Usage:\n *   const { value, result } = await callLlmJson<MyType>(\n *     { model: 'gpt-4o', messages: [...], jsonSchema: { name: 'x', schema: {...} } },\n *     { baseUrl: 'https://router.tangle.tools/v1', apiKey: process.env.KEY },\n *   )\n *\n * `createChatClient` wraps this implementation for provider-neutral package\n * entry points. Direct callers can use `callLlm` or `callLlmJson`.\n */\n\nimport {\n  type CostReceiptInput,\n  type CustomTokenPricing,\n  costForTokenPricing,\n  type MaximumCharge,\n} from './cost-ledger'\nimport { AgentEvalError, CaptureIntegrityError } from './errors'\nimport {\n  defaultProviderRedactor,\n  type ProviderRedactor,\n  providerFromBaseUrl,\n  type RawProviderEvent,\n  type RawProviderSink,\n} from './trace/raw-provider-sink'\n\n// ─── Types ──────────────────────────────────────────────────────────────\n\nexport interface LlmMessage {\n  role: 'system' | 'user' | 'assistant'\n  /**\n   * Either a plain text content string OR a multimodal content array\n   * (text + image_url parts) for vision-capable models.\n   */\n  content:\n    | string\n    | Array<\n        | { type: 'text'; text: string }\n        | { type: 'image_url'; image_url: { url: string; detail?: 'auto' | 'low' | 'high' } }\n      >\n}\n\nexport type LlmThinkingMode = 'enabled' | 'disabled'\n\nexport interface LlmCallRequest {\n  model: string\n  messages: LlmMessage[]\n  /** Optional JSON-mode response format (response_format: json_object). */\n  jsonMode?: boolean\n  /** Optional structured output via JSON Schema. Falls back to json_object on 400. */\n  jsonSchema?: { name: string; schema: Record<string, unknown> }\n  temperature?: number\n  maxTokens?: number\n  /** OpenAI-compatible reasoning mode. Omitted when the provider default should apply. */\n  thinking?: LlmThinkingMode\n  /** Per-call timeout, default 300s. */\n  timeoutMs?: number\n}\n\n/** Conservative priced bound for the exact text request sent to a provider.\n * Returns undefined when output or multimodal input is not bounded, causing a\n * capped CostLedger to reject the call before execution. Pass\n * `customTokenPricing` when package pricing does not cover the model or endpoint. */\nexport function maximumChargeForLlmRequest(\n  request: Pick<LlmCallRequest, 'model' | 'messages' | 'jsonSchema' | 'maxTokens' | 'thinking'>,\n  options: LlmClientOptions = {},\n): MaximumCharge | undefined {\n  if (request.maxTokens === undefined) return undefined\n  if (!Number.isInteger(request.maxTokens) || request.maxTokens <= 0) {\n    throw new RangeError(`maximumChargeForLlmRequest: maxTokens must be a positive integer`)\n  }\n  if (\n    request.messages.some(\n      (message) =>\n        Array.isArray(message.content) && message.content.some((part) => part.type === 'image_url'),\n    )\n  ) {\n    return undefined\n  }\n\n  const attempts = resolveMaximumAttempts(options.maximumAttempts)\n  const forceJsonObject = options.jsonSchemaTransport === 'json-object'\n  // A byte-level tokenizer cannot emit more input tokens than request bytes.\n  // Pricing the complete body also covers role/schema framing omitted from content-only estimates.\n  const requestBytes = new TextEncoder().encode(\n    JSON.stringify(buildBody(request, forceJsonObject, options.thinking)),\n  ).byteLength\n  // A rejected response schema can trigger one JSON-mode batch with the same output limit.\n  const batches = request.jsonSchema && !forceJsonObject ? 2 : 1\n  const usage = {\n    inputTokens: requestBytes * attempts * batches,\n    outputTokens: request.maxTokens * attempts * batches,\n  }\n  return options.customTokenPricing\n    ? { customTokenPricing: options.customTokenPricing, ...usage }\n    : { model: request.model, ...usage }\n}\n\nexport interface LlmUsage {\n  promptTokens: number\n  completionTokens: number\n  totalTokens: number\n  /** False when the provider omitted or malformed prompt/completion usage. */\n  captured?: boolean\n  /** Reasoning-token subset of completionTokens, when reported. */\n  reasoningTokens?: number\n  /** Proxies populate this when prompt caching is on. */\n  cachedPromptTokens?: number\n}\n\nexport interface LlmCallResult {\n  /** The text content of the first choice. Empty string if none. */\n  content: string\n  usage: LlmUsage\n  /**\n   * Cost in USD. Uses the provider's reported cost when present, otherwise\n   * caller-supplied token pricing. `null` when neither is available.\n   */\n  costUsd: number | null\n  /** Model name actually used (echoed from response). */\n  model: string\n  /** Wall-clock duration of the HTTP call (last attempt, if retried). */\n  durationMs: number\n  /**\n   * `finish_reason` echoed from the first choice (`stop`, `length`,\n   * `content_filter`, `tool_calls`, ...). `null` when the provider omits it.\n   * Exposed so a free-form `callLlm` caller CAN detect a truncated answer\n   * (`length`) instead of treating a cut-off completion as complete. Note:\n   * `callLlm` does not itself reject on it — acting on this signal is the\n   * caller's responsibility (in-repo free-form drivers do not yet enforce it).\n   */\n  finishReason?: string | null\n  /**\n   * True when `content.trim()` is empty. An empty completion is a silent zero\n   * for free-form `callLlm` callers; this flag is the signal a caller can\n   * inspect to fail loud rather than proceed on an empty string. `callLlm`\n   * surfaces it but does not throw on it.\n   */\n  contentEmpty?: boolean\n  /** Raw response body. */\n  raw: Record<string, unknown>\n}\n\nexport type LlmCallMetadata = Pick<LlmCallResult, 'usage' | 'costUsd' | 'model' | 'durationMs'>\n\n/** Convert a provider result into the canonical paid-call receipt input. */\nexport function costReceiptFromLlm(\n  result: LlmCallResult,\n  customTokenPricing?: CustomTokenPricing,\n): CostReceiptInput {\n  const cachedTokens = result.usage.cachedPromptTokens ?? 0\n  const inputTokens = Math.max(0, result.usage.promptTokens - cachedTokens)\n  const configuredCostUsd =\n    result.costUsd === null && customTokenPricing && result.usage.captured !== false\n      ? costForTokenPricing(customTokenPricing, {\n          inputTokens,\n          ...(cachedTokens > 0 ? { cachedTokens } : {}),\n          outputTokens: result.usage.completionTokens,\n        })\n      : undefined\n  return {\n    model: result.model,\n    inputTokens,\n    outputTokens: result.usage.completionTokens,\n    reasoningTokens: result.usage.reasoningTokens,\n    cachedTokens: cachedTokens > 0 ? cachedTokens : undefined,\n    actualCostUsd: result.costUsd ?? configuredCostUsd,\n    usageUnknown: result.usage.captured === false,\n  }\n}\n\n/** Structured-response failures retain their completed provider receipt. */\nexport function costReceiptFromLlmError(\n  error: Error,\n  customTokenPricing?: CustomTokenPricing,\n): CostReceiptInput | undefined {\n  return error instanceof LlmResponseError\n    ? costReceiptFromLlm(error.result, customTokenPricing)\n    : undefined\n}\n\nexport class LlmCallError extends AgentEvalError {\n  constructor(\n    message: string,\n    public readonly status: number,\n    public readonly body: string,\n    public readonly model: string,\n  ) {\n    super('judge', message)\n  }\n}\n\n/** A provider response completed and incurred measurable usage, but its content\n *  could not satisfy the caller's response contract. The response envelope is\n *  retained so accounting can commit the receipt before the error propagates. */\nexport class LlmResponseError extends AgentEvalError {\n  constructor(\n    message: string,\n    public readonly result: LlmCallResult,\n    options?: { cause?: unknown },\n  ) {\n    super('judge', message, options)\n  }\n}\n\nexport interface LlmClientOptions {\n  /** Base URL (without trailing slash). Must end at the `/v1` prefix. */\n  baseUrl?: string\n  /** Bearer token — either `apiKey` or `bearer` populates `Authorization: Bearer ...`. */\n  apiKey?: string\n  bearer?: string\n  /** Override for the `Authorization` header (e.g. `X-Auth: ...`). Takes precedence over apiKey/bearer. */\n  authHeader?: { name: string; value: string }\n  /** Stable provider idempotency key, reused across retries of this logical call. */\n  idempotencyKey?: string\n  /** Default timeout in ms. Per-call can override. */\n  defaultTimeoutMs?: number\n  /**\n   * Caller-supplied abort signal — e.g. a campaign-wide cancel. Linked to\n   * each attempt's per-attempt timeout controller, so aborting it cancels\n   * the in-flight fetch. A caller abort is FATAL: it is not retried even\n   * though an AbortError otherwise matches the transient patterns.\n   */\n  signal?: AbortSignal\n  /**\n   * Cross-attempt wall-clock budget in ms, measured from the first attempt.\n   * Before launching each attempt the loop checks the remaining budget and\n   * stops retrying once it is exhausted, rather than waiting the full\n   * per-attempt timeout on every retry. Bounds total time independent of\n   * total attempts × `timeoutMs`.\n   */\n  deadlineMs?: number\n  /** Total provider attempts. Default 3. */\n  maximumAttempts?: number\n  /** Token rates used when the provider omits cost or package pricing does not cover the model. */\n  customTokenPricing?: CustomTokenPricing\n  /**\n   * Transport for requests that declare `jsonSchema`. `native` sends\n   * `response_format: json_schema`; `json-object` sends the broadly supported\n   * JSON mode and relies on the caller to include the schema in model-visible\n   * instructions. Default: `native`.\n   */\n  jsonSchemaTransport?: 'native' | 'json-object'\n  /**\n   * JSON payload parsing policy. `extract` accepts fenced or prose-prefixed JSON.\n   * `exact` requires the complete response content to be one JSON value.\n   * Default: `extract`.\n   */\n  jsonPayloadMode?: 'extract' | 'exact'\n  /** Default provider reasoning mode. A per-call request value takes precedence. */\n  thinking?: LlmThinkingMode\n  /** Fetch implementation — defaults to global `fetch`. Override for custom transport (e.g. tests). */\n  fetch?: typeof fetch\n  /**\n   * Optional raw HTTP capture sink. When provided, every request, response,\n   * and error (across all retry attempts) is recorded to the sink, with auth\n   * headers and credential-shaped body fields redacted by default. This is\n   * the layer-1 forensics primitive: structured `LlmSpan`s record intent,\n   * raw events record what actually crossed the wire.\n   */\n  rawSink?: RawProviderSink\n  /**\n   * Logical provider id attached to raw events. When omitted, derived from\n   * `baseUrl` via `providerFromBaseUrl`.\n   */\n  provider?: string\n  /** Trace context attached to raw events; populated by emitter-aware callers. */\n  traceContext?: { runId?: string; spanId?: string }\n  /** Override the redaction strategy for this call. Defaults to `defaultProviderRedactor`. */\n  redactor?: ProviderRedactor\n}\n\n// ─── Internals ──────────────────────────────────────────────────────────\n\nconst DEFAULT_BASE_URL = 'https://router.tangle.tools/v1'\n// Flagship / reasoning models routinely take several minutes on large prompts (a\n// reflection over many failures, a long tool transcript). A tight cap aborts a\n// legitimately-slow but healthy call — and because every retry attempt re-uses\n// the same window, such a model aborts on ALL attempts and the loop throws. The\n// default is generous enough to let those complete, bounded enough that a truly\n// hung call still fails over after retries, and tunable per deployment via\n// TANGLE_LLM_TIMEOUT_MS. Per-call `req.timeoutMs` / `opts.defaultTimeoutMs`\n// still win for callers that know their model's latency.\nconst DEFAULT_TIMEOUT_MS = Number(process.env.TANGLE_LLM_TIMEOUT_MS) || 300_000\nconst DEFAULT_MAXIMUM_ATTEMPTS =\n  process.env.TANGLE_LLM_MAXIMUM_ATTEMPTS === undefined\n    ? 3\n    : Number(process.env.TANGLE_LLM_MAXIMUM_ATTEMPTS)\n\nfunction resolveMaximumAttempts(configured: number | undefined): number {\n  const attempts = configured ?? DEFAULT_MAXIMUM_ATTEMPTS\n  if (!Number.isInteger(attempts) || attempts <= 0) {\n    throw new RangeError('LLM maximum attempts must be a positive integer')\n  }\n  return attempts\n}\n\nfunction providerTokenCount(value: unknown): number | undefined {\n  return typeof value === 'number' && Number.isSafeInteger(value) && value >= 0 ? value : undefined\n}\n\nconst RETRYABLE_STATUS = new Set([429, 502, 503, 504])\n\n/**\n * Transient transport/network error signatures, matched against an error's\n * name, message, and `code`. Covers fetch/undici network failures, aborts\n * and timeouts, and — critically — HTTP/2 transport faults a keep-alive\n * connection raises mid-response: `terminated`, `NGHTTP2_INTERNAL_ERROR`,\n * `UND_ERR_*`, `other side closed`. Those last ones carry no clean HTTP\n * status; unrecognised, they escape the retry loop and surface as an\n * uncaught rejection.\n */\nconst TRANSIENT_ERROR_PATTERNS: readonly RegExp[] = [\n  /AbortError/i,\n  /TimeoutError/i,\n  /this operation was aborted/i,\n  /fetch failed/i,\n  /ECONNRESET/i,\n  /ETIMEDOUT/i,\n  /EAI_AGAIN/i,\n  /socket hang up/i,\n  /stream.*ended.*unexpectedly/i,\n  /terminated/i,\n  /other side closed/i,\n  /NGHTTP2/i,\n  /UND_ERR/i,\n]\n\n/**\n * True when an error is a transient transport/network fault worth retrying,\n * as opposed to a deterministic failure (4xx schema reject, JSON parse) that\n * a retry cannot fix. Inspects `LlmCallError.status`, then the error's\n * name/message/code, then recurses into `error.cause` — undici nests the\n * real socket fault one or more levels under `.cause`.\n *\n * This is the retry classifier for the package: `callLlm` and\n * `withJudgeRetry` both route through it, so connection failures are treated\n * consistently across transports.\n */\nexport function isTransientLlmError(err: unknown): boolean {\n  return classifyTransient(err, 0)\n}\n\nfunction classifyTransient(err: unknown, depth: number): boolean {\n  if (err instanceof LlmCallError) return RETRYABLE_STATUS.has(err.status)\n  if (!(err instanceof Error)) return false\n  // Foreign transport errors can carry a numeric HTTP status without being an\n  // LlmCallError. A retryable status is decisive.