@tangle-network/agent-eval 0.128.2 → 0.130.0

package/dist/chunk-VZSRQ272.js.map

@@ -1 +0,0 @@
-{"version":3,"sources":["../src/rl/active-curriculum.ts"],"sourcesContent":["/**\n * Adaptive curriculum / active scenario selection.\n *\n * Fixed scenario sets waste sample budget on cells the policy already\n * passes (no information left) and cells the policy never passes (no\n * gradient available either). Active learning over scenarios fixes this\n * by allocating the next sample budget to cells where the policy's\n * outcome is *uncertain* — those carry the most decision-relevant signal.\n *\n * This module ships two complementary strategies:\n *\n *   1. **Variance-based** — score each (variant, scenario) cell by the\n *      empirical variance of past observations. Allocate next-round budget\n *      proportional to variance. Standard active-learning-by-uncertainty\n *      heuristic; works well when the policy is non-deterministic and\n *      cells differ in observation noise.\n *\n *   2. **Bandit-based (Thompson sampling)** — model each (variant,\n *      scenario) cell as a Beta-Bernoulli arm; sample a posterior; pick\n *      cells whose posterior mean is closest to the per-scenario decision\n *      threshold. The right primitive when scenarios are\n *      \"pass/fail\" rather than continuous, and when promotion gates fire\n *      at a known threshold (e.g., 0.5).\n *\n * The output is a *next-round budget allocation* — a list of (variant,\n * scenario, count) triples. The consumer's matrix runner consumes the\n * allocation, runs those cells, feeds the new observations back. Loop.\n *\n * Out of scope (deliberate): scenario *generation* — that's the\n * adversarial primitive's job. This module allocates over an existing\n * scenario pool.\n */\n\nimport type { RunRecord } from '../run-record'\n\nexport interface CellObservation {\n  variantId: string\n  scenarioId: string\n  /** Observed score in [0, 1]. */\n  score: number\n  /** For Bernoulli arms — derive from the score with a threshold if needed. */\n  pass?: boolean\n}\n\nexport interface CurriculumAllocation {\n  variantId: string\n  scenarioId: string\n  /** How many additional reps to run on this cell. */\n  count: number\n  /** Strategy-specific reason for the allocation. */\n  reason: string\n}\n\nexport interface VarianceCurriculumOptions {\n  /** Total reps to allocate across all cells. */\n  budget: number\n  /**\n   * Smoothing prior on variance — keeps the allocator from concentrating\n   * on a cell with one observation just because its 1-sample variance is\n   * 0. Default 0.05.\n   */\n  variancePrior?: number\n  /**\n   * Minimum reps per cell — even when the variance estimate is low, give\n   * every cell at least this many. Default 1.\n   */\n  floorPerCell?: number\n}\n\n/**\n * Variance-proportional allocation. For each cell, estimate variance from\n * past observations + a prior, then allocate the budget proportional to\n * (sqrt(variance) + 1/sqrt(n)) — a classical optimal-allocation rule\n * (Neyman 1934) that balances \"explore noisy cells\" with \"explore\n * under-sampled cells.