\n  const status = (err as { status?: unknown }).status\n  if (typeof status === 'number' && RETRYABLE_STATUS.has(status)) return true\n  const code = (err as { code?: unknown }).code\n  const haystack = `${err.name}\\n${err.message}\\n${typeof code === 'string' ? code : ''}`\n  if (TRANSIENT_ERROR_PATTERNS.some((p) => p.test(haystack))) return true\n  const cause = (err as { cause?: unknown }).cause\n  if (depth < 4 && cause instanceof Error && cause !== err) {\n    return classifyTransient(cause, depth + 1)\n  }\n  return false\n}\n\nfunction parseRetryAfter(headers: Headers): number | null {\n  const h = headers.get('retry-after')\n  if (!h) return null\n  const asNumber = Number(h)\n  if (Number.isFinite(asNumber) && asNumber > 0) return asNumber * 1000\n  const asDate = Date.parse(h)\n  if (Number.isFinite(asDate)) return Math.max(0, asDate - Date.now())\n  return null\n}\n\n/** Exponential backoff: 500ms, 1s, 2s, 4s, ... capped at 16s. Attempt is 0-indexed. */\nexport function backoffMs(attempt: number): number {\n  return Math.min(500 * 2 ** attempt, 16_000)\n}\n\nfunction buildHeaders(opts: LlmClientOptions): Record<string, string> {\n  const headers: Record<string, string> = {\n    'Content-Type': 'application/json',\n    Accept: 'application/json',\n  }\n  if (opts.authHeader) {\n    headers[opts.authHeader.name] = opts.authHeader.value\n  } else if (opts.bearer || opts.apiKey) {\n    headers.Authorization = `Bearer ${opts.bearer ?? opts.apiKey}`\n  }\n  if (opts.idempotencyKey) headers['Idempotency-Key'] = opts.idempotencyKey\n  return headers\n}\n\nfunction isSchemaRejection(status: number, body: string): boolean {\n  if (status !== 400) return false\n  const lower = body.toLowerCase()\n  return (\n    lower.includes('response_format') ||\n    lower.includes('json_schema') ||\n    lower.includes('is unavailable') ||\n    lower.includes('not supported')\n  )\n}\n\nfunction isTemperatureOneRejection(status: number, body: string): boolean {\n  if (status !== 400 || !/temperature/i.test(body)) return false\n  return (\n    /temperature[^.\\n]{0,120}\\b(?:only|must|should|required|requires?)\\b[^.\\n]{0,40}\\b1(?:\\.0+)?\\b/i.test(\n      body,\n    ) || /\\bonly\\s+1(?:\\.0+)?\\s+is\\s+allowed\\b[^.\\n]{0,120}\\btemperature\\b/i.test(body)\n  )\n}\n\nfunction buildBody(\n  req: LlmCallRequest,\n  forceJsonObject: boolean,\n  defaultThinking?: LlmThinkingMode,\n): Record<string, unknown> {\n  const body: Record<string, unknown> = {\n    model: req.model,\n    messages: req.messages,\n    temperature: req.temperature ?? 0,\n  }\n  if (req.maxTokens != null) {\n    if (usesMaxCompletionTokens(req.model)) body.max_completion_tokens = req.maxTokens\n    else body.max_tokens = req.maxTokens\n  }\n  const thinking = req.thinking ?? defaultThinking\n  if (thinking !== undefined) {\n    body.thinking = { type: thinking }\n  }\n\n  if (req.jsonSchema && !forceJsonObject) {\n    body.response_format = {\n      type: 'json_schema',\n      json_schema: { name: req.jsonSchema.name, schema: req.jsonSchema.schema, strict: true },\n    }\n  } else if (req.jsonMode || req.jsonSchema) {\n    body.response_format = { type: 'json_object' }\n  }\n\n  return body\n}\n\nfunction usesMaxCompletionTokens(model: string): boolean {\n  return /^gpt-5(?:[.-]|$)/i.test(model)\n}\n\nasync function sleep(ms: number): Promise<void> {\n  return new Promise((resolve) => setTimeout(resolve, ms))\n}\n\n/**\n * Combine the per-attempt timeout signal with an optional caller signal into\n * one signal the fetch listens on. Prefers the native `AbortSignal.any`; falls\n * back to manual wiring on runtimes that predate it. The caller signal is also\n * propagated to the timeout controller so aborting it cancels the in-flight\n * fetch immediately.\n */\nfunction linkSignals(timeoutController: AbortController, caller?: AbortSignal): AbortSignal {\n  if (!caller) return timeoutController.signal\n  if (typeof (AbortSignal as { any?: unknown }).any === 'function') {\n    return AbortSignal.any([timeoutController.signal, caller])\n  }\n  if (caller.aborted) {\n    timeoutController.abort()\n  } else {\n    caller.addEventListener('abort', () => timeoutController.abort(), { once: true })\n  }\n  return timeoutController.signal\n}\n\n/** True once the cross-attempt wall-clock budget (if any) is exhausted. */\nfunction deadlineExceeded(start: number, deadlineMs: number | undefined): boolean {\n  return deadlineMs != null && Date.now() - start >= deadlineMs\n}\n\n// ─── Public API ─────────────────────────────────────────────────────────\n\n/**\n * Strip a ```json / ``` code fence if the model emitted one.\n * Idempotent for naked JSON. Some models (claude-code via router, certain\n * deepseek models) wrap output even under json_object.\n */\nexport function stripFencedJson(raw: string): string {\n  const trimmed = raw.trim()\n  const m = trimmed.match(/^```(?:json)?\\s*\\n?([\\s\\S]*?)\\n?```\\s*$/)\n  return m ? m[1]!.trim() : trimmed\n}\n\nexport function extractJsonPayload(raw: string): string {\n  const stripped = stripFencedJson(raw)\n  try {\n    JSON.parse(stripped)\n    return stripped\n  } catch {\n    // A response that declares a JSON root must parse as that complete root.\n    // Scanning onward could turn a truncated object into one of its valid nested\n    // arrays or objects and silently change the response schema.\n    if (stripped.startsWith('{') || stripped.startsWith('[')) return stripped\n  }\n\n  // Only prose-leading responses may contain a recoverable JSON payload.\n  const starts = [...stripped.matchAll(/[[{]/g)]\n    .map((match) => match.index)\n    .filter((index) => index != null)\n  for (const start of starts) {\n    const candidate = extractBalancedJson(stripped, start)\n    if (!candidate) continue\n    try {\n      JSON.parse(candidate)\n      return candidate\n    } catch {\n      // Keep scanning; earlier braces may belong to prose.\n    }\n  }\n\n  return stripped\n}\n\nfunction extractBalancedJson(input: string, start: number): string | null {\n  const opener = input[start]\n  const closer = opener === '{' ? '}' : opener === '[' ? ']' : null\n  if (!closer) return null\n\n  const stack: string[] = [closer]\n  let isInString = false\n  let isEscaped = false\n\n  for (let i = start + 1; i < input.length; i++) {\n    const char = input[i]!\n    if (isEscaped) {\n      isEscaped = false\n      continue\n    }\n    if (char === '\\\\') {\n      isEscaped = isInString\n      continue\n    }\n    if (char === '\"') {\n      isInString = !isInString\n      continue\n    }\n    if (isInString) continue\n\n    if (char === '{') stack.push('}')\n    else if (char === '[') stack.push(']')\n    else if (char === stack[stack.length - 1]) {\n      stack.pop()\n      if (stack.length === 0) return input.slice(start, i + 1)\n    }\n  }\n\n  return null\n}\n\n/**\n * Low-level call. Returns raw content + usage + cost. Retries on transient\n * failures; does NOT degrade schema here — callers that want graceful\n * degrade use `callLlmJson`.\n */\nexport async function callLlm(\n  req: LlmCallRequest,\n  opts: LlmClientOptions = {},\n): Promise<LlmCallResult> {\n  const baseUrl = (opts.baseUrl ?? DEFAULT_BASE_URL).replace(/\\/+$/, '')\n  const url = `${baseUrl}/chat/completions`\n  const endpoint = '/chat/completions'\n  const timeoutMs = req.timeoutMs ?? opts.defaultTimeoutMs ?? DEFAULT_TIMEOUT_MS\n  const maximumAttempts = resolveMaximumAttempts(opts.maximumAttempts)\n  const fetchFn = opts.fetch ?? globalThis.fetch\n  const headers = buildHeaders(opts)\n  const provider = opts.provider ?? providerFromBaseUrl(baseUrl)\n  const sink = opts.rawSink\n  const redactor = opts.redactor ?? defaultProviderRedactor\n  const traceContext = opts.traceContext\n  const callerSignal = opts.signal\n  const deadlineMs = opts.deadlineMs\n  const deadlineStart = Date.now()\n  if (opts.customTokenPricing) {\n    costForTokenPricing(opts.customTokenPricing, { inputTokens: 0, outputTokens: 0 })\n  }\n\n  let lastErr: unknown\n  let effectiveRequest = req\n  for (let attempt = 0; attempt < maximumAttempts; attempt++) {\n    // A caller cancel is fatal — never retried. Checking before each attempt\n    // means an already-aborted signal short-circuits without firing fetch.\n    if (callerSignal?.aborted) {\n      throw new DOMException('callLlm aborted by caller signal', 'AbortError')\n    }\n    // Stop retrying once the cross-attempt budget is spent rather than burning\n    // a full per-attempt timeout on each remaining retry.\n    if (attempt > 0 && deadlineExceeded(deadlineStart, deadlineMs)) {\n      throw lastErr instanceof Error ? lastErr : new Error(String(lastErr))\n    }\n    const controller = new AbortController()\n    const attemptSignal = linkSignals(controller, callerSignal)\n    const timeoutHandle = setTimeout(() => controller.abort(), timeoutMs)\n    const started = Date.now()\n    const requestBody = buildBody(\n      effectiveRequest,\n      opts.jsonSchemaTransport === 'json-object',\n      opts.thinking,\n    )\n    let attemptErrorRecorded = false\n    if (sink) {\n      await recordRaw(sink, redactor, {\n        eventId: cryptoEventId(),\n        runId: traceContext?.runId,\n        spanId: traceContext?.spanId,\n        provider,\n        model: req.model,\n        endpoint,\n        baseUrl,\n        attemptIndex: attempt,\n        direction: 'request',\n        timestamp: started,\n        requestHeaders: headers,\n        requestBody,\n        redactedFields: [],\n      })\n    }\n\n    try {\n      const res = await fetchFn(url, {\n        method: 'POST',\n        headers,\n        body: JSON.stringify(requestBody),\n        signal: attemptSignal,\n      })\n      clearTimeout(timeoutHandle)\n      const responseHeaders = sink ? headersToObject(res.headers) : undefined\n\n      if (!res.ok) {\n        const body = await res.text()\n        if (sink) {\n          await recordRaw(sink, redactor, {\n            eventId: cryptoEventId(),\n            runId: traceContext?.runId,\n            spanId: traceContext?.spanId,\n            provider,\n            model: req.model,\n            endpoint,\n            baseUrl,\n            attemptIndex: attempt,\n            direction: 'error',\n            timestamp: Date.now(),\n            durationMs: Date.now() - started,\n            statusCode: res.status,\n            responseHeaders,\n            responseBody: body,\n            errorMessage: `HTTP ${res.status}`,\n            redactedFields: [],\n          })\n          attemptErrorRecorded = true\n        }\n        const err = new LlmCallError(\n          `LLM call ${res.status}: ${body.slice(0, 300)}`,\n          res.status,\n          body,\n          req.model,\n        )\n        if (\n          isTemperatureOneRejection(res.status, body) &&\n          effectiveRequest.temperature !== 1 &&\n          attempt < maximumAttempts - 1 &&\n          !deadlineExceeded(deadlineStart, deadlineMs)\n        ) {\n          lastErr = err\n          effectiveRequest = { ...effectiveRequest, temperature: 1 }\n          continue\n        }\n        if (\n          RETRYABLE_STATUS.has(res.status) &&\n          attempt < maximumAttempts - 1 &&\n          !deadlineExceeded(deadlineStart, deadlineMs)\n        ) {\n          lastErr = err\n          const retryAfter = parseRetryAfter(res.headers)\n          await sleep(retryAfter ?? backoffMs(attempt))\n          continue\n        }\n        throw err\n      }\n\n      const text = await res.text()\n      let json: Record<string, unknown>\n      try {\n        json = JSON.parse(text) as Record<string, unknown>\n      } catch (parseErr) {\n        if (sink) {\n          await recordRaw(sink, redactor, {\n            eventId: cryptoEventId(),\n            runId: traceContext?.runId,\n            spanId: traceContext?.spanId,\n            provider,\n            model: req.model,\n            endpoint,\n            baseUrl,\n            attemptIndex: attempt,\n            direction: 'error',\n            timestamp: Date.now(),\n            durationMs: Date.now() - started,\n            statusCode: res.status,\n            responseHeaders,\n            responseBody: text,\n            errorMessage: `non-JSON response: ${parseErr instanceof Error ? parseErr.message : String(parseErr)}`,\n            redactedFields: [],\n          })\n          attemptErrorRecorded = true\n        }\n        throw parseErr\n      }\n      if (sink) {\n        await recordRaw(sink, redactor, {\n          eventId: cryptoEventId(),\n          runId: traceContext?.runId,\n          spanId: traceContext?.spanId,\n          provider,\n          model: req.model,\n          endpoint,\n          baseUrl,\n          attemptIndex: attempt,\n          direction: 'response',\n          timestamp: Date.now(),\n          durationMs: Date.now() - started,\n          statusCode: res.status,\n          responseHeaders,\n          responseBody: json,\n          redactedFields: [],\n        })\n      }\n      const choice = (\n        json.choices as\n          | Array<{ message?: { content?: string }; finish_reason?: string | null }>\n          | undefined\n      )?.[0]\n      const usageRaw =\n        json.usage && typeof json.usage === 'object' && !Array.isArray(json.usage)\n          ? (json.usage as Record<string, unknown>)\n          : undefined\n      const promptTokens = providerTokenCount(usageRaw?.prompt_tokens)\n      const completionTokens = providerTokenCount(usageRaw?.completion_tokens)\n      const totalTokens = providerTokenCount(usageRaw?.total_tokens)\n      const completionDetails =\n        usageRaw?.completion_tokens_details &&\n        typeof usageRaw.completion_tokens_details === 'object' &&\n        !Array.isArray(usageRaw.completion_tokens_details)\n          ? (usageRaw.completion_tokens_details as Record<string, unknown>)\n          : undefined\n      const reasoningRaw = completionDetails?.reasoning_tokens\n      const reasoningTokens =\n        reasoningRaw === undefined ? undefined : providerTokenCount(reasoningRaw)\n      const cachedRaw =\n        usageRaw?.prompt_tokens_details &&\n        typeof usageRaw.prompt_tokens_details === 'object' &&\n        !Array.isArray(usageRaw.prompt_tokens_details)\n          ? (usageRaw.prompt_tokens_details as Record<string, unknown>).cached_tokens\n          : undefined\n      const cachedPromptTokens = cachedRaw === undefined ? undefined : providerTokenCount(cachedRaw)\n      const usageCaptured =\n        promptTokens !== undefined &&\n        completionTokens !== undefined &&\n        (reasoningRaw === undefined ||\n          (reasoningTokens !== undefined && reasoningTokens <= completionTokens)) &&\n        (cachedRaw === undefined ||\n          (cachedPromptTokens !== undefined && cachedPromptTokens <= promptTokens)) &&\n        (totalTokens === undefined || totalTokens === promptTokens + completionTokens)\n      const costFromProxy = (json._response_cost ?? json.cost_usd) as number | undefined\n      const content = choice?.message?.content ?? ''\n\n      const configuredCost =\n        typeof costFromProxy !== 'number' && usageCaptured && opts.customTokenPricing\n          ? costForTokenPricing(opts.customTokenPricing, {\n              inputTokens: promptTokens! - (cachedPromptTokens ?? 0),\n              ...(cachedPromptTokens ? { cachedTokens: cachedPromptTokens } : {}),\n              outputTokens: completionTokens!,\n            })\n          : undefined\n\n      return {\n        content,\n        finishReason: choice?.finish_reason ?? null,\n        contentEmpty: content.trim().length === 0,\n        usage: {\n          promptTokens: promptTokens ?? 0,\n          completionTokens: completionTokens ?? 0,\n          totalTokens: totalTokens ?? (promptTokens ?? 0) + (completionTokens ?? 0),\n          captured: usageCaptured,\n          reasoningTokens,\n          cachedPromptTokens,\n        },\n        costUsd: typeof costFromProxy === 'number' ? costFromProxy : (configuredCost ?? null),\n        model: (json.model as string) ?? req.model,\n        durationMs: Date.now() - started,\n        raw: json,\n      }\n    } catch (err) {\n      clearTimeout(timeoutHandle)\n      lastErr = err\n      // A caller cancel is fatal even though an AbortError matches the\n      // transient patterns — a cancelled call must surface immediately, not\n      // be retried against the same dead intent.\n      if (callerSignal?.aborted) {\n        if (sink && !attemptErrorRecorded) {\n          await recordRaw(sink, redactor, {\n            eventId: cryptoEventId(),\n            runId: traceContext?.runId,\n            spanId: traceContext?.spanId,\n            provider,\n            model: req.model,\n            endpoint,\n            baseUrl,\n            attemptIndex: attempt,\n            direction: 'error',\n            timestamp: Date.now(),\n            durationMs: Date.now() - started,\n            errorMessage: err instanceof Error ? err.message : String(err),\n            redactedFields: [],\n          })\n        }\n        throw err\n      }\n      if (sink && !attemptErrorRecorded) {\n        // Record only if neither the !res.ok branch nor the JSON.parse catch\n        // already produced an error event for this attempt. Covers network\n        // failures, timeouts, and aborts.\n        await recordRaw(sink, redactor, {\n          eventId: cryptoEventId(),\n          runId: traceContext?.runId,\n          spanId: traceContext?.spanId,\n          provider,\n          model: req.model,\n          endpoint,\n          baseUrl,\n          attemptIndex: attempt,\n          direction: 'error',\n          timestamp: Date.now(),\n          durationMs: Date.now() - started,\n          errorMessage: err instanceof Error ? err.message : String(err),\n          redactedFields: [],\n        })\n      }\n      if (\n        attempt < maximumAttempts - 1 &&\n        isTransientLlmError(err) &&\n        !deadlineExceeded(deadlineStart, deadlineMs)\n      ) {\n        await sleep(backoffMs(attempt))\n        continue\n      }\n      throw err\n    }\n  }\n  throw lastErr instanceof Error ? lastErr : new Error(String(lastErr))\n}\n\nasync function recordRaw(\n  sink: RawProviderSink,\n  redactor: ProviderRedactor,\n  event: RawProviderEvent,\n): Promise<void> {\n  // Errors from sinks must not crash the LLM call. Forensic capture is\n  // best-effort; the structured trace is the system of record.\n  try {\n    await sink.record(redactor(event))\n  } catch {\n    // Intentionally swallowed.\n  }\n}\n\nfunction headersToObject(h: Headers): Record<string, string> {\n  const out: Record<string, string> = {}\n  h.forEach((value, key) => {\n    out[key] = value\n  })\n  return out\n}\n\nfunction cryptoEventId(): string {\n  if (typeof globalThis.crypto?.randomUUID === 'function') return globalThis.crypto.randomUUID()\n  return `${Date.now().toString(36)}-${Math.random().toString(36).slice(2, 10)}`\n}\n\n/**\n * Structured-output call. Returns parsed JSON plus the raw result envelope.\n * Degrades `jsonSchema` → `jsonMode` on a 400 that names the schema param —\n * critical for deepseek-v3/v4, kimi-k2.6, and other models that don't accept\n * the `response_format.json_schema` shape but DO accept `json_object`.\n */\nexport async function callLlmJson<T = unknown>(\n  req: LlmCallRequest,\n  opts: LlmClientOptions = {},\n): Promise<{ value: T; result: LlmCallResult }> {\n  const result = await callLlmStructured(req, opts)\n  const value = parseJsonResult<T>(result, opts.jsonPayloadMode ?? 'extract')\n  return { value, result }\n}\n\n/** Shared schema-to-JSON-mode fallback that preserves the raw result. */\nasync function callLlmStructured(\n  req: LlmCallRequest,\n  opts: LlmClientOptions = {},\n): Promise<LlmCallResult> {\n  try {\n    return await callLlm({ ...req, jsonMode: req.jsonMode ?? !req.jsonSchema }, opts)\n  } catch (err) {\n    if (\n      opts.jsonSchemaTransport !== 'json-object' &&\n      err instanceof LlmCallError &&\n      isSchemaRejection(err.status, err.body) &&\n      req.jsonSchema\n    ) {\n      const degradedReq: LlmCallRequest = { ...req, jsonMode: true, jsonSchema: undefined }\n      return await callLlm(degradedReq, opts)\n    }\n    throw err\n  }\n}\n\nfunction parseJsonResult<T>(\n  result: LlmCallResult,\n  jsonPayloadMode: NonNullable<LlmClientOptions['jsonPayloadMode']>,\n): T {\n  try {\n    if (result.finishReason === 'length') {\n      throw new Error(\n        `LLM returned truncated JSON content (model=${result.model}, finishReason=length)`,\n      )\n    }\n    return parseJsonSafely<T>(result.content, result.model, jsonPayloadMode)\n  } catch (error) {\n    if (error instanceof LlmResponseError) throw error\n    const cause = error instanceof Error ? error : new Error(String(error))\n    throw new LlmResponseError(cause.message, result, { cause })\n  }\n}\n\nfunction parseJsonSafely<T>(\n  content: string,\n  model: string,\n  jsonPayloadMode: NonNullable<LlmClientOptions['jsonPayloadMode']>,\n): T {\n  const payload = jsonPayloadMode === 'exact' ? content : extractJsonPayload(content)\n  try {\n    return JSON.parse(payload) as T\n  } catch (err) {\n    throw new Error(\n      `LLM returned non-JSON content (model=${model}): ${\n        err instanceof Error ? err.message : String(err)\n      }\\n--- raw content ---\\n${content.slice(0, 800)}`,\n    )\n  }\n}\n\n// ─── Route assertion ────────────────────────────────────────────────────\n\nexport type LlmRouteAssertionReason =\n  | 'no_explicit_base_url'\n  | 'base_url_blocked'\n  | 'base_url_not_allowed'\n  | 'no_auth'\n  | 'wrong_provider'\n\nexport class LlmRouteAssertionError extends CaptureIntegrityError {\n  constructor(\n    message: string,\n    public readonly reason: LlmRouteAssertionReason,\n    public readonly baseUrl: string,\n  ) {\n    super(message)\n  }\n}\n\nexport interface LlmRouteRequirements {\n  /**\n   * Throw if `opts.baseUrl` is undefined, i.e. the call would fall back to\n   * `DEFAULT_BASE_URL`. Set this for evaluation runs where silently using\n   * the public/free-tier router is a defect — the launch reviewer needs to\n   * know exactly which provider answered.\n   */\n  requireExplicitBaseUrl?: boolean\n  /**\n   * Allowlist of acceptable base URLs. Strings match by prefix\n   * (case-insensitive); RegExps test against the full base URL.\n   */\n  allowedBaseUrls?: Array<string | RegExp>\n  /** Blocklist that takes precedence over `allowedBaseUrls`. */\n  blockedBaseUrls?: Array<string | RegExp>\n  /** Throw if no auth header / api key is configured. */\n  requireAuth?: boolean\n  /**\n   * Logical provider id the configured `baseUrl` is expected to match (via\n   * `providerFromBaseUrl`). Mainly useful when paired with `requireExplicitBaseUrl`.\n   */\n  expectedProvider?: string\n}\n\n/**\n * Fail-loud assertion that the configured LLM client points at the route\n * the caller intends. Designed for the matrix-runner preflight: invoke\n * once before any LLM call to catch misconfiguration before a sweep burns\n * dollars on the wrong provider.\n *\n * Throws `LlmRouteAssertionError`. Pure — no I/O — so it's safe to call\n * from constructors and CI gates.\n */\nexport function assertLlmRoute(opts: LlmClientOptions, req: LlmRouteRequirements = {}): void {\n  const baseUrlExplicit = opts.baseUrl !== undefined\n  const baseUrl = (opts.baseUrl ?? DEFAULT_BASE_URL).replace(/\\/+$/, '')\n\n  if (req.requireExplicitBaseUrl && !baseUrlExplicit) {\n    throw new LlmRouteAssertionError(\n      `assertLlmRoute: requireExplicitBaseUrl set but opts.baseUrl is undefined; would fall back to ${DEFAULT_BASE_URL}.`,\n      'no_explicit_base_url',\n      baseUrl,\n    )\n  }\n\n  if (req.blockedBaseUrls?.some((p) => matchUrl(baseUrl, p))) {\n    throw new LlmRouteAssertionError(\n      `assertLlmRoute: baseUrl ${baseUrl} matches a blocked pattern.`,\n      'base_url_blocked',\n      baseUrl,\n    )\n  }\n\n  if (req.allowedBaseUrls && req.allowedBaseUrls.length > 0) {\n    const ok = req.allowedBaseUrls.some((p) => matchUrl(baseUrl, p))\n    if (!ok) {\n      throw new LlmRouteAssertionError(\n        `assertLlmRoute: baseUrl ${baseUrl} is not in the allowed list (${req.allowedBaseUrls.map(describePattern).join(', ')}).`,\n        'base_url_not_allowed',\n        baseUrl,\n      )\n    }\n  }\n\n  if (req.requireAuth && !opts.apiKey && !opts.bearer && !opts.authHeader) {\n    throw new LlmRouteAssertionError(\n      `assertLlmRoute: requireAuth set but no apiKey, bearer, or authHeader was supplied.`,\n      'no_auth',\n      baseUrl,\n    )\n  }\n\n  if (req.expectedProvider) {\n    const actual = opts.provider ?? providerFromBaseUrl(baseUrl)\n    if (actual !== req.expectedProvider) {\n      throw new LlmRouteAssertionError(\n        `assertLlmRoute: expected provider ${req.expectedProvider} but baseUrl ${baseUrl} resolves to ${actual}.`,\n        'wrong_provider',\n        baseUrl,\n      )\n    }\n  }\n}\n\nfunction matchUrl(url: string, pattern: string | RegExp): boolean {\n  if (pattern instanceof RegExp) return pattern.test(url)\n  return url.toLowerCase().startsWith(pattern.toLowerCase())\n}\n\nfunction describePattern(p: string | RegExp): string {\n  return p instanceof RegExp ? p.source : p\n}\n\n/**\n * Probe whether a model is reachable. Returns latency + null error on\n * success; `ok=false` + error message on any failure (HTTP, timeout,\n * network, parse). Designed for sweep preflights — fail loud at the\n * boundary before burning a 30-leaf run on a misconfigured router.\n *\n * Sends a tiny `ping` message with `maxTokens=64`. Reasoning models\n * (glm-5.1, deepseek-v4) can burn the entire budget on internal reasoning\n * for short prompts, so don't tighten this further. We don't validate\n * content; HTTP 200 means reachable.\n */\nexport async function probeLlm(\n  model: string,\n  opts: LlmClientOptions & { timeoutMs?: number } = {},\n): Promise<{ ok: boolean; latencyMs: number; error: string | null }> {\n  const start = Date.now()\n  try {\n    await callLlm(\n      {\n        model,\n        messages: [{ role: 'user', content: 'ping' }],\n        maxTokens: 64,\n        timeoutMs: opts.timeoutMs ?? 30_000,\n      },\n      opts,\n    )\n    return { ok: true, latencyMs: Date.now() - start, error: null }\n  } catch (err) {\n    return {\n      ok: false,\n      latencyMs: Date.now() - start,\n      error: err instanceof Error ? err.message : String(err),\n    }\n  }\n}\n\n/**\n * Stateful client — construct once with defaults, call many times.\n * Thin wrapper around the free functions; exists for callers that want\n * to inject a single configured instance into multiple primitives.