\"\n */\nexport function varianceBasedCurriculum(\n  observations: CellObservation[],\n  candidateCells: Array<{ variantId: string; scenarioId: string }>,\n  opts: VarianceCurriculumOptions,\n): CurriculumAllocation[] {\n  const variancePrior = opts.variancePrior ?? 0.05\n  const floor = opts.floorPerCell ?? 1\n  const budget = opts.budget\n\n  const grouped = new Map<string, number[]>()\n  for (const o of observations) {\n    const k = `${o.variantId}::${o.scenarioId}`\n    const arr = grouped.get(k) ?? []\n    arr.push(o.score)\n    grouped.set(k, arr)\n  }\n\n  const cellStats = candidateCells.map((c) => {\n    const k = `${c.variantId}::${c.scenarioId}`\n    const samples = grouped.get(k) ?? []\n    const n = samples.length\n    const mean = n === 0 ? 0.5 : samples.reduce((s, v) => s + v, 0) / n\n    const variance =\n      n < 2\n        ? variancePrior\n        : samples.reduce((s, v) => s + (v - mean) ** 2, 0) / (n - 1) + variancePrior\n    // Neyman optimal allocation: weight ∝ √variance; add √(1/n) to break\n    // ties toward under-sampled cells.\n    const weight = Math.sqrt(variance) + 1 / Math.sqrt(Math.max(1, n))\n    return { variantId: c.variantId, scenarioId: c.scenarioId, n, mean, variance, weight }\n  })\n\n  // Reserve floor*N for the floor; allocate the rest proportional to weight.\n  const floorTotal = floor * cellStats.length\n  if (floorTotal >= budget) {\n    const each = Math.max(1, Math.floor(budget / Math.max(1, cellStats.length)))\n    return cellStats.map((c) => ({\n      variantId: c.variantId,\n      scenarioId: c.scenarioId,\n      count: each,\n      reason: `floor allocation (budget tight; n=${c.n})`,\n    }))\n  }\n  const remaining = budget - floorTotal\n  const totalWeight = cellStats.reduce((s, c) => s + c.weight, 0)\n  return cellStats.map((c) => {\n    const proportional = totalWeight === 0 ? 0 : Math.round((c.weight / totalWeight) * remaining)\n    return {\n      variantId: c.variantId,\n      scenarioId: c.scenarioId,\n      count: floor + proportional,\n      reason: `variance ${c.variance.toFixed(3)} (n=${c.n}, mean=${c.mean.toFixed(3)})`,\n    }\n  })\n}\n\nexport interface ThompsonCurriculumOptions {\n  budget: number\n  /**\n   * The per-scenario decision threshold. Cells whose posterior mean is\n   * closest to this get the most budget — that's where the next observation\n   * has the highest information value for the gate decision. Default 0.5.\n   */\n  decisionThreshold?: number\n  /** Beta prior parameters. Default α=β=1 (uniform). */\n  priorAlpha?: number\n  priorBeta?: number\n  /** Seed the Thompson sampler. Default unset (Math.random). */\n  seed?: number\n}\n\n/**\n * Thompson-sampling-style allocation for pass/fail cells. For each cell:\n *\n *   - Maintain Beta(α + passes, β + failures) posterior on pass-rate\n *   - Allocation weight ∝ exp(-((sampledMean - threshold) / σ)^2):\n *     cells whose sampled posterior straddles the decision boundary get\n *     the most weight; cells already clearly above or below get less.\n *\n * This is the right primitive when promotion gates fire at a known\n * threshold and you want to sharpen the posterior near the boundary.\n */\nexport function thompsonCurriculum(\n  observations: CellObservation[],\n  candidateCells: Array<{ variantId: string; scenarioId: string }>,\n  opts: ThompsonCurriculumOptions,\n): CurriculumAllocation[] {\n  const threshold = opts.decisionThreshold ?? 0.5\n  const alpha0 = opts.priorAlpha ?? 1\n  const beta0 = opts.priorBeta ?? 1\n  const rng = makeRng(opts.seed)\n\n  const grouped = new Map<string, { passes: number; failures: number }>()\n  for (const o of observations) {\n    const k = `${o.variantId}::${o.scenarioId}`\n    const cur = grouped.get(k) ?? { passes: 0, failures: 0 }\n    const pass = o.pass ?? o.score >= threshold\n    if (pass) cur.passes += 1\n    else cur.failures += 1\n    grouped.set(k, cur)\n  }\n\n  const stats = candidateCells.map((c) => {\n    const k = `${c.variantId}::${c.scenarioId}`\n    const cur = grouped.get(k) ?? { passes: 0, failures: 0 }\n    const a = alpha0 + cur.passes\n    const b = beta0 + cur.failures\n    // Sample a single Beta draw — the Thompson signal.\n    const sampled = sampleBeta(a, b, rng)\n    const distance = Math.abs(sampled - threshold)\n    // Information-near-threshold weight: closer = higher.\n    // Use Gaussian-shaped kernel with σ tuned to posterior std.\n    const variance = (a * b) / ((a + b) ** 2 * (a + b + 1))\n    const sigma = Math.max(0.05, Math.sqrt(variance))\n    const weight = Math.exp(-((distance / sigma) ** 2))\n    return {\n      variantId: c.variantId,\n      scenarioId: c.scenarioId,\n      n: cur.passes + cur.failures,\n      sampled,\n      sigma,\n      weight,\n      a,\n      b,\n    }\n  })\n\n  const totalWeight = stats.reduce((s, c) => s + c.weight, 0)\n  return stats.map((c) => {\n    const proportional = totalWeight === 0 ? 0 : Math.round((c.weight / totalWeight) * opts.budget)\n    return {\n      variantId: c.variantId,\n      scenarioId: c.scenarioId,\n      count: Math.max(0, proportional),\n      reason: `Beta(${c.a.toFixed(1)},${c.b.toFixed(1)}) sample=${c.sampled.toFixed(3)} (target ${threshold})`,\n    }\n  })\n}\n\n/** Convenience: extract `CellObservation[]` directly from `RunRecord[]`. */\nexport function observationsFromRunRecords(\n  runs: RunRecord[],\n  opts: { passThreshold?: number; useHoldout?: boolean } = {},\n): CellObservation[] {\n  const threshold = opts.passThreshold ?? 0.5\n  const useHoldout = opts.useHoldout ?? true\n  const out: CellObservation[] = []\n  for (const r of runs) {\n    if (!r.scenarioId) continue\n    const score = useHoldout\n      ? (r.outcome.holdoutScore ?? r.outcome.searchScore)\n      : (r.outcome.searchScore ?? r.outcome.holdoutScore)\n    if (typeof score !