\n */\nexport class LlmClient {\n  readonly maximumAttempts: number\n  private readonly opts: LlmClientOptions\n\n  constructor(opts: LlmClientOptions = {}) {\n    this.opts = opts\n    this.maximumAttempts = resolveMaximumAttempts(opts.maximumAttempts)\n  }\n\n  call(req: LlmCallRequest, per?: LlmClientOptions): Promise<LlmCallResult> {\n    const options = { ...this.opts, ...per }\n    return req.jsonSchema ? callLlmStructured(req, options) : callLlm(req, options)\n  }\n\n  callJson<T = unknown>(\n    req: LlmCallRequest,\n    per?: LlmClientOptions,\n  ): Promise<{ value: T; result: LlmCallResult }> {\n    return callLlmJson<T>(req, { ...this.opts, ...per })\n  }\n}\n"],"mappings":";;;;;;;;;;;;;;;;;;;;;;;;;;;;;AA0EA,SAAgB,2BACd,SACA,UAA4B,CAAC,GACF;CAC3B,IAAI,QAAQ,cAAc,KAAA,GAAW,OAAO,KAAA;CAC5C,IAAI,CAAC,OAAO,UAAU,QAAQ,SAAS,KAAK,QAAQ,aAAa,GAC/D,MAAM,IAAI,WAAW,kEAAkE;CAEzF,IACE,QAAQ,SAAS,MACd,YACC,MAAM,QAAQ,QAAQ,OAAO,KAAK,QAAQ,QAAQ,MAAM,SAAS,KAAK,SAAS,WAAW,CAC9F,GAEA;CAGF,MAAM,WAAW,uBAAuB,QAAQ,eAAe;CAC/D,MAAM,kBAAkB,QAAQ,wBAAwB;CAGxD,MAAM,eAAe,IAAI,YAAY,CAAC,CAAC,OACrC,KAAK,UAAU,UAAU,SAAS,iBAAiB,QAAQ,QAAQ,CAAC,CACtE,CAAC,CAAC;CAEF,MAAM,UAAU,QAAQ,cAAc,CAAC,kBAAkB,IAAI;CAC7D,MAAM,QAAQ;EACZ,aAAa,eAAe,WAAW;EACvC,cAAc,QAAQ,YAAY,WAAW;CAC/C;CACA,OAAO,QAAQ,qBACX;EAAE,oBAAoB,QAAQ;EAAoB,GAAG;CAAM,IAC3D;EAAE,OAAO,QAAQ;EAAO,GAAG;CAAM;AACvC;;AAkDA,SAAgB,mBACd,QACA,oBACkB;CAClB,MAAM,eAAe,OAAO,MAAM,sBAAsB;CACxD,MAAM,cAAc,KAAK,IAAI,GAAG,OAAO,MAAM,eAAe,YAAY;CACxE,MAAM,oBACJ,OAAO,YAAY,QAAQ,sBAAsB,OAAO,MAAM,aAAa,QACvE,oBAAoB,oBAAoB;EACtC;EACA,GAAI,eAAe,IAAI,EAAE,aAAa,IAAI,CAAC;EAC3C,cAAc,OAAO,MAAM;CAC7B,CAAC,IACD,KAAA;CACN,OAAO;EACL,OAAO,OAAO;EACd;EACA,cAAc,OAAO,MAAM;EAC3B,iBAAiB,OAAO,MAAM;EAC9B,cAAc,eAAe,IAAI,eAAe,KAAA;EAChD,eAAe,OAAO,WAAW;EACjC,cAAc,OAAO,MAAM,aAAa;CAC1C;AACF;;AAGA,SAAgB,wBACd,OACA,oBAC8B;CAC9B,OAAO,iBAAiB,mBACpB,mBAAmB,MAAM,QAAQ,kBAAkB,IACnD,KAAA;AACN;AAEA,IAAa,eAAb,cAAkC,eAAe;CAG7B;CACA;CACA;CAJlB,YACE,SACA,QACA,MACA,OACA;EACA,MAAM,SAAS,OAAO;EAJN,KAAA,SAAA;EACA,KAAA,OAAA;EACA,KAAA,QAAA;CAGlB;AACF;;;;AAKA,IAAa,mBAAb,cAAsC,eAAe;CAGjC;CAFlB,YACE,SACA,QACA,SACA;EACA,MAAM,SAAS,SAAS,OAAO;EAHf,KAAA,SAAA;CAIlB;AACF;AAuEA,MAAM,mBAAmB;AASzB,MAAM,qBAAqB,OAAO,QAAQ,IAAI,qBAAqB,KAAK;AACxE,MAAM,2BACJ,QAAQ,IAAI,gCAAgC,KAAA,IACxC,IACA,OAAO,QAAQ,IAAI,2BAA2B;AAEpD,SAAS,uBAAuB,YAAwC;CACtE,MAAM,WAAW,cAAc;CAC/B,IAAI,CAAC,OAAO,UAAU,QAAQ,KAAK,YAAY,GAC7C,MAAM,IAAI,WAAW,iDAAiD;CAExE,OAAO;AACT;AAEA,SAAS,mBAAmB,OAAoC;CAC9D,OAAO,OAAO,UAAU,YAAY,OAAO,cAAc,KAAK,KAAK,SAAS,IAAI,QAAQ,KAAA;AAC1F;AAEA,MAAM,mCAAmB,IAAI,IAAI;CAAC;CAAK;CAAK;CAAK;AAAG,CAAC;;;;;;;;;;AAWrD,MAAM,2BAA8C;CAClD;CACA;CACA;CACA;CACA;CACA;CACA;CACA;CACA;CACA;CACA;CACA;CACA;AACF;;;;;;;;;;;;AAaA,SAAgB,oBAAoB,KAAuB;CACzD,OAAO,kBAAkB,KAAK,CAAC;AACjC;AAEA,SAAS,kBAAkB,KAAc,OAAwB;CAC/D,IAAI,eAAe,cAAc,OAAO,iBAAiB,IAAI,IAAI,MAAM;CACvE,IAAI,EAAE,eAAe,QAAQ,OAAO;CAGpC,MAAM,SAAU,IAA6B;CAC7C,IAAI,OAAO,WAAW,YAAY,iBAAiB,IAAI,MAAM,GAAG,OAAO;CACvE,MAAM,OAAQ,IAA2B;CACzC,MAAM,WAAW,GAAG,IAAI,KAAK,IAAI,IAAI,QAAQ,IAAI,OAAO,SAAS,WAAW,OAAO;CACnF,IAAI,yBAAyB,MAAM,MAAM,EAAE,KAAK,QAAQ,CAAC,GAAG,OAAO;CACnE,MAAM,QAAS,IAA4B;CAC3C,IAAI,QAAQ,KAAK,iBAAiB,SAAS,UAAU,KACnD,OAAO,kBAAkB,OAAO,QAAQ,CAAC;CAE3C,OAAO;AACT;AAEA,SAAS,gBAAgB,SAAiC;CACxD,MAAM,IAAI,QAAQ,IAAI,aAAa;CACnC,IAAI,CAAC,GAAG,OAAO;CACf,MAAM,WAAW,OAAO,CAAC;CACzB,IAAI,OAAO,SAAS,QAAQ,KAAK,WAAW,GAAG,OAAO,WAAW;CACjE,MAAM,SAAS,KAAK,MAAM,CAAC;CAC3B,IAAI,OAAO,SAAS,MAAM,GAAG,OAAO,KAAK,IAAI,GAAG,SAAS,KAAK,IAAI,CAAC;CACnE,OAAO;AACT;;AAGA,SAAgB,UAAU,SAAyB;CACjD,OAAO,KAAK,IAAI,MAAM,KAAK,SAAS,IAAM;AAC5C;AAEA,SAAS,aAAa,MAAgD;CACpE,MAAM,UAAkC;EACtC,gBAAgB;EAChB,QAAQ;CACV;CACA,IAAI,KAAK,YACP,QAAQ,KAAK,WAAW,QAAQ,KAAK,WAAW;MAC3C,IAAI,KAAK,UAAU,KAAK,QAC7B,QAAQ,gBAAgB,UAAU,KAAK,UAAU,KAAK;CAExD,IAAI,KAAK,gBAAgB,QAAQ,qBAAqB,KAAK;CAC3D,OAAO;AACT;AAEA,SAAS,kBAAkB,QAAgB,MAAuB;CAChE,IAAI,WAAW,KAAK,OAAO;CAC3B,MAAM,QAAQ,KAAK,YAAY;CAC/B,OACE,MAAM,SAAS,iBAAiB,KAChC,MAAM,SAAS,aAAa,KAC5B,MAAM,SAAS,gBAAgB,KAC/B,MAAM,SAAS,eAAe;AAElC;AAEA,SAAS,0BAA0B,QAAgB,MAAuB;CACxE,IAAI,WAAW,OAAO,CAAC,eAAe,KAAK,IAAI,GAAG,OAAO;CACzD,OACE,iGAAiG,KAC/F,IACF,KAAK,oEAAoE,KAAK,IAAI;AAEtF;AAEA,SAAS,UACP,KACA,iBACA,iBACyB;CACzB,MAAM,OAAgC;EACpC,OAAO,IAAI;EACX,UAAU,IAAI;EACd,aAAa,IAAI,eAAe;CAClC;CACA,IAAI,IAAI,aAAa,MACnB,IAAI,wBAAwB,IAAI,KAAK,GAAG,KAAK,wBAAwB,IAAI;MACpE,KAAK,aAAa,IAAI;CAE7B,MAAM,WAAW,IAAI,YAAY;CACjC,IAAI,aAAa,KAAA,GACf,KAAK,WAAW,EAAE,MAAM,SAAS;CAGnC,IAAI,IAAI,cAAc,CAAC,iBACrB,KAAK,kBAAkB;EACrB,MAAM;EACN,aAAa;GAAE,MAAM,IAAI,WAAW;GAAM,QAAQ,IAAI,WAAW;GAAQ,QAAQ;EAAK;CACxF;MACK,IAAI,IAAI,YAAY,IAAI,YAC7B,KAAK,kBAAkB,EAAE,MAAM,cAAc;CAG/C,OAAO;AACT;AAEA,SAAS,wBAAwB,OAAwB;CACvD,OAAO,oBAAoB,KAAK,KAAK;AACvC;AAEA,eAAe,MAAM,IAA2B;CAC9C,OAAO,IAAI,SAAS,YAAY,WAAW,SAAS,EAAE,CAAC;AACzD;;;;;;;;AASA,SAAS,YAAY,mBAAoC,QAAmC;CAC1F,IAAI,CAAC,QAAQ,OAAO,kBAAkB;CACtC,IAAI,OAAQ,YAAkC,QAAQ,YACpD,OAAO,YAAY,IAAI,CAAC,kBAAkB,QAAQ,MAAM,CAAC;CAE3D,IAAI,OAAO,SACT,kBAAkB,MAAM;MAExB,OAAO,iBAAiB,eAAe,kBAAkB,MAAM,GAAG,EAAE,MAAM,KAAK,CAAC;CAElF,OAAO,kBAAkB;AAC3B;;AAGA,SAAS,iBAAiB,OAAe,YAAyC;CAChF,OAAO,cAAc,QAAQ,KAAK,IAAI,IAAI,SAAS;AACrD;;;;;;AASA,SAAgB,gBAAgB,KAAqB;CACnD,MAAM,UAAU,IAAI,KAAK;CACzB,MAAM,IAAI,QAAQ,MAAM,yCAAyC;CACjE,OAAO,IAAI,EAAE,EAAE,CAAE,KAAK,IAAI;AAC5B;AAEA,SAAgB,mBAAmB,KAAqB;CACtD,MAAM,WAAW,gBAAgB,GAAG;CACpC,IAAI;EACF,KAAK,MAAM,QAAQ;EACnB,OAAO;CACT,QAAQ;EAIN,IAAI,SAAS,WAAW,GAAG,KAAK,SAAS,WAAW,GAAG,GAAG,OAAO;CACnE;CAGA,MAAM,SAAS,CAAC,GAAG,SAAS,SAAS,OAAO,CAAC,CAAC,CAC3C,KAAK,UAAU,MAAM,KAAK,CAAC,CAC3B,QAAQ,UAAU,SAAS,IAAI;CAClC,KAAK,MAAM,SAAS,QAAQ;EAC1B,MAAM,YAAY,oBAAoB,UAAU,KAAK;EACrD,IAAI,CAAC,WAAW;EAChB,IAAI;GACF,KAAK,MAAM,SAAS;GACpB,OAAO;EACT,QAAQ,CAER;CACF;CAEA,OAAO;AACT;AAEA,SAAS,oBAAoB,OAAe,OAA8B;CACxE,MAAM,SAAS,MAAM;CACrB,MAAM,SAAS,WAAW,MAAM,MAAM,WAAW,MAAM,MAAM;CAC7D,IAAI,CAAC,QAAQ,OAAO;CAEpB,MAAM,QAAkB,CAAC,MAAM;CAC/B,IAAI,aAAa;CACjB,IAAI,YAAY;CAEhB,KAAK,IAAI,IAAI,QAAQ,GAAG,IAAI,MAAM,QAAQ,KAAK;EAC7C,MAAM,OAAO,MAAM;EACnB,IAAI,WAAW;GACb,YAAY;GACZ;EACF;EACA,IAAI,SAAS,MAAM;GACjB,YAAY;GACZ;EACF;EACA,IAAI,SAAS,MAAK;GAChB,aAAa,CAAC;GACd;EACF;EACA,IAAI,YAAY;EAEhB,IAAI,SAAS,KAAK,MAAM,KAAK,GAAG;OAC3B,IAAI,SAAS,KAAK,MAAM,KAAK,GAAG;OAChC,IAAI,SAAS,MAAM,MAAM,SAAS,IAAI;GACzC,MAAM,IAAI;GACV,IAAI,MAAM,WAAW,GAAG,OAAO,MAAM,MAAM,OAAO,IAAI,CAAC;EACzD;CACF;CAEA,OAAO;AACT;;;;;;AAOA,eAAsB,QACpB,KACA,OAAyB,CAAC,GACF;CACxB,MAAM,WAAW,KAAK,WAAW,iBAAA,CAAkB,QAAQ,QAAQ,EAAE;CACrE,MAAM,MAAM,GAAG,QAAQ;CACvB,MAAM,WAAW;CACjB,MAAM,YAAY,IAAI,aAAa,KAAK,oBAAoB;CAC5D,MAAM,kBAAkB,uBAAuB,KAAK,eAAe;CACnE,MAAM,UAAU,KAAK,SAAS,WAAW;CACzC,MAAM,UAAU,aAAa,IAAI;CACjC,MAAM,WAAW,KAAK,YAAY,oBAAoB,OAAO;CAC7D,MAAM,OAAO,KAAK;CAClB,MAAM,WAAW,KAAK,YAAY;CAClC,MAAM,eAAe,KAAK;CAC1B,MAAM,eAAe,KAAK;CAC1B,MAAM,aAAa,KAAK;CACxB,MAAM,gBAAgB,KAAK,IAAI;CAC/B,IAAI,KAAK,oBACP,oBAAoB,KAAK,oBAAoB;EAAE,aAAa;EAAG,cAAc;CAAE,CAAC;CAGlF,IAAI;CACJ,IAAI,mBAAmB;CACvB,KAAK,IAAI,UAAU,GAAG,UAAU,iBAAiB,WAAW;EAG1D,IAAI,cAAc,SAChB,MAAM,IAAI,aAAa,oCAAoC,YAAY;EAIzE,IAAI,UAAU,KAAK,iBAAiB,eAAe,UAAU,GAC3D,MAAM,mBAAmB,QAAQ,UAAU,IAAI,MAAM,OAAO,OAAO,CAAC;EAEtE,MAAM,aAAa,IAAI,gBAAgB;EACvC,MAAM,gBAAgB,YAAY,YAAY,YAAY;EAC1D,MAAM,gBAAgB,iBAAiB,WAAW,MAAM,GAAG,SAAS;EACpE,MAAM,UAAU,KAAK,IAAI;EACzB,MAAM,cAAc,UAClB,kBACA,KAAK,wBAAwB,eAC7B,KAAK,QACP;EACA,IAAI,uBAAuB;EAC3B,IAAI,MACF,MAAM,UAAU,MAAM,UAAU;GAC9B,SAAS,cAAc;GACvB,OAAO,cAAc;GACrB,QAAQ,cAAc;GACtB;GACA,OAAO,IAAI;GACX;GACA;GACA,cAAc;GACd,WAAW;GACX,WAAW;GACX,gBAAgB;GAChB;GACA,gBAAgB,CAAC;EACnB,CAAC;EAGH,IAAI;GACF,MAAM,MAAM,MAAM,QAAQ,KAAK;IAC7B,QAAQ;IACR;IACA,MAAM,KAAK,UAAU,WAAW;IAChC,QAAQ;GACV,CAAC;GACD,aAAa,aAAa;GAC1B,MAAM,kBAAkB,OAAO,gBAAgB,IAAI,OAAO,IAAI,KAAA;GAE9D,IAAI,CAAC,IAAI,IAAI;IACX,MAAM,OAAO,MAAM,IAAI,KAAK;IAC5B,IAAI,MAAM;KACR,MAAM,UAAU,MAAM,UAAU;MAC9B,SAAS,cAAc;MACvB,OAAO,cAAc;MACrB,QAAQ,cAAc;MACtB;MACA,OAAO,IAAI;MACX;MACA;MACA,cAAc;MACd,WAAW;MACX,WAAW,KAAK,IAAI;MACpB,YAAY,KAAK,IAAI,IAAI;MACzB,YAAY,IAAI;MAChB;MACA,cAAc;MACd,cAAc,QAAQ,IAAI;MAC1B,gBAAgB,CAAC;KACnB,CAAC;KACD,uBAAuB;IACzB;IACA,MAAM,MAAM,IAAI,aACd,YAAY,IAAI,OAAO,IAAI,KAAK,MAAM,GAAG,GAAG,KAC5C,IAAI,QACJ,MACA,IAAI,KACN;IACA,IACE,0BAA0B,IAAI,QAAQ,IAAI,KAC1C,iBAAiB,gBAAgB,KACjC,UAAU,kBAAkB,KAC5B,CAAC,iBAAiB,eAAe,UAAU,GAC3C;KACA,UAAU;KACV,mBAAmB;MAAE,GAAG;MAAkB,aAAa;KAAE;KACzD;IACF;IACA,IACE,iBAAiB,IAAI,IAAI,MAAM,KAC/B,UAAU,kBAAkB,KAC5B,CAAC,iBAAiB,eAAe,UAAU,GAC3C;KACA,UAAU;KAEV,MAAM,MADa,gBAAgB,IAAI,OAClB,KAAK,UAAU,OAAO,CAAC;KAC5C;IACF;IACA,MAAM;GACR;GAEA,MAAM,OAAO,MAAM,IAAI,KAAK;GAC5B,IAAI;GACJ,IAAI;IACF,OAAO,KAAK,MAAM,IAAI;GACxB,SAAS,UAAU;IACjB,IAAI,MAAM;KACR,MAAM,UAAU,MAAM,UAAU;MAC9B,SAAS,cAAc;MACvB,OAAO,cAAc;MACrB,QAAQ,cAAc;MACtB;MACA,OAAO,IAAI;MACX;MACA;MACA,cAAc;MACd,WAAW;MACX,WAAW,KAAK,IAAI;MACpB,YAAY,KAAK,IAAI,IAAI;MACzB,YAAY,IAAI;MAChB;MACA,cAAc;MACd,cAAc,sBAAsB,oBAAoB,QAAQ,SAAS,UAAU,OAAO,QAAQ;MAClG,gBAAgB,CAAC;KACnB,CAAC;KACD,uBAAuB;IACzB;IACA,MAAM;GACR;GACA,IAAI,MACF,MAAM,UAAU,MAAM,UAAU;IAC9B,SAAS,cAAc;IACvB,OAAO,cAAc;IACrB,QAAQ,cAAc;IACtB;IACA,OAAO,IAAI;IACX;IACA;IACA,cAAc;IACd,WAAW;IACX,WAAW,KAAK,IAAI;IACpB,YAAY,KAAK,IAAI,IAAI;IACzB,YAAY,IAAI;IAChB;IACA,cAAc;IACd,gBAAgB,CAAC;GACnB,CAAC;GAEH,MAAM,SACJ,KAAK,UAGH;GACJ,MAAM,WACJ,KAAK,SAAS,OAAO,KAAK,UAAU,YAAY,CAAC,MAAM,QAAQ,KAAK,KAAK,IACpE,KAAK,QACN,KAAA;GACN,MAAM,eAAe,mBAAmB,UAAU,aAAa;GAC/D,MAAM,mBAAmB,mBAAmB,UAAU,iBAAiB;GACvE,MAAM,cAAc,mBAAmB,UAAU,YAAY;GAO7D,MAAM,gBALJ,UAAU,6BACV,OAAO,SAAS,8BAA8B,YAC9C,CAAC,MAAM,QAAQ,SAAS,yBAAyB,IAC5C,SAAS,4BACV,KAAA,EAAA,EACkC;GACxC,MAAM,kBACJ,iBAAiB,KAAA,IAAY,KAAA,IAAY,mBAAmB,YAAY;GAC1E,MAAM,YACJ,UAAU,yBACV,OAAO,SAAS,0BAA0B,YAC1C,CAAC,MAAM,QAAQ,SAAS,qBAAqB,IACxC,SAAS,sBAAkD,gBAC5D,KAAA;GACN,MAAM,qBAAqB,cAAc,KAAA,IAAY,KAAA,IAAY,mBAAmB,SAAS;GAC7F,MAAM,gBACJ,iBAAiB,KAAA,KACjB,qBAAqB,KAAA,MACpB,iBAAiB,KAAA,KACf,oBAAoB,KAAA,KAAa,mBAAmB,sBACtD,cAAc,KAAA,KACZ,uBAAuB,KAAA,KAAa,sBAAsB,kBAC5D,gBAAgB,KAAA,KAAa,gBAAgB,eAAe;GAC/D,MAAM,gBAAiB,KAAK,kBAAkB,KAAK;GACnD,MAAM,UAAU,QAAQ,SAAS,WAAW;GAE5C,MAAM,iBACJ,OAAO,kBAAkB,YAAY,iBAAiB,KAAK,qBACvD,oBAAoB,KAAK,oBAAoB;IAC3C,aAAa,gBAAiB,sBAAsB;IACpD,GAAI,qBAAqB,EAAE,cAAc,mBAAmB,IAAI,CAAC;IACjE,cAAc;GAChB,CAAC,IACD,KAAA;GAEN,OAAO;IACL;IACA,cAAc,QAAQ,iBAAiB;IACvC,cAAc,QAAQ,KAAK,CAAC,CAAC,WAAW;IACxC,OAAO;KACL,cAAc,gBAAgB;KAC9B,kBAAkB,oBAAoB;KACtC,aAAa,gBAAgB,gBAAgB,MAAM,oBAAoB;KACvE,UAAU;KACV;KACA;IACF;IACA,SAAS,OAAO,kBAAkB,WAAW,gBAAiB,kBAAkB;IAChF,OAAQ,KAAK,SAAoB,IAAI;IACrC,YAAY,KAAK,IAAI,IAAI;IACzB,KAAK;GACP;EACF,SAAS,KAAK;GACZ,aAAa,aAAa;GAC1B,UAAU;GAIV,IAAI,cAAc,SAAS;IACzB,IAAI,QAAQ,CAAC,sBACX,MAAM,UAAU,MAAM,UAAU;KAC9B,SAAS,cAAc;KACvB,OAAO,cAAc;KACrB,QAAQ,cAAc;KACtB;KACA,OAAO,IAAI;KACX;KACA;KACA,cAAc;KACd,WAAW;KACX,WAAW,KAAK,IAAI;KACpB,YAAY,KAAK,IAAI,IAAI;KACzB,cAAc,eAAe,QAAQ,IAAI,UAAU,OAAO,GAAG;KAC7D,gBAAgB,CAAC;IACnB,CAAC;IAEH,MAAM;GACR;GACA,IAAI,QAAQ,CAAC,sBAIX,MAAM,UAAU,MAAM,UAAU;IAC9B,SAAS,cAAc;IACvB,OAAO,cAAc;IACrB,QAAQ,cAAc;IACtB;IACA,OAAO,IAAI;IACX;IACA;IACA,cAAc;IACd,WAAW;IACX,WAAW,KAAK,IAAI;IACpB,YAAY,KAAK,IAAI,IAAI;IACzB,cAAc,eAAe,QAAQ,IAAI,UAAU,OAAO,GAAG;IAC7D,gBAAgB,CAAC;GACnB,CAAC;GAEH,IACE,UAAU,kBAAkB,KAC5B,oBAAoB,GAAG,KACvB,CAAC,iBAAiB,eAAe,UAAU,GAC3C;IACA,MAAM,MAAM,UAAU,OAAO,CAAC;IAC9B;GACF;GACA,MAAM;EACR;CACF;CACA,MAAM,mBAAmB,QAAQ,UAAU,IAAI,MAAM,OAAO,OAAO,CAAC;AACtE;AAEA,eAAe,UACb,MACA,UACA,OACe;CAGf,IAAI;EACF,MAAM,KAAK,OAAO,SAAS,KAAK,CAAC;CACnC,QAAQ,CAER;AACF;AAEA,SAAS,gBAAgB,GAAoC;CAC3D,MAAM,MAA8B,CAAC;CACrC,EAAE,SAAS,OAAO,QAAQ;EACxB,IAAI,OAAO;CACb,CAAC;CACD,OAAO;AACT;AAEA,SAAS,gBAAwB;CAC/B,IAAI,OAAO,WAAW,QAAQ,eAAe,YAAY,OAAO,WAAW,OAAO,WAAW;CAC7F,OAAO,GAAG,KAAK,IAAI,CAAC,CAAC,SAAS,EAAE,EAAE,GAAG,KAAK,OAAO,CAAC,CAAC,SAAS,EAAE,CAAC,CAAC,MAAM,GAAG,EAAE;AAC7E;;;;;;;AAQA,eAAsB,YACpB,KACA,OAAyB,CAAC,GACoB;CAC9C,MAAM,SAAS,MAAM,kBAAkB,KAAK,IAAI;CAEhD,OAAO;EAAE,OADK,gBAAmB,QAAQ,KAAK,mBAAmB,SACpD;EAAG;CAAO;AACzB;;AAGA,eAAe,kBACb,KACA,OAAyB,CAAC,GACF;CACxB,IAAI;EACF,OAAO,MAAM,QAAQ;GAAE,GAAG;GAAK,UAAU,IAAI,YAAY,CAAC,IAAI;EAAW,GAAG,IAAI;CAClF,SAAS,KAAK;EACZ,IACE,KAAK,wBAAwB,iBAC7B,eAAe,gBACf,kBAAkB,IAAI,QAAQ,IAAI,IAAI,KACtC,IAAI,YAGJ,OAAO,MAAM,QAAQ;GADiB,GAAG;GAAK,UAAU;GAAM,YAAY,KAAA;EAC3C,GAAG,IAAI;EAExC,MAAM;CACR;AACF;AAEA,SAAS,gBACP,QACA,iBACG;CACH,IAAI;EACF,IAAI,OAAO,iBAAiB,UAC1B,MAAM,IAAI,MACR,8CAA8C,OAAO,MAAM,uBAC7D;EAEF,OAAO,gBAAmB,OAAO,SAAS,OAAO,OAAO,eAAe;CACzE,SAAS,OAAO;EACd,IAAI,iBAAiB,kBAAkB,MAAM;EAC7C,MAAM,QAAQ,iBAAiB,QAAQ,QAAQ,IAAI,MAAM,OAAO,KAAK,CAAC;EACtE,MAAM,IAAI,iBAAiB,MAAM,SAAS,QAAQ,EAAE,MAAM,CAAC;CAC7D;AACF;AAEA,SAAS,gBACP,SACA,OACA,iBACG;CACH,MAAM,UAAU,oBAAoB,UAAU,UAAU,mBAAmB,OAAO;CAClF,IAAI;EACF,OAAO,KAAK,MAAM,OAAO;CAC3B,SAAS,KAAK;EACZ,MAAM,IAAI,MACR,wCAAwC,MAAM,KAC5C,eAAe,QAAQ,IAAI,UAAU,OAAO,GAAG,EAChD,yBAAyB,QAAQ,MAAM,GAAG,GAAG,GAChD;CACF;AACF;AAWA,IAAa,yBAAb,cAA4C,sBAAsB;CAG9C;CACA;CAHlB,YACE,SACA,QACA,SACA;EACA,MAAM,OAAO;EAHG,KAAA,SAAA;EACA,KAAA,UAAA;CAGlB;AACF;;;;;;;;;;AAmCA,SAAgB,eAAe,MAAwB,MAA4B,CAAC,GAAS;CAC3F,MAAM,kBAAkB,KAAK,YAAY,KAAA;CACzC,MAAM,WAAW,KAAK,WAAW,iBAAA,CAAkB,QAAQ,QAAQ,EAAE;CAErE,IAAI,IAAI,0BAA0B,CAAC,iBACjC,MAAM,IAAI,uBACR,gGAAgG,iBAAiB,IACjH,wBACA,OACF;CAGF,IAAI,IAAI,iBAAiB,MAAM,MAAM,SAAS,SAAS,CAAC,CAAC,GACvD,MAAM,IAAI,uBACR,2BAA2B,QAAQ,8BACnC,oBACA,OACF;CAGF,IAAI,IAAI,mBAAmB,IAAI,gBAAgB,SAAS,GAElD;MAAA,CADO,IAAI,gBAAgB,MAAM,MAAM,SAAS,SAAS,CAAC,CACxD,GACJ,MAAM,IAAI,uBACR,2BAA2B,QAAQ,+BAA+B,IAAI,gBAAgB,IAAI,eAAe,CAAC,CAAC,KAAK,IAAI,EAAE,KACtH,wBACA,OACF;CAAA;CAIJ,IAAI,IAAI,eAAe,CAAC,KAAK,UAAU,CAAC,KAAK,UAAU,CAAC,KAAK,YAC3D,MAAM,IAAI,uBACR,sFACA,WACA,OACF;CAGF,IAAI,IAAI,kBAAkB;EACxB,MAAM,SAAS,KAAK,YAAY,oBAAoB,OAAO;EAC3D,IAAI,WAAW,IAAI,kBACjB,MAAM,IAAI,uBACR,qCAAqC,IAAI,iBAAiB,eAAe,QAAQ,eAAe,OAAO,IACvG,kBACA,OACF;CAEJ;AACF;AAEA,SAAS,SAAS,KAAa,SAAmC;CAChE,IAAI,mBAAmB,QAAQ,OAAO,QAAQ,KAAK,GAAG;CACtD,OAAO,IAAI,YAAY,CAAC,CAAC,WAAW,QAAQ,YAAY,CAAC;AAC3D;AAEA,SAAS,gBAAgB,GAA4B;CACnD,OAAO,aAAa,SAAS,EAAE,SAAS;AAC1C;;;;;;;;;;;;AAaA,eAAsB,SACpB,OACA,OAAkD,CAAC,GACgB;CACnE,MAAM,QAAQ,KAAK,IAAI;CACvB,IAAI;EACF,MAAM,QACJ;GACE;GACA,UAAU,CAAC;IAAE,MAAM;IAAQ,SAAS;GAAO,CAAC;GAC5C,WAAW;GACX,WAAW,KAAK,aAAa;EAC/B,GACA,IACF;EACA,OAAO;GAAE,IAAI;GAAM,WAAW,KAAK,IAAI,IAAI;GAAO,OAAO;EAAK;CAChE,SAAS,KAAK;EACZ,OAAO;GACL,IAAI;GACJ,WAAW,KAAK,IAAI,IAAI;GACxB,OAAO,eAAe,QAAQ,IAAI,UAAU,OAAO,GAAG;EACxD;CACF;AACF;;;;;;AAOA,IAAa,YAAb,MAAuB;CACrB;CACA;CAEA,YAAY,OAAyB,CAAC,GAAG;EACvC,KAAK,OAAO;EACZ,KAAK,kBAAkB,uBAAuB,KAAK,eAAe;CACpE;CAEA,KAAK,KAAqB,KAAgD;EACxE,MAAM,UAAU;GAAE,GAAG,KAAK;GAAM,GAAG;EAAI;EACvC,OAAO,IAAI,aAAa,kBAAkB,KAAK,OAAO,IAAI,QAAQ,KAAK,OAAO;CAChF;CAEA,SACE,KACA,KAC8C;EAC9C,OAAO,YAAe,KAAK;GAAE,GAAG,KAAK;GAAM,GAAG;EAAI,CAAC;CACrD;AACF"}