== 'number' || !Number.isFinite(score)) continue\n    out.push({\n      variantId: r.candidateId,\n      scenarioId: r.scenarioId,\n      score,\n      pass: score >= threshold,\n    })\n  }\n  return out\n}\n\n// ── Helpers ──────────────────────────────────────────────────────────────\n\nfunction makeRng(seed?: number): () => number {\n  if (seed === undefined) return Math.random\n  let s = seed >>> 0\n  return () => {\n    s = (s + 0x6d2b79f5) >>> 0\n    let t = s\n    t = Math.imul(t ^ (t >>> 15), t | 1)\n    t ^= t + Math.imul(t ^ (t >>> 7), t | 61)\n    return ((t ^ (t >>> 14)) >>> 0) / 4294967296\n  }\n}\n\n/**\n * Sample from Beta(α, β) via the Marsaglia–Tsang method using two Gamma\n * variates. Accuracy is good for α, β > 1; we floor the parameters at 1\n * to avoid degenerate cases.\n */\nfunction sampleBeta(alpha: number, beta: number, rng: () => number): number {\n  const a = Math.max(1, alpha)\n  const b = Math.max(1, beta)\n  const x = sampleGamma(a, rng)\n  const y = sampleGamma(b, rng)\n  return x / (x + y)\n}\n\nfunction sampleGamma(shape: number, rng: () => number): number {\n  // Marsaglia–Tsang for shape ≥ 1.\n  const d = shape - 1 / 3\n  const c = 1 / Math.sqrt(9 * d)\n  while (true) {\n    let x: number\n    let v: number\n    do {\n      const u1 = rng() || 1e-12\n      const u2 = rng() || 1e-12\n      // Box-Muller for a normal sample.\n      x = Math.sqrt(-2 * Math.log(u1)) * Math.cos(2 * Math.PI * u2)\n      v = 1 + c * x\n    } while (v <= 0)\n    v = v * v * v\n    const u = rng()\n    if (u < 1 - 0.0331 * x ** 4) return d * v\n    if (Math.log(u) < 0.5 * x * x + d * (1 - v + Math.log(v))) return d * v\n  }\n}\n"],"mappings":";AA4EO,SAAS,wBACd,cACA,gBACA,MACwB;AACxB,QAAM,gBAAgB,KAAK,iBAAiB;AAC5C,QAAM,QAAQ,KAAK,gBAAgB;AACnC,QAAM,SAAS,KAAK;AAEpB,QAAM,UAAU,oBAAI,IAAsB;AAC1C,aAAW,KAAK,cAAc;AAC5B,UAAM,IAAI,GAAG,EAAE,SAAS,KAAK,EAAE,UAAU;AACzC,UAAM,MAAM,QAAQ,IAAI,CAAC,KAAK,CAAC;AAC/B,QAAI,KAAK,EAAE,KAAK;AAChB,YAAQ,IAAI,GAAG,GAAG;AAAA,EACpB;AAEA,QAAM,YAAY,eAAe,IAAI,CAAC,MAAM;AAC1C,UAAM,IAAI,GAAG,EAAE,SAAS,KAAK,EAAE,UAAU;AACzC,UAAM,UAAU,QAAQ,IAAI,CAAC,KAAK,CAAC;AACnC,UAAM,IAAI,QAAQ;AAClB,UAAM,OAAO,MAAM,IAAI,MAAM,QAAQ,OAAO,CAAC,GAAG,MAAM,IAAI,GAAG,CAAC,IAAI;AAClE,UAAM,WACJ,IAAI,IACA,gBACA,QAAQ,OAAO,CAAC,GAAG,MAAM,KAAK,IAAI,SAAS,GAAG,CAAC,KAAK,IAAI,KAAK;AAGnE,UAAM,SAAS,KAAK,KAAK,QAAQ,IAAI,IAAI,KAAK,KAAK,KAAK,IAAI,GAAG,CAAC,CAAC;AACjE,WAAO,EAAE,WAAW,EAAE,WAAW,YAAY,EAAE,YAAY,GAAG,MAAM,UAAU,OAAO;AAAA,EACvF,CAAC;AAGD,QAAM,aAAa,QAAQ,UAAU;AACrC,MAAI,cAAc,QAAQ;AACxB,UAAM,OAAO,KAAK,IAAI,GAAG,KAAK,MAAM,SAAS,KAAK,IAAI,GAAG,UAAU,MAAM,CAAC,CAAC;AAC3E,WAAO,UAAU,IAAI,CAAC,OAAO;AAAA,MAC3B,WAAW,EAAE;AAAA,MACb,YAAY,EAAE;AAAA,MACd,OAAO;AAAA,MACP,QAAQ,qCAAqC,EAAE,CAAC;AAAA,IAClD,EAAE;AAAA,EACJ;AACA,QAAM,YAAY,SAAS;AAC3B,QAAM,cAAc,UAAU,OAAO,CAAC,GAAG,MAAM,IAAI,EAAE,QAAQ,CAAC;AAC9D,SAAO,UAAU,IAAI,CAAC,MAAM;AAC1B,UAAM,eAAe,gBAAgB,IAAI,IAAI,KAAK,MAAO,EAAE,SAAS,cAAe,SAAS;AAC5F,WAAO;AAAA,MACL,WAAW,EAAE;AAAA,MACb,YAAY,EAAE;AAAA,MACd,OAAO,QAAQ;AAAA,MACf,QAAQ,YAAY,EAAE,SAAS,QAAQ,CAAC,CAAC,OAAO,EAAE,CAAC,UAAU,EAAE,KAAK,QAAQ,CAAC,CAAC;AAAA,IAChF;AAAA,EACF,CAAC;AACH;AA4BO,SAAS,mBACd,cACA,gBACA,MACwB;AACxB,QAAM,YAAY,KAAK,qBAAqB;AAC5C,QAAM,SAAS,KAAK,cAAc;AAClC,QAAM,QAAQ,KAAK,aAAa;AAChC,QAAM,MAAM,QAAQ,KAAK,IAAI;AAE7B,QAAM,UAAU,oBAAI,IAAkD;AACtE,aAAW,KAAK,cAAc;AAC5B,UAAM,IAAI,GAAG,EAAE,SAAS,KAAK,EAAE,UAAU;AACzC,UAAM,MAAM,QAAQ,IAAI,CAAC,KAAK,EAAE,QAAQ,GAAG,UAAU,EAAE;AACvD,UAAM,OAAO,EAAE,QAAQ,EAAE,SAAS;AAClC,QAAI,KAAM,KAAI,UAAU;AAAA,QACnB,KAAI,YAAY;AACrB,YAAQ,IAAI,GAAG,GAAG;AAAA,EACpB;AAEA,QAAM,QAAQ,eAAe,IAAI,CAAC,MAAM;AACtC,UAAM,IAAI,GAAG,EAAE,SAAS,KAAK,EAAE,UAAU;AACzC,UAAM,MAAM,QAAQ,IAAI,CAAC,KAAK,EAAE,QAAQ,GAAG,UAAU,EAAE;AACvD,UAAM,IAAI,SAAS,IAAI;AACvB,UAAM,IAAI,QAAQ,IAAI;AAEtB,UAAM,UAAU,WAAW,GAAG,GAAG,GAAG;AACpC,UAAM,WAAW,KAAK,IAAI,UAAU,SAAS;AAG7C,UAAM,WAAY,IAAI,MAAO,IAAI,MAAM,KAAK,IAAI,IAAI;AACpD,UAAM,QAAQ,KAAK,IAAI,MAAM,KAAK,KAAK,QAAQ,CAAC;AAChD,UAAM,SAAS,KAAK,IAAI,GAAG,WAAW,UAAU,EAAE;AAClD,WAAO;AAAA,MACL,WAAW,EAAE;AAAA,MACb,YAAY,EAAE;AAAA,MACd,GAAG,IAAI,SAAS,IAAI;AAAA,MACpB;AAAA,MACA;AAAA,MACA;AAAA,MACA;AAAA,MACA;AAAA,IACF;AAAA,EACF,CAAC;AAED,QAAM,cAAc,MAAM,OAAO,CAAC,GAAG,MAAM,IAAI,EAAE,QAAQ,CAAC;AAC1D,SAAO,MAAM,IAAI,CAAC,MAAM;AACtB,UAAM,eAAe,gBAAgB,IAAI,IAAI,KAAK,MAAO,EAAE,SAAS,cAAe,KAAK,MAAM;AAC9F,WAAO;AAAA,MACL,WAAW,EAAE;AAAA,MACb,YAAY,EAAE;AAAA,MACd,OAAO,KAAK,IAAI,GAAG,YAAY;AAAA,MAC/B,QAAQ,QAAQ,EAAE,EAAE,QAAQ,CAAC,CAAC,IAAI,EAAE,EAAE,QAAQ,CAAC,CAAC,YAAY,EAAE,QAAQ,QAAQ,CAAC,CAAC,YAAY,SAAS;AAAA,IACvG;AAAA,EACF,CAAC;AACH;AAGO,SAAS,2BACd,MACA,OAAyD,CAAC,GACvC;AACnB,QAAM,YAAY,KAAK,iBAAiB;AACxC,QAAM,aAAa,KAAK,cAAc;AACtC,QAAM,MAAyB,CAAC;AAChC,aAAW,KAAK,MAAM;AACpB,QAAI,CAAC,EAAE,WAAY;AACnB,UAAM,QAAQ,aACT,EAAE,QAAQ,gBAAgB,EAAE,QAAQ,cACpC,EAAE,QAAQ,eAAe,EAAE,QAAQ;AACxC,QAAI,OAAO,UAAU,YAAY,CAAC,OAAO,SAAS,KAAK,EAAG;AAC1D,QAAI,KAAK;AAAA,MACP,WAAW,EAAE;AAAA,MACb,YAAY,EAAE;AAAA,MACd;AAAA,MACA,MAAM,SAAS;AAAA,IACjB,CAAC;AAAA,EACH;AACA,SAAO;AACT;AAIA,SAAS,QAAQ,MAA6B;AAC5C,MAAI,SAAS,OAAW,QAAO,KAAK;AACpC,MAAI,IAAI,SAAS;AACjB,SAAO,MAAM;AACX,QAAK,IAAI,eAAgB;AACzB,QAAI,IAAI;AACR,QAAI,KAAK,KAAK,IAAK,MAAM,IAAK,IAAI,CAAC;AACnC,SAAK,IAAI,KAAK,KAAK,IAAK,MAAM,GAAI,IAAI,EAAE;AACxC,aAAS,IAAK,MAAM,QAAS,KAAK;AAAA,EACpC;AACF;AAOA,SAAS,WAAW,OAAe,MAAc,KAA2B;AAC1E,QAAM,IAAI,KAAK,IAAI,GAAG,KAAK;AAC3B,QAAM,IAAI,KAAK,IAAI,GAAG,IAAI;AAC1B,QAAM,IAAI,YAAY,GAAG,GAAG;AAC5B,QAAM,IAAI,YAAY,GAAG,GAAG;AAC5B,SAAO,KAAK,IAAI;AAClB;AAEA,SAAS,YAAY,OAAe,KAA2B;AAE7D,QAAM,IAAI,QAAQ,IAAI;AACtB,QAAM,IAAI,IAAI,KAAK,KAAK,IAAI,CAAC;AAC7B,SAAO,MAAM;AACX,QAAI;AACJ,QAAI;AACJ,OAAG;AACD,YAAM,KAAK,IAAI,KAAK;AACpB,YAAM,KAAK,IAAI,KAAK;AAEpB,UAAI,KAAK,KAAK,KAAK,KAAK,IAAI,EAAE,CAAC,IAAI,KAAK,IAAI,IAAI,KAAK,KAAK,EAAE;AAC5D,UAAI,IAAI,IAAI;AAAA,IACd,SAAS,KAAK;AACd,QAAI,IAAI,IAAI;AACZ,UAAM,IAAI,IAAI;AACd,QAAI,IAAI,IAAI,SAAS,KAAK,EAAG,QAAO,IAAI;AACxC,QAAI,KAAK,IAAI,CAAC,IAAI,MAAM,IAAI,IAAI,KAAK,IAAI,IAAI,KAAK,IAAI,CAAC,GAAI,QAAO,IAAI;AAAA,EACxE;AACF;","names":[]}

package/dist/chunk-WGXIEX7P.js