+{"version":3,"file":"llm-client-DHx8pzyJ.js","names":[],"sources":["../src/llm-client.ts"],"sourcesContent":["/**\n * LLM client with graceful degrade.\n *\n * OpenAI-compatible `/v1/chat/completions` client with:\n *   - Exponential-backoff retry on 429 + 5xx gateway errors (502/503/504).\n *   - Retry on transient network errors (fetch failed, AbortError, ECONNRESET).\n *   - One retry at temperature 1 when a model explicitly requires it.\n *   - Graceful json_schema → json_object degrade on 400 with schema-reject body.\n *   - Fenced-JSON stripping (```json ... ```) for models that wrap structured output.\n *   - Configurable base URL + api key / bearer, works with LiteLLM proxies, OpenAI\n *     directly, cli-bridge subscriptions, and any router that speaks the spec.\n *\n * Usage:\n *   const { value, result } = await callLlmJson<MyType>(\n *     { model: 'gpt-4o', messages: [...], jsonSchema: { name: 'x', schema: {...} } },\n *     { baseUrl: 'https://router.tangle.tools/v1', apiKey: process.env.KEY },\n *   )\n *\n * `createChatClient` wraps this implementation for provider-neutral package\n * entry points. Direct callers can use `callLlm` or `callLlmJson`.\n */\n\nimport {\n  type CostReceiptInput,\n  type CustomTokenPricing,\n  costForTokenPricing,\n  type MaximumCharge,\n} from './cost-ledger'\nimport { AgentEvalError, CaptureIntegrityError } from './errors'\nimport {\n  defaultProviderRedactor,\n  type ProviderRedactor,\n  providerFromBaseUrl,\n  type RawProviderEvent,\n  type RawProviderSink,\n} from './trace/raw-provider-sink'\n\n// ─── Types ──────────────────────────────────────────────────────────────\n\nexport interface LlmMessage {\n  role: 'system' | 'user' | 'assistant'\n  /**\n   * Either a plain text content string OR a multimodal content array\n   * (text + image_url parts) for vision-capable models.\n   */\n  content:\n    | string\n    | Array<\n        | { type: 'text'; text: string }\n        | { type: 'image_url'; image_url: { url: string; detail?: 'auto' | 'low' | 'high' } }\n      >\n}\n\nexport type LlmThinkingMode = 'enabled' | 'disabled'\n\nexport interface LlmCallRequest {\n  model: string\n  messages: LlmMessage[]\n  /** Optional JSON-mode response format (response_format: json_object). */\n  jsonMode?: boolean\n  /** Optional structured output via JSON Schema. Falls back to json_object on 400. */\n  jsonSchema?: { name: string; schema: Record<string, unknown> }\n  temperature?: number\n  maxTokens?: number\n  /** OpenAI-compatible reasoning mode. Omitted when the provider default should apply. */\n  thinking?: LlmThinkingMode\n  /** Per-call timeout, default 300s. */\n  timeoutMs?: number\n}\n\n/** Conservative priced bound for the exact text request sent to a provider.\n * Returns undefined when output or multimodal input is not bounded, causing a\n * capped CostLedger to reject the call before execution. Pass\n * `customTokenPricing` when package pricing does not cover the model or endpoint. */\nexport function maximumChargeForLlmRequest(\n  request: Pick<LlmCallRequest, 'model' | 'messages' | 'jsonSchema' | 'maxTokens' | 'thinking'>,\n  options: LlmClientOptions = {},\n): MaximumCharge | undefined {\n  if (request.maxTokens === undefined) return undefined\n  if (!Number.isInteger(request.maxTokens) || request.maxTokens <= 0) {\n    throw new RangeError(`maximumChargeForLlmRequest: maxTokens must be a positive integer`)\n  }\n  if (\n    request.messages.some(\n      (message) =>\n        Array.isArray(message.content) && message.content.some((part) => part.type === 'image_url'),\n    )\n  ) {\n    return undefined\n  }\n\n  const attempts = resolveMaximumAttempts(options.maximumAttempts)\n  const forceJsonObject = options.jsonSchemaTransport === 'json-object'\n  // A byte-level tokenizer cannot emit more input tokens than request bytes.\n  // Pricing the complete body also covers role/schema framing omitted from content-only estimates.\n  const requestBytes = new TextEncoder().encode(\n    JSON.stringify(buildBody(request, forceJsonObject, options.thinking)),\n  ).byteLength\n  // A rejected response schema can trigger one JSON-mode batch with the same output limit.\n  const batches = request.jsonSchema && !forceJsonObject ? 2 : 1\n  const usage = {\n    inputTokens: requestBytes * attempts * batches,\n    outputTokens: request.maxTokens * attempts * batches,\n  }\n  return options.customTokenPricing\n    ? { customTokenPricing: options.customTokenPricing, ...usage }\n    : { model: request.model, ...usage }\n}\n\nexport interface LlmUsage {\n  promptTokens: number\n  completionTokens: number\n  totalTokens: number\n  /** False when the provider omitted or malformed prompt/completion usage. */\n  captured?: boolean\n  /** Reasoning-token subset of completionTokens, when reported. */\n  reasoningTokens?: number\n  /** Proxies populate this when prompt caching is on. */\n  cachedPromptTokens?: number\n}\n\nexport interface LlmCallResult {\n  /** The text content of the first choice. Empty string if none. */\n  content: string\n  usage: LlmUsage\n  /**\n   * Cost in USD. Uses the provider's reported cost when present, otherwise\n   * caller-supplied token pricing. `null` when neither is available.\n   */\n  costUsd: number | null\n  /** Model name actually used (echoed from response). */\n  model: string\n  /** Wall-clock duration of the HTTP call (last attempt, if retried). */\n  durationMs: number\n  /**\n   * `finish_reason` echoed from the first choice (`stop`, `length`,\n   * `content_filter`, `tool_calls`, ...). `null` when the provider omits it.\n   * Exposed so a free-form `callLlm` caller CAN detect a truncated answer\n   * (`length`) instead of treating a cut-off completion as complete. Note:\n   * `callLlm` does not itself reject on it — acting on this signal is the\n   * caller's responsibility (in-repo free-form drivers do not yet enforce it).\n   */\n  finishReason?: string | null\n  /**\n   * True when `content.trim()` is empty. An empty completion is a silent zero\n   * for free-form `callLlm` callers; this flag is the signal a caller can\n   * inspect to fail loud rather than proceed on an empty string. `callLlm`\n   * surfaces it but does not throw on it.\n   */\n  contentEmpty?: boolean\n  /** Raw response body. */\n  raw: Record<string, unknown>\n}\n\nexport type LlmCallMetadata = Pick<LlmCallResult, 'usage' | 'costUsd' | 'model' | 'durationMs'>\n\n/** Convert a provider result into the canonical paid-call receipt input. */\nexport function costReceiptFromLlm(\n  result: LlmCallResult,\n  customTokenPricing?: CustomTokenPricing,\n): CostReceiptInput {\n  const cachedTokens = result.usage.cachedPromptTokens ?? 0\n  const inputTokens = Math.max(0, result.usage.promptTokens - cachedTokens)\n  const configuredCostUsd =\n    result.costUsd === null && customTokenPricing && result.usage.captured !== false\n      ? costForTokenPricing(customTokenPricing, {\n          inputTokens,\n          ...(cachedTokens > 0 ? { cachedTokens } : {}),\n          outputTokens: result.usage.completionTokens,\n        })\n      : undefined\n  return {\n    model: result.model,\n    inputTokens,\n    outputTokens: result.usage.completionTokens,\n    reasoningTokens: result.usage.reasoningTokens,\n    cachedTokens: cachedTokens > 0 ? cachedTokens : undefined,\n    actualCostUsd: result.costUsd ?? configuredCostUsd,\n    usageUnknown: result.usage.captured === false,\n  }\n}\n\n/** Structured-response failures retain their completed provider receipt. */\nexport function costReceiptFromLlmError(\n  error: Error,\n  customTokenPricing?: CustomTokenPricing,\n): CostReceiptInput | undefined {\n  return error instanceof LlmResponseError\n    ? costReceiptFromLlm(error.result, customTokenPricing)\n    : undefined\n}\n\nexport class LlmCallError extends AgentEvalError {\n  constructor(\n    message: string,\n    public readonly status: number,\n    public readonly body: string,\n    public readonly model: string,\n  ) {\n    super('judge', message)\n  }\n}\n\n/** A provider response completed and incurred measurable usage, but its content\n *  could not satisfy the caller's response contract. The response envelope is\n *  retained so accounting can commit the receipt before the error propagates. */\nexport class LlmResponseError extends AgentEvalError {\n  constructor(\n    message: string,\n    public readonly result: LlmCallResult,\n    options?: { cause?: unknown },\n  ) {\n    super('judge', message, options)\n  }\n}\n\nexport interface LlmClientOptions {\n  /** Base URL (without trailing slash). Must end at the `/v1` prefix. */\n  baseUrl?: string\n  /** Bearer token — either `apiKey` or `bearer` populates `Authorization: Bearer ...`. */\n  apiKey?: string\n  bearer?: string\n  /** Override for the `Authorization` header (e.g. `X-Auth: ...`). Takes precedence over apiKey/bearer. */\n  authHeader?: { name: string; value: string }\n  /** Stable provider idempotency key, reused across retries of this logical call. */\n  idempotencyKey?: string\n  /** Default timeout in ms. Per-call can override. */\n  defaultTimeoutMs?: number\n  /**\n   * Caller-supplied abort signal — e.g. a campaign-wide cancel. Linked to\n   * each attempt's per-attempt timeout controller, so aborting it cancels\n   * the in-flight fetch. A caller abort is FATAL: it is not retried even\n   * though an AbortError otherwise matches the transient patterns.\n   */\n  signal?: AbortSignal\n  /**\n   * Cross-attempt wall-clock budget in ms, measured from the first attempt.\n   * Before launching each attempt the loop checks the remaining budget and\n   * stops retrying once it is exhausted, rather than waiting the full\n   * per-attempt timeout on every retry. Bounds total time independent of\n   * total attempts × `timeoutMs`.\n   */\n  deadlineMs?: number\n  /** Total provider attempts. Default 3. */\n  maximumAttempts?: number\n  /** Token rates used when the provider omits cost or package pricing does not cover the model. */\n  customTokenPricing?: CustomTokenPricing\n  /**\n   * Transport for requests that declare `jsonSchema`. `native` sends\n   * `response_format: json_schema`; `json-object` sends the broadly supported\n   * JSON mode and relies on the caller to include the schema in model-visible\n   * instructions. Default: `native`.\n   */\n  jsonSchemaTransport?: 'native' | 'json-object'\n  /**\n   * JSON payload parsing policy. `extract` accepts fenced or prose-prefixed JSON.\n   * `exact` requires the complete response content to be one JSON value.\n   * Default: `extract`.\n   */\n  jsonPayloadMode?: 'extract' | 'exact'\n  /** Default provider reasoning mode. A per-call request value takes precedence. */\n  thinking?: LlmThinkingMode\n  /** Fetch implementation — defaults to global `fetch`. Override for custom transport (e.g. tests). */\n  fetch?: typeof fetch\n  /**\n   * Optional raw HTTP capture sink. When provided, every request, response,\n   * and error (across all retry attempts) is recorded to the sink, with auth\n   * headers and credential-shaped body fields redacted by default. This is\n   * the layer-1 forensics primitive: structured `LlmSpan`s record intent,\n   * raw events record what actually crossed the wire.\n   */\n  rawSink?: RawProviderSink\n  /**\n   * Logical provider id attached to raw events. When omitted, derived from\n   * `baseUrl` via `providerFromBaseUrl`.\n   */\n  provider?: string\n  /** Trace context attached to raw events; populated by emitter-aware callers. */\n  traceContext?: { runId?: string; spanId?: string }\n  /** Override the redaction strategy for this call. Defaults to `defaultProviderRedactor`. */\n  redactor?: ProviderRedactor\n}\n\n// ─── Internals ──────────────────────────────────────────────────────────\n\nconst DEFAULT_BASE_URL = 'https://router.tangle.tools/v1'\n// Flagship / reasoning models routinely take several minutes on large prompts (a\n// reflection over many failures, a long tool transcript). A tight cap aborts a\n// legitimately-slow but healthy call — and because every retry attempt re-uses\n// the same window, such a model aborts on ALL attempts and the loop throws. The\n// default is generous enough to let those complete, bounded enough that a truly\n// hung call still fails over after retries, and tunable per deployment via\n// TANGLE_LLM_TIMEOUT_MS. Per-call `req.timeoutMs` / `opts.defaultTimeoutMs`\n// still win for callers that know their model's latency.\nconst DEFAULT_TIMEOUT_MS = Number(process.env.TANGLE_LLM_TIMEOUT_MS) || 300_000\nconst DEFAULT_MAXIMUM_ATTEMPTS =\n  process.env.TANGLE_LLM_MAXIMUM_ATTEMPTS === undefined\n    ? 3\n    : Number(process.env.TANGLE_LLM_MAXIMUM_ATTEMPTS)\n\nfunction resolveMaximumAttempts(configured: number | undefined): number {\n  const attempts = configured ?? DEFAULT_MAXIMUM_ATTEMPTS\n  if (!Number.isInteger(attempts) || attempts <= 0) {\n    throw new RangeError('LLM maximum attempts must be a positive integer')\n  }\n  return attempts\n}\n\nfunction providerTokenCount(value: unknown): number | undefined {\n  return typeof value === 'number' && Number.isSafeInteger(value) && value >= 0 ? value : undefined\n}\n\nconst RETRYABLE_STATUS = new Set([429, 502, 503, 504])\n\n/**\n * Transient transport/network error signatures, matched against an error's\n * name, message, and `code`. Covers fetch/undici network failures, aborts\n * and timeouts, and — critically — HTTP/2 transport faults a keep-alive\n * connection raises mid-response: `terminated`, `NGHTTP2_INTERNAL_ERROR`,\n * `UND_ERR_*`, `other side closed`. Those last ones carry no clean HTTP\n * status; unrecognised, they escape the retry loop and surface as an\n * uncaught rejection.\n */\nconst TRANSIENT_ERROR_PATTERNS: readonly RegExp[] = [\n  /AbortError/i,\n  /TimeoutError/i,\n  /this operation was aborted/i,\n  /fetch failed/i,\n  /ECONNRESET/i,\n  /ETIMEDOUT/i,\n  /EAI_AGAIN/i,\n  /socket hang up/i,\n  /stream.*ended.*unexpectedly/i,\n  /terminated/i,\n  /other side closed/i,\n  /NGHTTP2/i,\n  /UND_ERR/i,\n]\n\n/**\n * True when an error is a transient transport/network fault worth retrying,\n * as opposed to a deterministic failure (4xx schema reject, JSON parse) that\n * a retry cannot fix. Inspects `LlmCallError.status`, then the error's\n * name/message/code, then recurses into `error.cause` — undici nests the\n * real socket fault one or more levels under `.cause`.\n *\n * This is the retry classifier for the package: `callLlm` and\n * `withJudgeRetry` both route through it, so connection failures are treated\n * consistently across transports.\n */\nexport function isTransientLlmError(err: unknown): boolean {\n  return classifyTransient(err, 0)\n}\n\nfunction classifyTransient(err: unknown, depth: number): boolean {\n  if (err instanceof LlmCallError) return RETRYABLE_STATUS.has(err.status)\n  if (!(err instanceof Error)) return false\n  // Foreign transport errors can carry a numeric HTTP status without being an\n  // LlmCallError. A retryable status is decisive.\n  const status = (err as { status?: unknown }).status\n  if (typeof status === 'number' && RETRYABLE_STATUS.has(status)) return true\n  const code = (err as { code?: unknown }).code\n  const haystack = `${err.name}\\n${err.message}\\n${typeof code === 'string' ? code : ''}`\n  if (TRANSIENT_ERROR_PATTERNS.some((p) => p.test(haystack))) return true\n  const cause = (err as { cause?: unknown }).cause\n  if (depth < 4 && cause instanceof Error && cause !