@@ -1,116 +0,0 @@
-// src/concurrency.ts
-var Mutex = class {
-  locked = false;
-  waiters = [];
-  async acquire() {
-    if (!this.locked) {
-      this.locked = true;
-      return () => this.release();
-    }
-    return new Promise((resolve) => {
-      this.waiters.push(() => {
-        resolve(() => this.release());
-      });
-    });
-  }
-  release() {
-    const next = this.waiters.shift();
-    if (next) {
-      next();
-    } else {
-      this.locked = false;
-    }
-  }
-  async runExclusive(fn) {
-    const release = await this.acquire();
-    try {
-      return await fn();
-    } finally {
-      release();
-    }
-  }
-  /** True iff someone holds the lock right now. Diagnostics only. */
-  get isLocked() {
-    return this.locked;
-  }
-  /** Pending waiter count. Diagnostics only. */
-  get pending() {
-    return this.waiters.length;
-  }
-};
-async function mapConcurrent(items, concurrency, map) {
-  if (!Number.isInteger(concurrency) || concurrency < 1) {
-    throw new Error(`mapConcurrent: concurrency must be a positive integer, got ${concurrency}`);
-  }
-  if (items.length === 0) return [];
-  const results = new Array(items.length);
-  let nextIndex = 0;
-  let stopped = false;
-  let failed = false;
-  let failure;
-  const worker = async () => {
-    while (!stopped) {
-      const index = nextIndex;
-      nextIndex += 1;
-      if (index >= items.length) return;
-      try {
-        results[index] = await map(items[index], index);
-      } catch (error) {
-        stopped = true;
-        if (!failed) {
-          failed = true;
-          failure = error;
-        }
-      }
-    }
-  };
-  await Promise.all(Array.from({ length: Math.min(concurrency, items.length) }, () => worker()));
-  if (failed) throw failure;
-  return results;
-}
-
-// src/run-score.ts
-var DEFAULT_RUN_SCORE_WEIGHTS = {
-  success: 4,
-  goalProgress: 2,
-  repoGroundedness: 1.5,
-  driftPenalty: -1.5,
-  toolUseQuality: 1,
-  patchQuality: 1.25,
-  testReality: 1.5,
-  finalGate: 3,
-  reviewerBlockers: -2,
-  costUsd: -0.2,
-  wallSeconds: -0.1
-};
-function aggregateRunScore(score, weights = {}) {
-  const w = { ...DEFAULT_RUN_SCORE_WEIGHTS, ...weights };
-  return w.success * clamp01(score.success) + w.goalProgress * clamp01(score.goalProgress) + w.repoGroundedness * clamp01(score.repoGroundedness) + w.driftPenalty * clamp01(score.driftPenalty) + w.toolUseQuality * clamp01(score.toolUseQuality) + w.patchQuality * clamp01(score.patchQuality) + w.testReality * clamp01(score.testReality) + w.finalGate * clamp01(score.finalGate) + w.reviewerBlockers * clamp01(score.reviewerBlockers) + w.costUsd * Math.max(0, finiteOrZero(score.costUsd)) + w.wallSeconds * Math.max(0, finiteOrZero(score.wallSeconds) / 60);
-}
-function clamp01(value) {
-  if (!Number.isFinite(value)) return 0;
-  return Math.max(0, Math.min(1, value));
-}
-function finiteOrZero(value) {
-  return Number.isFinite(value) ? value : 0;
-}
-
-// src/abort-signal.ts
-function combineAbortSignals(...signals) {
-  const active = [
-    ...new Set(signals.filter((signal) => signal !== void 0))
-  ];
-  if (active.length === 0) return void 0;
-  if (active.length === 1) return active[0];
-  return AbortSignal.any(active);
-}
-
-export {
-  combineAbortSignals,
-  Mutex,
-  mapConcurrent,
-  DEFAULT_RUN_SCORE_WEIGHTS,
-  aggregateRunScore,
-  clamp01
-};
-//# sourceMappingURL=chunk-WGXIEX7P.js.map

package/dist/chunk-WGXIEX7P.js.map

@@ -1 +0,0 @@
-{"version":3,"sources":["../src/concurrency.ts","../src/run-score.ts","../src/abort-signal.ts"],"sourcesContent":["/**\n * concurrency — small primitives the evolution loop needs.\n *\n * `Mutex` is a zero-dep async lock with FIFO fairness. The evolution loop\n * uses it to serialise checkout/build/commit sequences inside a single\n * pool slot, and to gate concurrent JSONL writers (see\n * `lockedJsonlReferenceReplayStore`).\n *\n * Deliberately minimal — no priority queue, no timeouts. If you need\n * those, swap to `async-mutex` at the call site.\n */\n\nexport class Mutex {\n  private locked = false\n  private readonly waiters: Array<() => void> = []\n\n  async acquire(): Promise<() => void> {\n    if (!this.locked) {\n      this.locked = true\n      return () => this.release()\n    }\n    return new Promise<() => void>((resolve) => {\n      this.waiters.push(() => {\n        resolve(() => this.release())\n      })\n    })\n  }\n\n  private release(): void {\n    const next = this.waiters.shift()\n    if (next) {\n      next()\n    } else {\n      this.locked = false\n    }\n  }\n\n  async runExclusive<T>(fn: () => Promise<T> | T): Promise<T> {\n    const release = await this.acquire()\n    try {\n      return await fn()\n    } finally {\n      release()\n    }\n  }\n\n  /** True iff someone holds the lock right now. Diagnostics only. */\n  get isLocked(): boolean {\n    return this.locked\n  }\n\n  /** Pending waiter count. Diagnostics only. */\n  get pending(): number {\n    return this.waiters.length\n  }\n}\n\n/**\n * Map independent work with a fixed worker count while preserving input order.\n * After the first rejection, no new items start; already-running work is allowed\n * to settle before the returned promise rejects. Partial results are discarded.\n */\nexport async function mapConcurrent<T, R>(\n  items: readonly T[],\n  concurrency: number,\n  map: (item: T, index: number) => Promise<R>,\n): Promise<R[]> {\n  if (!Number.isInteger(concurrency) || concurrency < 1) {\n    throw new Error(`mapConcurrent: concurrency must be a positive integer, got ${concurrency}`)\n  }\n  if (items.length === 0) return []\n\n  const results = new Array<R>(items.length)\n  let nextIndex = 0\n  let stopped = false\n  let failed = false\n  let failure: unknown\n\n  const worker = async (): Promise<void> => {\n    while (!stopped) {\n      const index = nextIndex\n      nextIndex += 1\n      if (index >= items.length) return\n\n      