== err) {\n    return classifyTransient(cause, depth + 1)\n  }\n  return false\n}\n\nfunction parseRetryAfter(headers: Headers): number | null {\n  const h = headers.get('retry-after')\n  if (!h) return null\n  const asNumber = Number(h)\n  if (Number.isFinite(asNumber) && asNumber > 0) return asNumber * 1000\n  const asDate = Date.parse(h)\n  if (Number.isFinite(asDate)) return Math.max(0, asDate - Date.now())\n  return null\n}\n\n/** Exponential backoff: 500ms, 1s, 2s, 4s, ... capped at 16s. Attempt is 0-indexed. */\nexport function backoffMs(attempt: number): number {\n  return Math.min(500 * 2 ** attempt, 16_000)\n}\n\nfunction buildHeaders(opts: LlmClientOptions): Record<string, string> {\n  const headers: Record<string, string> = {\n    'Content-Type': 'application/json',\n    Accept: 'application/json',\n  }\n  if (opts.authHeader) {\n    headers[opts.authHeader.name] = opts.authHeader.value\n  } else if (opts.bearer || opts.apiKey) {\n    headers.Authorization = `Bearer ${opts.bearer ?? opts.apiKey}`\n  }\n  if (opts.idempotencyKey) headers['Idempotency-Key'] = opts.idempotencyKey\n  return headers\n}\n\nfunction isSchemaRejection(status: number, body: string): boolean {\n  if (status !== 400) return false\n  const lower = body.toLowerCase()\n  return (\n    lower.includes('response_format') ||\n    lower.includes('json_schema') ||\n    lower.includes('is unavailable') ||\n    lower.includes('not supported')\n  )\n}\n\nfunction isTemperatureOneRejection(status: number, body: string): boolean {\n  if (status !== 400 || !/temperature/i.test(body)) return false\n  return (\n    /temperature[^.\\n]{0,120}\\b(?:only|must|should|required|requires?)\\b[^.\\n]{0,40}\\b1(?:\\.0+)?\\b/i.test(\n      body,\n    ) || /\\bonly\\s+1(?:\\.0+)?\\s+is\\s+allowed\\b[^.\\n]{0,120}\\btemperature\\b/i.test(body)\n  )\n}\n\nfunction buildBody(\n  req: LlmCallRequest,\n  forceJsonObject: boolean,\n  defaultThinking?: LlmThinkingMode,\n): Record<string, unknown> {\n  const body: Record<string, unknown> = {\n    model: req.model,\n    messages: req.messages,\n    temperature: req.temperature ?? 0,\n  }\n  if (req.maxTokens != null) {\n    if (usesMaxCompletionTokens(req.model)) body.max_completion_tokens = req.maxTokens\n    else body.max_tokens = req.maxTokens\n  }\n  const thinking = req.thinking ?? defaultThinking\n  if (thinking !== undefined) {\n    body.thinking = { type: thinking }\n  }\n\n  if (req.jsonSchema && !forceJsonObject) {\n    body.response_format = {\n      type: 'json_schema',\n      json_schema: { name: req.jsonSchema.name, schema: req.jsonSchema.schema, strict: true },\n    }\n  } else if (req.jsonMode || req.jsonSchema) {\n    body.response_format = { type: 'json_object' }\n  }\n\n  return body\n}\n\nfunction usesMaxCompletionTokens(model: string): boolean {\n  return /^gpt-5(?:[.-]|$)/i.test(model)\n}\n\nasync function sleep(ms: number): Promise<void> {\n  return new Promise((resolve) => setTimeout(resolve, ms))\n}\n\n/**\n * Combine the per-attempt timeout signal with an optional caller signal into\n * one signal the fetch listens on. Prefers the native `AbortSignal.any`; falls\n * back to manual wiring on runtimes that predate it. The caller signal is also\n * propagated to the timeout controller so aborting it cancels the in-flight\n * fetch immediately.\n */\nfunction linkSignals(timeoutController: AbortController, caller?: AbortSignal): AbortSignal {\n  if (!caller) return timeoutController.signal\n  if (typeof (AbortSignal as { any?: unknown }).any === 'function') {\n    return AbortSignal.any([timeoutController.signal, caller])\n  }\n  if (caller.aborted) {\n    timeoutController.abort()\n  } else {\n    caller.addEventListener('abort', () => timeoutController.abort(), { once: true })\n  }\n  return timeoutController.signal\n}\n\n/** True once the cross-attempt wall-clock budget (if any) is exhausted. */\nfunction deadlineExceeded(start: number, deadlineMs: number | undefined): boolean {\n  return deadlineMs != null && Date.now() - start >= deadlineMs\n}\n\n// ─── Public API ─────────────────────────────────────────────────────────\n\n/**\n * Strip a ```json / ``` code fence if the model emitted one.\n * Idempotent for naked JSON. Some models (claude-code via router, certain\n * deepseek models) wrap output even under json_object.\n */\nexport function stripFencedJson(raw: string): string {\n  const trimmed = raw.trim()\n  const m = trimmed.match(/^```(?:json)?\\s*\\n?([\\s\\S]*?)\\n?```\\s*$/)\n  return m ? m[1]!.trim() : trimmed\n}\n\nexport function extractJsonPayload(raw: string): string {\n  const stripped = stripFencedJson(raw)\n  try {\n    JSON.parse(stripped)\n    return stripped\n  } catch {\n    // A response that declares a JSON root must parse as that complete root.\n    // Scanning onward could turn a truncated object into one of its valid nested\n    // arrays or objects and silently change the response schema.\n    if (stripped.startsWith('{') || stripped.startsWith('[')) return stripped\n  }\n\n  // Only prose-leading responses may contain a recoverable JSON payload.\n  const starts = [...stripped.matchAll(/[[{]/g)]\n    .map((match) => match.index)\n    .filter((index) => index != null)\n  for (const start of starts) {\n    const candidate = extractBalancedJson(stripped, start)\n    if (!candidate) continue\n    try {\n      JSON.parse(candidate)\n      return candidate\n    } catch {\n      // Keep scanning; earlier braces may belong to prose.\n    }\n  }\n\n  return stripped\n}\n\nfunction extractBalancedJson(input: string, start: number): string | null {\n  const opener = input[start]\n  const closer = opener === '{' ? '}' : opener === '[' ? ']' : null\n  if (!closer) return null\n\n  const stack: string[] = [closer]\n  let isInString = false\n  let isEscaped = false\n\n  for (let i = start + 1; i < input.length; i++) {\n    const char = input[i]!\n    if (isEscaped) {\n      isEscaped = false\n      continue\n    }\n    if (char === '\\\\') {\n      isEscaped = isInString\n      continue\n    }\n    if (char === '\"') {\n      isInString = !isInString\n      continue\n    }\n    if (isInString) continue\n\n    if (char === '{') stack.push('}')\n    else if (char === '[') stack.push(']')\n    else if (char === stack[stack.length - 1]) {\n      stack.pop()\n      if (stack.length === 0) return input.slice(start, i + 1)\n    }\n  }\n\n  return null\n}\n\n/**\n * Low-level call. Returns raw content + usage + cost. Retries on transient\n * failures; does NOT degrade schema here — callers that want graceful\n * degrade use `callLlmJson`.\n */\nexport async function callLlm(\n  req: LlmCallRequest,\n  opts: LlmClientOptions = {},\n): Promise<LlmCallResult> {\n  const baseUrl = (opts.baseUrl ?? DEFAULT_BASE_URL).replace(/\\/+$/, '')\n  const url = `${baseUrl}/chat/completions`\n  const endpoint = '/chat/completions'\n  const timeoutMs = req.timeoutMs ?? opts.defaultTimeoutMs ?? DEFAULT_TIMEOUT_MS\n  const maximumAttempts = resolveMaximumAttempts(opts.maximumAttempts)\n  const fetchFn = opts.fetch ?? globalThis.fetch\n  const headers = buildHeaders(opts)\n  const provider = opts.provider ?? providerFromBaseUrl(baseUrl)\n  const sink = opts.rawSink\n  const redactor = opts.redactor ?? defaultProviderRedactor\n  const traceContext = opts.traceContext\n  const callerSignal = opts.signal\n  const deadlineMs = opts.deadlineMs\n  const deadlineStart = Date.now()\n  if (opts.customTokenPricing) {\n    costForTokenPricing(opts.customTokenPricing, { inputTokens: 0, outputTokens: 0 })\n  }\n\n  let lastErr: unknown\n  let effectiveRequest = req\n  for (let attempt = 0; attempt < maximumAttempts; attempt++) {\n    // A caller cancel is fatal — never retried. Checking before each attempt\n    // means an already-aborted signal short-circuits without firing fetch.\n    if (callerSignal?.aborted) {\n      throw new DOMException('callLlm aborted by caller signal', 'AbortError')\n    }\n    // Stop retrying once the cross-attempt budget is spent rather than burning\n    // a full per-attempt timeout on each remaining retry.\n    if (attempt > 0 && deadlineExceeded(deadlineStart, deadlineMs)) {\n      throw lastErr instanceof Error ? lastErr : new Error(String(lastErr))\n    }\n    const controller = new AbortController()\n    const attemptSignal = linkSignals(controller, callerSignal)\n    const timeoutHandle = setTimeout(() => controller.abort(), timeoutMs)\n    const started = Date.now()\n    const requestBody = buildBody(\n      effectiveRequest,\n      opts.jsonSchemaTransport === 'json-object',\n      opts.thinking,\n    )\n    let attemptErrorRecorded = false\n    if (sink) {\n      await recordRaw(sink, redactor, {\n        eventId: cryptoEventId(),\n        runId: traceContext?.runId,\n        spanId: traceContext?.spanId,\n        provider,\n        model: req.model,\n        endpoint,\n        baseUrl,\n        attemptIndex: attempt,\n        direction: 'request',\n        timestamp: started,\n        requestHeaders: headers,\n        requestBody,\n        redactedFields: [],\n      })\n    }\n\n    try {\n      const res = await fetchFn(url, {\n        method: 'POST',\n        headers,\n        body: JSON.stringify(requestBody),\n        signal: attemptSignal,\n      })\n      clearTimeout(timeoutHandle)\n      const responseHeaders = sink ? headersToObject(res.headers) : undefined\n\n      if (!res.ok) {\n        const body = await res.text()\n        if (sink) {\n          await recordRaw(sink, redactor, {\n            eventId: cryptoEventId(),\n            runId: traceContext?.runId,\n            spanId: traceContext?.spanId,\n            provider,\n            model: req.model,\n            endpoint,\n            baseUrl,\n            attemptIndex: attempt,\n            direction: 'error',\n            timestamp: Date.now(),\n            durationMs: Date.now() - started,\n            statusCode: res.status,\n            responseHeaders,\n            responseBody: body,\n            errorMessage: `HTTP ${res.status}`,\n            redactedFields: [],\n          })\n          attemptErrorRecorded = true\n        }\n        const err = new LlmCallError(\n          `LLM call ${res.status}: ${body.slice(0, 300)}`,\n          res.status,\n          body,\n          req.model,\n        )\n        if (\n          isTemperatureOneRejection(res.status, body) &&\n          effectiveRequest.temperature !== 1 &&\n          attempt < maximumAttempts - 1 &&\n          !deadlineExceeded(deadlineStart, deadlineMs)\n        ) {\n          lastErr = err\n          effectiveRequest = { ...effectiveRequest, temperature: 1 }\n          continue\n        }\n        if (\n          RETRYABLE_STATUS.has(res.status) &&\n          attempt < maximumAttempts - 1 &&\n          !deadlineExceeded(deadlineStart, deadlineMs)\n        ) {\n          lastErr = err\n          const retryAfter = parseRetryAfter(res.headers)\n          await sleep(retryAfter ?? backoffMs(attempt))\n          continue\n        }\n        throw err\n      }\n\n      const text = await res.text()\n      let json: Record<string, unknown>\n      try {\n        json = JSON.parse(text) as Record<string, unknown>\n      } catch (parseErr) {\n        if (sink) {\n          await recordRaw(sink, redactor, {\n            eventId: cryptoEventId(),\n            runId: traceContext?.runId,\n            spanId: traceContext?.spanId,\n            provider,\n            model: req.model,\n            endpoint,\n            baseUrl,\n            attemptIndex: attempt,\n            direction: 'error',\n            timestamp: Date.now(),\n            durationMs: Date.now() - started,\n            statusCode: res.status,\n            responseHeaders,\n            responseBody: text,\n            errorMessage: `non-JSON response: ${parseErr instanceof Error ? parseErr.message : String(parseErr)}`,\n            redactedFields: [],\n          })\n          attemptErrorRecorded = true\n        }\n        throw parseErr\n      }\n      if (sink) {\n        await recordRaw(sink, redactor, {\n          eventId: cryptoEventId(),\n          runId: traceContext?.runId,\n          spanId: traceContext?.spanId,\n          provider,\n          model: req.model,\n          endpoint,\n          baseUrl,\n          attemptIndex: attempt,\n          direction: 'response',\n          timestamp: Date.now(),\n          durationMs: Date.now() - started,\n          statusCode: res.status,\n          responseHeaders,\n          responseBody: json,\n          redactedFields: [],\n        })\n      }\n      const choice = (\n        json.choices as\n          | Array<{ message?: { content?: string }; finish_reason?: string | null }>\n          | undefined\n      )?.[0]\n      const usageRaw =\n        json.usage && typeof json.usage === 'object' && !Array.isArray(json.usage)\n          ? (json.usage as Record<string, unknown>)\n          : undefined\n      const promptTokens = providerTokenCount(usageRaw?.prompt_tokens)\n      const completionTokens = providerTokenCount(usageRaw?.completion_tokens)\n      const totalTokens = providerTokenCount(usageRaw?.total_tokens)\n      const completionDetails =\n        usageRaw?.completion_tokens_details &&\n        typeof usageRaw.completion_tokens_details === 'object' &&\n        !Array.isArray(usageRaw.completion_tokens_details)\n          ? (usageRaw.completion_tokens_details as Record<string, unknown>)\n          : undefined\n      const reasoningRaw = completionDetails?.reasoning_tokens\n      const reasoningTokens =\n        reasoningRaw === undefined ? undefined : providerTokenCount(reasoningRaw)\n      const cachedRaw =\n        usageRaw?.prompt_tokens_details &&\n        typeof usageRaw.prompt_tokens_details === 'object' &&\n        !Array.isArray(usageRaw.prompt_tokens_details)\n          ? (usageRaw.prompt_tokens_details as Record<string, unknown>).cached_tokens\n          : undefined\n      const cachedPromptTokens = cachedRaw === undefined ? undefined : providerTokenCount(cachedRaw)\n      const usageCaptured =\n        promptTokens !== undefined &&\n        completionTokens !== undefined &&\n        (reasoningRaw === undefined ||\n          (reasoningTokens !== undefined && reasoningTokens <= completionTokens)) &&\n        (cachedRaw === undefined ||\n          (cachedPromptTokens !== undefined && cachedPromptTokens <= promptTokens)) &&\n        (totalTokens === undefined || totalTokens === promptTokens + completionTokens)\n      const costFromProxy = (json._response_cost ?? json.cost_usd) as number | undefined\n      const content = choice?.message?.content ?? ''\n\n      const configuredCost =\n        typeof costFromProxy !== 'number' && usageCaptured && opts.customTokenPricing\n          ? costForTokenPricing(opts.customTokenPricing, {\n              inputTokens: promptTokens! - (cachedPromptTokens ?? 0),\n              ...(cachedPromptTokens ? { cachedTokens: cachedPromptTokens } : {}),\n              outputTokens: completionTokens!,\n            })\n          : undefined\n\n      return {\n        content,\n        finishReason: choice?.finish_reason ?? null,\n        contentEmpty: content.trim().length === 0,\n        usage: {\n          promptTokens: promptTokens ?? 0,\n          completionTokens: completionTokens ?? 0,\n          totalTokens: totalTokens ?? (promptTokens ?? 0) + (completionTokens ?? 0),\n          captured: usageCaptured,\n          reasoningTokens,\n          cachedPromptTokens,\n        },\n        costUsd: typeof costFromProxy === 'number' ? costFromProxy : (configuredCost ?? null),\n        model: (json.model as string) ?? req.model,\n        durationMs: Date.now() - started,\n        raw: json,\n      }\n    } catch (err) {\n      clearTimeout(timeoutHandle)\n      lastErr = err\n      // A caller cancel is fatal even though an AbortError matches the\n      // transient patterns — a cancelled call must surface immediately, not\n      // be retried against the same dead intent.\n      if (callerSignal?.aborted) {\n        if (sink && !attemptErrorRecorded) {\n          await recordRaw(sink, redactor, {\n            eventId: cryptoEventId(),\n            runId: traceContext?.runId,\n            spanId: traceContext?.spanId,\n            provider,\n            model: req.model,\n            endpoint,\n            baseUrl,\n            attemptIndex: attempt,\n            direction: 'error',\n            timestamp: Date.now(),\n            durationMs: Date.now() - started,\n            errorMessage: err instanceof Error ? err.message : String(err),\n            redactedFields: [],\n          })\n        }\n        throw err\n      }\n      if (sink && !attemptErrorRecorded) {\n        // Record only if neither the !res.ok branch nor the JSON.parse catch\n        // already produced an error event for this attempt. Covers network\n        // failures, timeouts, and aborts.\n        await recordRaw(sink, redactor, {\n          eventId: cryptoEventId(),\n          runId: traceContext?.runId,\n          spanId: traceContext?.spanId,\n          provider,\n          model: req.model,\n          endpoint,\n          baseUrl,\n          attemptIndex: attempt,\n          direction: 'error',\n          timestamp: Date.now(),\n          durationMs: Date.now() - started,\n          errorMessage: err instanceof Error ? err.message : String(err),\n          redactedFields: [],\n        })\n      }\n      if (\n        attempt < maximumAttempts - 1 &&\n        isTransientLlmError(err) &&\n        !deadlineExceeded(deadlineStart, deadlineMs)\n      ) {\n        await sleep(backoffMs(attempt))\n        continue\n      }\n      throw err\n    }\n  }\n  throw lastErr instanceof Error ? lastErr : new Error(String(lastErr))\n}\n\nasync function recordRaw(\n  sink: RawProviderSink,\n  redactor: ProviderRedactor,\n  event: RawProviderEvent,\n): Promise<void> {\n  // Errors from sinks must not crash the LLM call. Forensic capture is\n  // best-effort; the structured trace is the system of record.\n  try {\n    await sink.record(redactor(event))\n  } catch {\n    // Intentionally swallowed.\n  }\n}\n\nfunction headersToObject(h: Headers): Record<string, string> {\n  const out: Record<string, string> = {}\n  h.forEach((value, key) => {\n    out[key] = value\n  })\n  return out\n}\n\nfunction cryptoEventId(): string {\n  if (typeof globalThis.crypto?.randomUUID === 'function') return globalThis.crypto.randomUUID()\n  return `${Date.now().toString(36)}-${Math.random().toString(36).slice(2, 10)}`\n}\n\n/**\n * Structured-output call. Returns parsed JSON plus the raw result envelope.\n * Degrades `jsonSchema` → `jsonMode` on a 400 that names the schema param —\n * critical for deepseek-v3/v4, kimi-k2.6, and other models that don't accept\n * the `response_format.json_schema` shape but DO accept `json_object`.\n */\nexport async function callLlmJson<T = unknown>(\n  req: LlmCallRequest,\n  opts: LlmClientOptions = {},\n): Promise<{ value: T; result: LlmCallResult }> {\n  const result = await callLlmStructured(req, opts)\n  const value = parseJsonResult<T>(result, opts.jsonPayloadMode ?? 'extract')\n  return { value, result }\n}\n\n/** Shared schema-to-JSON-mode fallback that preserves the raw result. */\nasync function callLlmStructured(\n  req: LlmCallRequest,\n  opts: LlmClientOptions = {},\n): Promise<LlmCallResult> {\n  try {\n    return await callLlm({ ...req, jsonMode: req.jsonMode ?? !req.jsonSchema }, opts)\n  } catch (err) {\n    if (\n      opts.jsonSchemaTransport !== 'json-object' &&\n      err instanceof LlmCallError &&\n      isSchemaRejection(err.status, err.body) &&\n      req.jsonSchema\n    ) {\n      const degradedReq: LlmCallRequest = { ...req, jsonMode: true, jsonSchema: undefined }\n      return await callLlm(degradedReq, opts)\n    }\n    throw err\n  }\n}\n\nfunction parseJsonResult<T>(\n  result: LlmCallResult,\n  jsonPayloadMode: NonNullable<LlmClientOptions['jsonPayloadMode']>,\n): T {\n  try {\n    if (result.finishReason === 'length') {\n      throw new Error(\n        `LLM returned truncated JSON content (model=${result.model}, finishReason=length)`,\n      )\n    }\n    return parseJsonSafely<T>(result.content, result.model, jsonPayloadMode)\n  } catch (error) {\n    if (error instanceof LlmResponseError) throw error\n    const cause = error instanceof Error ? error : new Error(String(error))\n    throw new LlmResponseError(cause.message, result, { cause })\n  }\n}\n\nfunction parseJsonSafely<T>(\n  content: string,\n  model: string,\n  jsonPayloadMode: NonNullable<LlmClientOptions['jsonPayloadMode']>,\n): T {\n  const payload = jsonPayloadMode === 'exact' ? content : extractJsonPayload(content)\n  try {\n    return JSON.parse(payload) as T\n  } catch (err) {\n    throw new Error(\n      `LLM returned non-JSON content (model=${model}): ${\n        err instanceof Error ? err.message : String(err)\n      }\\n--- raw content ---\\n${content.slice(0, 800)}`,\n    )\n  }\n}\n\n// ─── Route assertion ────────────────────────────────────────────────────\n\nexport type LlmRouteAssertionReason =\n  | 'no_explicit_base_url'\n  | 'base_url_blocked'\n  | 'base_url_not_allowed'\n  | 'no_auth'\n  | 'wrong_provider'\n\nexport class LlmRouteAssertionError extends CaptureIntegrityError {\n  constructor(\n    message: string,\n    public readonly reason: LlmRouteAssertionReason,\n    public readonly baseUrl: string,\n  ) {\n    super(message)\n  }\n}\n\nexport interface LlmRouteRequirements {\n  /**\n   * Throw if `opts.baseUrl` is undefined, i.e. the call would fall back to\n   * `DEFAULT_BASE_URL`. Set this for evaluation runs where silently using\n   * the public/free-tier router is a defect — the launch reviewer needs to\n   * know exactly which provider answered.\n   */\n  requireExplicitBaseUrl?: boolean\n  /**\n   * Allowlist of acceptable base URLs. Strings match by prefix\n   * (case-insensitive); RegExps test against the full base URL.\n   */\n  allowedBaseUrls?: Array<string | RegExp>\n  /** Blocklist that takes precedence over `allowedBaseUrls`. */\n  blockedBaseUrls?: Array<string | RegExp>\n  /** Throw if no auth header / api key is configured. */\n  requireAuth?: boolean\n  /**\n   * Logical provider id the configured `baseUrl` is expected to match (via\n   * `providerFromBaseUrl`). Mainly useful when paired with `requireExplicitBaseUrl`.\n   */\n  expectedProvider?: string\n}\n\n/**\n * Fail-loud assertion that the configured LLM client points at the route\n * the caller intends. Designed for the matrix-runner preflight: invoke\n * once before any LLM call to catch misconfiguration before a sweep burns\n * dollars on the wrong provider.\n *\n * Throws `LlmRouteAssertionError`. Pure — no I/O — so it's safe to call\n * from constructors and CI gates.\n */\nexport function assertLlmRoute(opts: LlmClientOptions, req: LlmRouteRequirements = {}): void {\n  const baseUrlExplicit = opts.baseUrl !== undefined\n  const baseUrl = (opts.baseUrl ?? DEFAULT_BASE_URL).replace(/\\/+$/, '')\n\n  if (req.requireExplicitBaseUrl && !baseUrlExplicit) {\n    throw new LlmRouteAssertionError(\n      `assertLlmRoute: requireExplicitBaseUrl set but opts.baseUrl is undefined; would fall back to ${DEFAULT_BASE_URL}.`,\n      'no_explicit_base_url',\n      baseUrl,\n    )\n  }\n\n  if (req.blockedBaseUrls?.some((p) => matchUrl(baseUrl, p))) {\n    throw new LlmRouteAssertionError(\n      `assertLlmRoute: baseUrl ${baseUrl} matches a blocked pattern.`,\n      'base_url_blocked',\n      baseUrl,\n    )\n  }\n\n  if (req.allowedBaseUrls && req.allowedBaseUrls.length > 0) {\n    const ok = req.allowedBaseUrls.some((p) => matchUrl(baseUrl, p))\n    if (!ok) {\n      throw new LlmRouteAssertionError(\n        `assertLlmRoute: baseUrl ${baseUrl} is not in the allowed list (${req.allowedBaseUrls.map(describePattern).join(', ')}).`,\n        'base_url_not_allowed',\n        baseUrl,\n      )\n    }\n  }\n\n  if (req.requireAuth && !opts.apiKey && !opts.bearer && !opts.authHeader) {\n    throw new LlmRouteAssertionError(\n      `assertLlmRoute: requireAuth set but no apiKey, bearer, or authHeader was supplied.`,\n      'no_auth',\n      baseUrl,\n    )\n  }\n\n  if (req.expectedProvider) {\n    const actual = opts.provider ?? providerFromBaseUrl(baseUrl)\n    if (actual !== req.expectedProvider) {\n      throw new LlmRouteAssertionError(\n        `assertLlmRoute: expected provider ${req.expectedProvider} but baseUrl ${baseUrl} resolves to ${actual}.`,\n        'wrong_provider',\n        baseUrl,\n      )\n    }\n  }\n}\n\nfunction matchUrl(url: string, pattern: string | RegExp): boolean {\n  if (pattern instanceof RegExp) return pattern.test(url)\n  return url.toLowerCase().startsWith(pattern.toLowerCase())\n}\n\nfunction describePattern(p: string | RegExp): string {\n  return p instanceof RegExp ? p.source : p\n}\n\n/**\n * Probe whether a model is reachable. Returns latency + null error on\n * success; `ok=false` + error message on any failure (HTTP, timeout,\n * network, parse). Designed for sweep preflights — fail loud at the\n * boundary before burning a 30-leaf run on a misconfigured router.\n *\n * Sends a tiny `ping` message with `maxTokens=64`. Reasoning models\n * (glm-5.1, deepseek-v4) can burn the entire budget on internal reasoning\n * for short prompts, so don't tighten this further. We don't validate\n * content; HTTP 200 means reachable.\n */\nexport async function probeLlm(\n  model: string,\n  opts: LlmClientOptions & { timeoutMs?: number } = {},\n): Promise<{ ok: boolean; latencyMs: number; error: string | null }> {\n  const start = Date.now()\n  try {\n    await callLlm(\n      {\n        model,\n        messages: [{ role: 'user', content: 'ping' }],\n        maxTokens: 64,\n        timeoutMs: opts.timeoutMs ?? 30_000,\n      },\n      opts,\n    )\n    return { ok: true, latencyMs: Date.now() - start, error: null }\n  } catch (err) {\n    return {\n      ok: false,\n      latencyMs: Date.now() - start,\n      error: err instanceof Error ? err.message : String(err),\n    }\n  }\n}\n\n/**\n * Stateful client — construct once with defaults, call many times.\n * Thin wrapper around the free functions; exists for callers that want\n * to inject a single configured instance into multiple primitives.\n */\nexport class LlmClient {\n  readonly maximumAttempts: number\n  private readonly opts: LlmClientOptions\n\n  constructor(opts: LlmClientOptions = {}) {\n    this.opts = opts\n    this.maximumAttempts = resolveMaximumAttempts(opts.maximumAttempts)\n  }\n\n  call(req: LlmCallRequest, per?: LlmClientOptions): Promise<LlmCallResult> {\n    const options = { ...this.opts, ...per }\n    return req.jsonSchema ? callLlmStructured(req, options) : callLlm(req, options)\n  }\n\n  callJson<T = unknown>(\n    req: LlmCallRequest,\n    per?: LlmClientOptions,\n  ): Promise<{ value: T; result: LlmCallResult }> {\n    return callLlmJson<T>(req, { ...this.opts, ...per })\n  }\n}\n"],"mappings":";;;;;;;;;;;;;;;;;;;;;;;;;;;;;AA0EA,SAAgB,2BACd,SACA,UAA4B,CAAC,GACF;CAC3B,IAAI,QAAQ,cAAc,KAAA,GAAW,OAAO,KAAA;CAC5C,IAAI,CAAC,OAAO,UAAU,QAAQ,SAAS,KAAK,QAAQ,aAAa,GAC/D,MAAM,IAAI,WAAW,kEAAkE;CAEzF,IACE,QAAQ,SAAS,MACd,YACC,MAAM,QAAQ,QAAQ,OAAO,KAAK,QAAQ,QAAQ,MAAM,SAAS,KAAK,SAAS,WAAW,CAC9F,GAEA;CAGF,MAAM,WAAW,uBAAuB,QAAQ,eAAe;CAC/D,MAAM,kBAAkB,QAAQ,wBAAwB;CAGxD,MAAM,eAAe,IAAI,YAAY,CAAC,CAAC,OACrC,KAAK,UAAU,UAAU,SAAS,iBAAiB,QAAQ,QAAQ,CAAC,CACtE,CAAC,CAAC;CAEF,MAAM,UAAU,QAAQ,cAAc,CAAC,kBAAkB,IAAI;CAC7D,MAAM,QAAQ;EACZ,aAAa,eAAe,WAAW;EACvC,cAAc,QAAQ,YAAY,WAAW;CAC/C;CACA,OAAO,QAAQ,qBACX;EAAE,oBAAoB,QAAQ;EAAoB,GAAG;CAAM,IAC3D;EAAE,OAAO,QAAQ;EAAO,GAAG;CAAM;AACvC;;AAkDA,SAAgB,mBACd,QACA,oBACkB;CAClB,MAAM,eAAe,OAAO,MAAM,sBAAsB;CACxD,MAAM,cAAc,KAAK,IAAI,GAAG,OAAO,MAAM,eAAe,YAAY;CACxE,MAAM,oBACJ,OAAO,YAAY,QAAQ,sBAAsB,OAAO,MAAM,aAAa,QACvE,oBAAoB,oBAAoB;EACtC;EACA,GAAI,eAAe,IAAI,EAAE,aAAa,IAAI,CAAC;EAC3C,cAAc,OAAO,MAAM;CAC7B,CAAC,IACD,KAAA;CACN,OAAO;EACL,OAAO,OAAO;EACd;EACA,cAAc,OAAO,MAAM;EAC3B,iBAAiB,OAAO,MAAM;EAC9B,cAAc,eAAe,IAAI,eAAe,KAAA;EAChD,eAAe,OAAO,WAAW;EACjC,cAAc,OAAO,MAAM,aAAa;CAC1C;AACF;;AAGA,SAAgB,wBACd,OACA,oBAC8B;CAC9B,OAAO,iBAAiB,mBACpB,mBAAmB,MAAM,QAAQ,kBAAkB,IACnD,KAAA;AACN;AAEA,IAAa,eAAb,cAAkC,eAAe;CAG7B;CACA;CACA;CAJlB,YACE,SACA,QACA,MACA,OACA;EACA,MAAM,SAAS,OAAO;EAJN,KAAA,SAAA;EACA,KAAA,OAAA;EACA,KAAA,QAAA;CAGlB;AACF;;;;AAKA,IAAa,mBAAb,cAAsC,eAAe;CAGjC;CAFlB,YACE,SACA,QACA,SACA;EACA,MAAM,SAAS,SAAS,OAAO;EAHf,KAAA,SAAA;CAIlB;AACF;AAuEA,MAAM,mBAAmB;AASzB,MAAM,qBAAqB,OAAO,QAAQ,IAAI,qBAAqB,KAAK;AACxE,MAAM,2BACJ,QAAQ,IAAI,gCAAgC,KAAA,IACxC,IACA,OAAO,QAAQ,IAAI,2BAA2B;AAEpD,SAAS,uBAAuB,YAAwC;CACtE,MAAM,WAAW,cAAc;CAC/B,IAAI,CAAC,OAAO,UAAU,QAAQ,KAAK,YAAY,GAC7C,MAAM,IAAI,WAAW,iDAAiD;CAExE,OAAO;AACT;AAEA,SAAS,mBAAmB,OAAoC;CAC9D,OAAO,OAAO,UAAU,YAAY,OAAO,cAAc,KAAK,KAAK,SAAS,IAAI,QAAQ,KAAA;AAC1F;AAEA,MAAM,mCAAmB,IAAI,IAAI;CAAC;CAAK;CAAK;CAAK;AAAG,CAAC;;;;;;;;;;AAWrD,MAAM,2BAA8C;CAClD;CACA;CACA;CACA;CACA;CACA;CACA;CACA;CACA;CACA;CACA;CACA;CACA;AACF;;;;;;;;;;;;AAaA,SAAgB,oBAAoB,KAAuB;CACzD,OAAO,kBAAkB,KAAK,CAAC;AACjC;AAEA,SAAS,kBAAkB,KAAc,OAAwB;CAC/D,IAAI,eAAe,cAAc,OAAO,iBAAiB,IAAI,IAAI,MAAM;CACvE,IAAI,EAAE,eAAe,QAAQ,OAAO;CAGpC,MAAM,SAAU,IAA6B;CAC7C,IAAI,OAAO,WAAW,YAAY,iBAAiB,IAAI,MAAM,GAAG,OAAO;CACvE,MAAM,OAAQ,IAA2B;CACzC,MAAM,WAAW,GAAG,IAAI,KAAK,IAAI,IAAI,QAAQ,IAAI,OAAO,SAAS,WAAW,OAAO;CACnF,IAAI,yBAAyB,MAAM,MAAM,EAAE,KAAK,QAAQ,CAAC,GAAG,OAAO;CACnE,MAAM,QAAS,IAA4B;CAC3C,IAAI,QAAQ,KAAK,iBAAiB,SAAS,UAAU,KACnD,OAAO,kBAAkB,OAAO,QAAQ,CAAC;CAE3C,OAAO;AACT;AAEA,SAAS,gBAAgB,SAAiC;CACxD,MAAM,IAAI,QAAQ,IAAI,aAAa;CACnC,IAAI,CAAC,GAAG,OAAO;CACf,MAAM,WAAW,OAAO,CAAC;CACzB,IAAI,OAAO,SAAS,QAAQ,KAAK,WAAW,GAAG,OAAO,WAAW;CACjE,MAAM,SAAS,KAAK,MAAM,CAAC;CAC3B,IAAI,OAAO,SAAS,MAAM,GAAG,OAAO,KAAK,IAAI,GAAG,SAAS,KAAK,IAAI,CAAC;CACnE,OAAO;AACT;;AAGA,SAAgB,UAAU,SAAyB;CACjD,OAAO,KAAK,IAAI,MAAM,KAAK,SAAS,IAAM;AAC5C;AAEA,SAAS,aAAa,MAAgD;CACpE,MAAM,UAAkC;EACtC,gBAAgB;EAChB,QAAQ;CACV;CACA,IAAI,KAAK,YACP,QAAQ,KAAK,WAAW,QAAQ,KAAK,WAAW;MAC3C,IAAI,KAAK,UAAU,KAAK,QAC7B,QAAQ,gBAAgB,UAAU,KAAK,UAAU,KAAK;CAExD,IAAI,KAAK,gBAAgB,QAAQ,qBAAqB,KAAK;CAC3D,OAAO;AACT;AAEA,SAAS,kBAAkB,QAAgB,MAAuB;CAChE,IAAI,WAAW,KAAK,OAAO;CAC3B,MAAM,QAAQ,KAAK,YAAY;CAC/B,OACE,MAAM,SAAS,iBAAiB,KAChC,MAAM,SAAS,aAAa,KAC5B,MAAM,SAAS,gBAAgB,KAC/B,MAAM,SAAS,eAAe;AAElC;AAEA,SAAS,0BAA0B,QAAgB,MAAuB;CACxE,IAAI,WAAW,OAAO,CAAC,eAAe,KAAK,IAAI,GAAG,OAAO;CACzD,OACE,iGAAiG,KAC/F,IACF,KAAK,oEAAoE,KAAK,IAAI;AAEtF;AAEA,SAAS,UACP,KACA,iBACA,iBACyB;CACzB,MAAM,OAAgC;EACpC,OAAO,IAAI;EACX,UAAU,IAAI;EACd,aAAa,IAAI,eAAe;CAClC;CACA,IAAI,IAAI,aAAa,MACnB,IAAI,wBAAwB,IAAI,KAAK,GAAG,KAAK,wBAAwB,IAAI;MACpE,KAAK,aAAa,IAAI;CAE7B,MAAM,WAAW,IAAI,YAAY;CACjC,IAAI,aAAa,KAAA,GACf,KAAK,WAAW,EAAE,MAAM,SAAS;CAGnC,IAAI,IAAI,cAAc,CAAC,iBACrB,KAAK,kBAAkB;EACrB,MAAM;EACN,aAAa;GAAE,MAAM,IAAI,WAAW;GAAM,QAAQ,IAAI,WAAW;GAAQ,QAAQ;EAAK;CACxF;MACK,IAAI,IAAI,YAAY,IAAI,YAC7B,KAAK,kBAAkB,EAAE,MAAM,cAAc;CAG/C,OAAO;AACT;AAEA,SAAS,wBAAwB,OAAwB;CACvD,OAAO,oBAAoB,KAAK,KAAK;AACvC;AAEA,eAAe,MAAM,IAA2B;CAC9C,OAAO,IAAI,SAAS,YAAY,WAAW,SAAS,EAAE,CAAC;AACzD;;;;;;;;AASA,SAAS,YAAY,mBAAoC,QAAmC;CAC1F,IAAI,CAAC,QAAQ,OAAO,kBAAkB;CACtC,IAAI,OAAQ,YAAkC,QAAQ,YACpD,OAAO,YAAY,IAAI,CAAC,kBAAkB,QAAQ,MAAM,CAAC;CAE3D,IAAI,OAAO,SACT,kBAAkB,MAAM;MAExB,OAAO,iBAAiB,eAAe,kBAAkB,MAAM,GAAG,EAAE,MAAM,KAAK,CAAC;CAElF,OAAO,kBAAkB;AAC3B;;AAGA,SAAS,iBAAiB,OAAe,YAAyC;CAChF,OAAO,cAAc,QAAQ,KAAK,IAAI,IAAI,SAAS;AACrD;;;;;;AASA,SAAgB,gBAAgB,KAAqB;CACnD,MAAM,UAAU,IAAI,KAAK;CACzB,MAAM,IAAI,QAAQ,MAAM,yCAAyC;CACjE,OAAO,IAAI,EAAE,EAAE,CAAE,KAAK,IAAI;AAC5B;AAEA,SAAgB,mBAAmB,KAAqB;CACtD,MAAM,WAAW,gBAAgB,GAAG;CACpC,IAAI;EACF,KAAK,MAAM,QAAQ;EACnB,OAAO;CACT,QAAQ;EAIN,IAAI,SAAS,WAAW,GAAG,KAAK,SAAS,WAAW,GAAG,GAAG,OAAO;CACnE;CAGA,MAAM,SAAS,CAAC,GAAG,SAAS,SAAS,OAAO,CAAC,CAAC,CAC3C,KAAK,UAAU,MAAM,KAAK,CAAC,CAC3B,QAAQ,UAAU,SAAS,IAAI;CAClC,KAAK,MAAM,SAAS,QAAQ;EAC1B,MAAM,YAAY,oBAAoB,UAAU,KAAK;EACrD,IAAI,CAAC,WAAW;EAChB,IAAI;GACF,KAAK,MAAM,SAAS;GACpB,OAAO;EACT,QAAQ,CAER;CACF;CAEA,OAAO;AACT;AAEA,SAAS,oBAAoB,OAAe,OAA8B;CACxE,MAAM,SAAS,MAAM;CACrB,MAAM,SAAS,WAAW,MAAM,MAAM,WAAW,MAAM,MAAM;CAC7D,IAAI,CAAC,QAAQ,OAAO;CAEpB,MAAM,QAAkB,CAAC,MAAM;CAC/B,IAAI,aAAa;CACjB,IAAI,YAAY;CAEhB,KAAK,IAAI,IAAI,QAAQ,GAAG,IAAI,MAAM,QAAQ,KAAK;EAC7C,MAAM,OAAO,MAAM;EACnB,IAAI,WAAW;GACb,YAAY;GACZ;EACF;EACA,IAAI,SAAS,MAAM;GACjB,YAAY;GACZ;EACF;EACA,IAAI,SAAS,MAAK;GAChB,aAAa,CAAC;GACd;EACF;EACA,IAAI,YAAY;EAEhB,IAAI,SAAS,KAAK,MAAM,KAAK,GAAG;OAC3B,IAAI,SAAS,KAAK,MAAM,KAAK,GAAG;OAChC,IAAI,SAAS,MAAM,MAAM,SAAS,IAAI;GACzC,MAAM,IAAI;GACV,IAAI,MAAM,WAAW,GAAG,OAAO,MAAM,MAAM,OAAO,IAAI,CAAC;EACzD;CACF;CAEA,OAAO;AACT;;;;;;AAOA,eAAsB,QACpB,KACA,OAAyB,CAAC,GACF;CACxB,MAAM,WAAW,KAAK,WAAW,iBAAA,CAAkB,QAAQ,QAAQ,EAAE;CACrE,MAAM,MAAM,GAAG,QAAQ;CACvB,MAAM,WAAW;CACjB,MAAM,YAAY,IAAI,aAAa,KAAK,oBAAoB;CAC5D,MAAM,kBAAkB,uBAAuB,KAAK,eAAe;CACnE,MAAM,UAAU,KAAK,SAAS,WAAW;CACzC,MAAM,UAAU,aAAa,IAAI;CACjC,MAAM,WAAW,KAAK,YAAY,oBAAoB,OAAO;CAC7D,MAAM,OAAO,KAAK;CAClB,MAAM,WAAW,KAAK,YAAY;CAClC,MAAM,eAAe,KAAK;CAC1B,MAAM,eAAe,KAAK;CAC1B,MAAM,aAAa,KAAK;CACxB,MAAM,gBAAgB,KAAK,IAAI;CAC/B,IAAI,KAAK,oBACP,oBAAoB,KAAK,oBAAoB;EAAE,aAAa;EAAG,cAAc;CAAE,CAAC;CAGlF,IAAI;CACJ,IAAI,mBAAmB;CACvB,KAAK,IAAI,UAAU,GAAG,UAAU,iBAAiB,WAAW;EAG1D,IAAI,cAAc,SAChB,MAAM,IAAI,aAAa,oCAAoC,YAAY;EAIzE,IAAI,UAAU,KAAK,iBAAiB,eAAe,UAAU,GAC3D,MAAM,mBAAmB,QAAQ,UAAU,IAAI,MAAM,OAAO,OAAO,CAAC;EAEtE,MAAM,aAAa,IAAI,gBAAgB;EACvC,MAAM,gBAAgB,YAAY,YAAY,YAAY;EAC1D,MAAM,gBAAgB,iBAAiB,WAAW,MAAM,GAAG,SAAS;EACpE,MAAM,UAAU,KAAK,IAAI;EACzB,MAAM,cAAc,UAClB,kBACA,KAAK,wBAAwB,eAC7B,KAAK,QACP;EACA,IAAI,uBAAuB;EAC3B,IAAI,MACF,MAAM,UAAU,MAAM,UAAU;GAC9B,SAAS,cAAc;GACvB,OAAO,cAAc;GACrB,QAAQ,cAAc;GACtB;GACA,OAAO,IAAI;GACX;GACA;GACA,cAAc;GACd,WAAW;GACX,WAAW;GACX,gBAAgB;GAChB;GACA,gBAAgB,CAAC;EACnB,CAAC;EAGH,IAAI;GACF,MAAM,MAAM,MAAM,QAAQ,KAAK;IAC7B,QAAQ;IACR;IACA,MAAM,KAAK,UAAU,WAAW;IAChC,QAAQ;GACV,CAAC;GACD,aAAa,aAAa;GAC1B,MAAM,kBAAkB,OAAO,gBAAgB,IAAI,OAAO,IAAI,KAAA;GAE9D,IAAI,CAAC,IAAI,IAAI;IACX,MAAM,OAAO,MAAM,IAAI,KAAK;IAC5B,IAAI,MAAM;KACR,MAAM,UAAU,MAAM,UAAU;MAC9B,SAAS,cAAc;MACvB,OAAO,cAAc;MACrB,QAAQ,cAAc;MACtB;MACA,OAAO,IAAI;MACX;MACA;MACA,cAAc;MACd,WAAW;MACX,WAAW,KAAK,IAAI;MACpB,YAAY,KAAK,IAAI,IAAI;MACzB,YAAY,IAAI;MAChB;MACA,cAAc;MACd,cAAc,QAAQ,IAAI;MAC1B,gBAAgB,CAAC;KACnB,CAAC;KACD,uBAAuB;IACzB;IACA,MAAM,MAAM,IAAI,aACd,YAAY,IAAI,OAAO,IAAI,KAAK,MAAM,GAAG,GAAG,KAC5C,IAAI,QACJ,MACA,IAAI,KACN;IACA,IACE,0BAA0B,IAAI,QAAQ,IAAI,KAC1C,iBAAiB,gBAAgB,KACjC,UAAU,kBAAkB,KAC5B,CAAC,iBAAiB,eAAe,UAAU,GAC3C;KACA,UAAU;KACV,mBAAmB;MAAE,GAAG;MAAkB,aAAa;KAAE;KACzD;IACF;IACA,IACE,iBAAiB,IAAI,IAAI,MAAM,KAC/B,UAAU,kBAAkB,KAC5B,CAAC,iBAAiB,eAAe,UAAU,GAC3C;KACA,UAAU;KAEV,MAAM,MADa,gBAAgB,IAAI,OAClB,KAAK,UAAU,OAAO,CAAC;KAC5C;IACF;IACA,MAAM;GACR;GAEA,MAAM,OAAO,MAAM,IAAI,KAAK;GAC5B,IAAI;GACJ,IAAI;IACF,OAAO,KAAK,MAAM,IAAI;GACxB,SAAS,UAAU;IACjB,IAAI,MAAM;KACR,MAAM,UAAU,MAAM,UAAU;MAC9B,SAAS,cAAc;MACvB,OAAO,cAAc;MACrB,QAAQ,cAAc;MACtB;MACA,OAAO,IAAI;MACX;MACA;MACA,cAAc;MACd,WAAW;MACX,WAAW,KAAK,IAAI;MACpB,YAAY,KAAK,IAAI,IAAI;MACzB,YAAY,IAAI;MAChB;MACA,cAAc;MACd,cAAc,sBAAsB,oBAAoB,QAAQ,SAAS,UAAU,OAAO,QAAQ;MAClG,gBAAgB,CAAC;KACnB,CAAC;KACD,uBAAuB;IACzB;IACA,MAAM;GACR;GACA,IAAI,MACF,MAAM,UAAU,MAAM,UAAU;IAC9B,SAAS,cAAc;IACvB,OAAO,cAAc;IACrB,QAAQ,cAAc;IACtB;IACA,OAAO,IAAI;IACX;IACA;IACA,cAAc;IACd,WAAW;IACX,WAAW,KAAK,IAAI;IACpB,YAAY,KAAK,IAAI,IAAI;IACzB,YAAY,IAAI;IAChB;IACA,cAAc;IACd,gBAAgB,CAAC;GACnB,CAAC;GAEH,MAAM,SACJ,KAAK,UAGH;GACJ,MAAM,WACJ,KAAK,SAAS,OAAO,KAAK,UAAU,YAAY,CAAC,MAAM,QAAQ,KAAK,KAAK,IACpE,KAAK,QACN,KAAA;GACN,MAAM,eAAe,mBAAmB,UAAU,aAAa;GAC/D,MAAM,mBAAmB,mBAAmB,UAAU,iBAAiB;GACvE,MAAM,cAAc,mBAAmB,UAAU,YAAY;GAO7D,MAAM,gBALJ,UAAU,6BACV,OAAO,SAAS,8BAA8B,YAC9C,CAAC,MAAM,QAAQ,SAAS,yBAAyB,IAC5C,SAAS,4BACV,KAAA,EAAA,EACkC;GACxC,MAAM,kBACJ,iBAAiB,KAAA,IAAY,KAAA,IAAY,mBAAmB,YAAY;GAC1E,MAAM,YACJ,UAAU,yBACV,OAAO,SAAS,0BAA0B,YAC1C,CAAC,MAAM,QAAQ,SAAS,qBAAqB,IACxC,SAAS,sBAAkD,gBAC5D,KAAA;GACN,MAAM,qBAAqB,cAAc,KAAA,IAAY,KAAA,IAAY,mBAAmB,SAAS;GAC7F,MAAM,gBACJ,iBAAiB,KAAA,KACjB,qBAAqB,KAAA,MACpB,iBAAiB,KAAA,KACf,oBAAoB,KAAA,KAAa,mBAAmB,sBACtD,cAAc,KAAA,KACZ,uBAAuB,KAAA,KAAa,sBAAsB,kBAC5D,gBAAgB,KAAA,KAAa,gBAAgB,eAAe;GAC/D,MAAM,gBAAiB,KAAK,kBAAkB,KAAK;GACnD,MAAM,UAAU,QAAQ,SAAS,WAAW;GAE5C,MAAM,iBACJ,OAAO,kBAAkB,YAAY,iBAAiB,KAAK,qBACvD,oBAAoB,KAAK,oBAAoB;IAC3C,aAAa,gBAAiB,sBAAsB;IACpD,GAAI,qBAAqB,EAAE,cAAc,mBAAmB,IAAI,CAAC;IACjE,cAAc;GAChB,CAAC,IACD,KAAA;GAEN,OAAO;IACL;IACA,cAAc,QAAQ,iBAAiB;IACvC,cAAc,QAAQ,KAAK,CAAC,CAAC,WAAW;IACxC,OAAO;KACL,cAAc,gBAAgB;KAC9B,kBAAkB,oBAAoB;KACtC,aAAa,gBAAgB,gBAAgB,MAAM,oBAAoB;KACvE,UAAU;KACV;KACA;IACF;IACA,SAAS,OAAO,kBAAkB,WAAW,gBAAiB,kBAAkB;IAChF,OAAQ,KAAK,SAAoB,IAAI;IACrC,YAAY,KAAK,IAAI,IAAI;IACzB,KAAK;GACP;EACF,SAAS,KAAK;GACZ,aAAa,aAAa;GAC1B,UAAU;GAIV,IAAI,cAAc,SAAS;IACzB,IAAI,QAAQ,CAAC,sBACX,MAAM,UAAU,MAAM,UAAU;KAC9B,SAAS,cAAc;KACvB,OAAO,cAAc;KACrB,QAAQ,cAAc;KACtB;KACA,OAAO,IAAI;KACX;KACA;KACA,cAAc;KACd,WAAW;KACX,WAAW,KAAK,IAAI;KACpB,YAAY,KAAK,IAAI,IAAI;KACzB,cAAc,eAAe,QAAQ,IAAI,UAAU,OAAO,GAAG;KAC7D,gBAAgB,CAAC;IACnB,CAAC;IAEH,MAAM;GACR;GACA,IAAI,QAAQ,CAAC,sBAIX,MAAM,UAAU,MAAM,UAAU;IAC9B,SAAS,cAAc;IACvB,OAAO,cAAc;IACrB,QAAQ,cAAc;IACtB;IACA,OAAO,IAAI;IACX;IACA;IACA,cAAc;IACd,WAAW;IACX,WAAW,KAAK,IAAI;IACpB,YAAY,KAAK,IAAI,IAAI;IACzB,cAAc,eAAe,QAAQ,IAAI,UAAU,OAAO,GAAG;IAC7D,gBAAgB,CAAC;GACnB,CAAC;GAEH,IACE,UAAU,kBAAkB,KAC5B,oBAAoB,GAAG,KACvB,CAAC,iBAAiB,eAAe,UAAU,GAC3C;IACA,MAAM,MAAM,UAAU,OAAO,CAAC;IAC9B;GACF;GACA,MAAM;EACR;CACF;CACA,MAAM,mBAAmB,QAAQ,UAAU,IAAI,MAAM,OAAO,OAAO,CAAC;AACtE;AAEA,eAAe,UACb,MACA,UACA,OACe;CAGf,IAAI;EACF,MAAM,KAAK,OAAO,SAAS,KAAK,CAAC;CACnC,QAAQ,CAER;AACF;AAEA,SAAS,gBAAgB,GAAoC;CAC3D,MAAM,MAA8B,CAAC;CACrC,EAAE,SAAS,OAAO,QAAQ;EACxB,IAAI,OAAO;CACb,CAAC;CACD,OAAO;AACT;AAEA,SAAS,gBAAwB;CAC/B,IAAI,OAAO,WAAW,QAAQ,eAAe,YAAY,OAAO,WAAW,OAAO,WAAW;CAC7F,OAAO,GAAG,KAAK,IAAI,CAAC,CAAC,SAAS,EAAE,EAAE,GAAG,KAAK,OAAO,CAAC,CAAC,SAAS,EAAE,CAAC,CAAC,MAAM,GAAG,EAAE;AAC7E;;;;;;;AAQA,eAAsB,YACpB,KACA,OAAyB,CAAC,GACoB;CAC9C,MAAM,SAAS,MAAM,kBAAkB,KAAK,IAAI;CAEhD,OAAO;EAAE,OADK,gBAAmB,QAAQ,KAAK,mBAAmB,SACpD;EAAG;CAAO;AACzB;;AAGA,eAAe,kBACb,KACA,OAAyB,CAAC,GACF;CACxB,IAAI;EACF,OAAO,MAAM,QAAQ;GAAE,GAAG;GAAK,UAAU,IAAI,YAAY,CAAC,IAAI;EAAW,GAAG,IAAI;CAClF,SAAS,KAAK;EACZ,IACE,KAAK,wBAAwB,iBAC7B,eAAe,gBACf,kBAAkB,IAAI,QAAQ,IAAI,IAAI,KACtC,IAAI,YAGJ,OAAO,MAAM,QAAQ;GADiB,GAAG;GAAK,UAAU;GAAM,YAAY,KAAA;EAC3C,GAAG,IAAI;EAExC,MAAM;CACR;AACF;AAEA,SAAS,gBACP,QACA,iBACG;CACH,IAAI;EACF,IAAI,OAAO,iBAAiB,UAC1B,MAAM,IAAI,MACR,8CAA8C,OAAO,MAAM,uBAC7D;EAEF,OAAO,gBAAmB,OAAO,SAAS,OAAO,OAAO,eAAe;CACzE,SAAS,OAAO;EACd,IAAI,iBAAiB,kBAAkB,MAAM;EAC7C,MAAM,QAAQ,iBAAiB,QAAQ,QAAQ,IAAI,MAAM,OAAO,KAAK,CAAC;EACtE,MAAM,IAAI,iBAAiB,MAAM,SAAS,QAAQ,EAAE,MAAM,CAAC;CAC7D;AACF;AAEA,SAAS,gBACP,SACA,OACA,iBACG;CACH,MAAM,UAAU,oBAAoB,UAAU,UAAU,mBAAmB,OAAO;CAClF,IAAI;EACF,OAAO,KAAK,MAAM,OAAO;CAC3B,SAAS,KAAK;EACZ,MAAM,IAAI,MACR,wCAAwC,MAAM,KAC5C,eAAe,QAAQ,IAAI,UAAU,OAAO,GAAG,EAChD,yBAAyB,QAAQ,MAAM,GAAG,GAAG,GAChD;CACF;AACF;AAWA,IAAa,yBAAb,cAA4C,sBAAsB;CAG9C;CACA;CAHlB,YACE,SACA,QACA,SACA;EACA,MAAM,OAAO;EAHG,KAAA,SAAA;EACA,KAAA,UAAA;CAGlB;AACF;;;;;;;;;;AAmCA,SAAgB,eAAe,MAAwB,MAA4B,CAAC,GAAS;CAC3F,MAAM,kBAAkB,KAAK,YAAY,KAAA;CACzC,MAAM,WAAW,KAAK,WAAW,iBAAA,CAAkB,QAAQ,QAAQ,EAAE;CAErE,IAAI,IAAI,0BAA0B,CAAC,iBACjC,MAAM,IAAI,uBACR,gGAAgG,iBAAiB,IACjH,wBACA,OACF;CAGF,IAAI,IAAI,iBAAiB,MAAM,MAAM,SAAS,SAAS,CAAC,CAAC,GACvD,MAAM,IAAI,uBACR,2BAA2B,QAAQ,8BACnC,oBACA,OACF;CAGF,IAAI,IAAI,mBAAmB,IAAI,gBAAgB,SAAS,GAElD;MAAA,CADO,IAAI,gBAAgB,MAAM,MAAM,SAAS,SAAS,CAAC,CACxD,GACJ,MAAM,IAAI,uBACR,2BAA2B,QAAQ,+BAA+B,IAAI,gBAAgB,IAAI,eAAe,CAAC,CAAC,KAAK,IAAI,EAAE,KACtH,wBACA,OACF;CAAA;CAIJ,IAAI,IAAI,eAAe,CAAC,KAAK,UAAU,CAAC,KAAK,UAAU,CAAC,KAAK,YAC3D,MAAM,IAAI,uBACR,sFACA,WACA,OACF;CAGF,IAAI,IAAI,kBAAkB;EACxB,MAAM,SAAS,KAAK,YAAY,oBAAoB,OAAO;EAC3D,IAAI,WAAW,IAAI,kBACjB,MAAM,IAAI,uBACR,qCAAqC,IAAI,iBAAiB,eAAe,QAAQ,eAAe,OAAO,IACvG,kBACA,OACF;CAEJ;AACF;AAEA,SAAS,SAAS,KAAa,SAAmC;CAChE,IAAI,mBAAmB,QAAQ,OAAO,QAAQ,KAAK,GAAG;CACtD,OAAO,IAAI,YAAY,CAAC,CAAC,WAAW,QAAQ,YAAY,CAAC;AAC3D;AAEA,SAAS,gBAAgB,GAA4B;CACnD,OAAO,aAAa,SAAS,EAAE,SAAS;AAC1C;;;;;;;;;;;;AAaA,eAAsB,SACpB,OACA,OAAkD,CAAC,GACgB;CACnE,MAAM,QAAQ,KAAK,IAAI;CACvB,IAAI;EACF,MAAM,QACJ;GACE;GACA,UAAU,CAAC;IAAE,MAAM;IAAQ,SAAS;GAAO,CAAC;GAC5C,WAAW;GACX,WAAW,KAAK,aAAa;EAC/B,GACA,IACF;EACA,OAAO;GAAE,IAAI;GAAM,WAAW,KAAK,IAAI,IAAI;GAAO,OAAO;EAAK;CAChE,SAAS,KAAK;EACZ,OAAO;GACL,IAAI;GACJ,WAAW,KAAK,IAAI,IAAI;GACxB,OAAO,eAAe,QAAQ,IAAI,UAAU,OAAO,GAAG;EACxD;CACF;AACF;;;;;;AAOA,IAAa,YAAb,MAAuB;CACrB;CACA;CAEA,YAAY,OAAyB,CAAC,GAAG;EACvC,KAAK,OAAO;EACZ,KAAK,kBAAkB,uBAAuB,KAAK,eAAe;CACpE;CAEA,KAAK,KAAqB,KAAgD;EACxE,MAAM,UAAU;GAAE,GAAG,KAAK;GAAM,GAAG;EAAI;EACvC,OAAO,IAAI,aAAa,kBAAkB,KAAK,OAAO,IAAI,QAAQ,KAAK,OAAO;CAChF;CAEA,SACE,KACA,KAC8C;EAC9C,OAAO,YAAe,KAAK;GAAE,GAAG,KAAK;GAAM,GAAG;EAAI,CAAC;CACrD;AACF"}

package/dist/openapi.json

@@ -2,7 +2,7 @@
   "openapi": "3.1.0",
   "info": {
     "title": "@tangle-network/agent-eval — wire protocol",
-    "version": "0.135.0",
+    "version": "0.135.1",
     "description": "HTTP and stdio RPC interface to agent-eval. The TypeScript runtime is the source of truth; this spec is the contract that cross-language clients (Python, Rust, Go) generate from.\n\nWire-protocol version: 1.0.0. Bumps on breaking changes to request/response schemas.",
     "contact": {
       "name": "Tangle Network",

package/dist/rl.js

@@ -7,7 +7,7 @@ import { t as isSplitEligible } from "./exporters-q9iL-2Jf.js";
 import { t as mintRolloutRows } from "./mint-DyRUc9k6.js";
 import { a as InMemoryTraceStore } from "./integrity-BzRbCHzi.js";
 import { c as campaignCellToRunRecord, i as filterDeterministicallyRewarded, n as extractVerifiableReward, r as extractVerifiableRewardsFromRecords, t as detectRewardHacking } from "./reward-hacking-DCdRK9TY.js";
-import { t as runEvalCampaign } from "./eval-campaign-CHFxPTVl.js";
+import { t as runEvalCampaign } from "./eval-campaign-aMG73hvX.js";
 import { t as evaluateInterimReleaseConfidence } from "./sequential-Br0mAPHA.js";
 import { t as rubricPredictiveValidity } from "./rubric-predictive-validity-D6Q6n9oq.js";
 import { n as thompsonCurriculum, r as varianceBasedCurriculum, t as observationsFromRunRecords } from "./active-curriculum-C4mk67HP.js";

package/dist/{semantic-concept-judge-Ca--u10C.js → semantic-concept-judge-Btozx3Vc.js}

@@ -1,6 +1,6 @@
 import { a as NotFoundError } from "./errors-8YnH8WlF.js";
-import { i as CostLedger } from "./cost-ledger-ZAa_P4r0.js";
-import { c as callLlmJson, f as maximumChargeForLlmRequest, l as costReceiptFromLlm, u as costReceiptFromLlmError } from "./llm-client-BNcP4v08.js";
+import { i as CostLedger } from "./cost-ledger-DHAjwNj7.js";
+import { c as callLlmJson, f as maximumChargeForLlmRequest, l as costReceiptFromLlm, u as costReceiptFromLlmError } from "./llm-client-DHx8pzyJ.js";
 import { a as clamp01, i as aggregateRunScore, s as computeFindingId } from "./proposal-findings-DCawte-y.js";
 import { t as Mutex } from "./concurrency-MUjT7VjM.js";
 import { appendFileSync, existsSync, mkdirSync, readFileSync, readdirSync, statSync } from "node:fs";
@@ -721,4 +721,4 @@ function createSemanticConceptJudge(options = {}) {
 //#endregion
 export { RunCritic as a, buildSkillUsageReport as c, defaultIsMaterial as d, diffFindings as f, runSemanticConceptJudge as i, emitSkillUsageFindings as l, SEMANTIC_CONCEPT_JUDGE_VERSION as n, SKILL_USAGE_ANALYST as o, LockedJsonlAppender as p, createSemanticConceptJudge as r, SkillUsageAnalyst as s, DEFAULT_COMPLEXITY_WEIGHTS as t, FindingsStore as u };
 
-//# sourceMappingURL=semantic-concept-judge-Ca--u10C.js.map
+//# sourceMappingURL=semantic-concept-judge-Btozx3Vc.js.map