try {\n        results[index] = await map(items[index]!, index)\n      } catch (error) {\n        stopped = true\n        if (!failed) {\n          failed = true\n          failure = error\n        }\n      }\n    }\n  }\n\n  await Promise.all(Array.from({ length: Math.min(concurrency, items.length) }, () => worker()))\n  if (failed) throw failure\n  return results\n}\n","export interface RunScore {\n  success: number\n  goalProgress: number\n  repoGroundedness: number\n  driftPenalty: number\n  toolUseQuality: number\n  patchQuality: number\n  testReality: number\n  finalGate: number\n  reviewerBlockers: number\n  costUsd: number\n  wallSeconds: number\n  notes?: string[]\n}\n\nexport interface RunScoreWeights {\n  success: number\n  goalProgress: number\n  repoGroundedness: number\n  driftPenalty: number\n  toolUseQuality: number\n  patchQuality: number\n  testReality: number\n  finalGate: number\n  reviewerBlockers: number\n  costUsd: number\n  wallSeconds: number\n}\n\nexport const DEFAULT_RUN_SCORE_WEIGHTS: RunScoreWeights = {\n  success: 4,\n  goalProgress: 2,\n  repoGroundedness: 1.5,\n  driftPenalty: -1.5,\n  toolUseQuality: 1,\n  patchQuality: 1.25,\n  testReality: 1.5,\n  finalGate: 3,\n  reviewerBlockers: -2,\n  costUsd: -0.2,\n  wallSeconds: -0.1,\n}\n\nexport function aggregateRunScore(score: RunScore, weights: Partial<RunScoreWeights> = {}): number {\n  const w = { ...DEFAULT_RUN_SCORE_WEIGHTS, ...weights }\n  return (\n    w.success * clamp01(score.success) +\n    w.goalProgress * clamp01(score.goalProgress) +\n    w.repoGroundedness * clamp01(score.repoGroundedness) +\n    w.driftPenalty * clamp01(score.driftPenalty) +\n    w.toolUseQuality * clamp01(score.toolUseQuality) +\n    w.patchQuality * clamp01(score.patchQuality) +\n    w.testReality * clamp01(score.testReality) +\n    w.finalGate * clamp01(score.finalGate) +\n    w.reviewerBlockers * clamp01(score.reviewerBlockers) +\n    w.costUsd * Math.max(0, finiteOrZero(score.costUsd)) +\n    w.wallSeconds * Math.max(0, finiteOrZero(score.wallSeconds) / 60)\n  )\n}\n\nexport function clamp01(value: number): number {\n  if (!Number.isFinite(value)) return 0\n  return Math.max(0, Math.min(1, value))\n}\n\nfunction finiteOrZero(value: number): number {\n  return Number.isFinite(value) ? value : 0\n}\n","/** Combine active cancellation sources without wrapping a single source. */\nexport function combineAbortSignals(\n  ...signals: Array<AbortSignal | undefined>\n): AbortSignal | undefined {\n  const active = [\n    ...new Set(signals.filter((signal): signal is AbortSignal => signal !== undefined)),\n  ]\n  if (active.length === 0) return undefined\n  if (active.length === 1) return active[0]\n  return AbortSignal.any(active)\n}\n"],"mappings":";AAYO,IAAM,QAAN,MAAY;AAAA,EACT,SAAS;AAAA,EACA,UAA6B,CAAC;AAAA,EAE/C,MAAM,UAA+B;AACnC,QAAI,CAAC,KAAK,QAAQ;AAChB,WAAK,SAAS;AACd,aAAO,MAAM,KAAK,QAAQ;AAAA,IAC5B;AACA,WAAO,IAAI,QAAoB,CAAC,YAAY;AAC1C,WAAK,QAAQ,KAAK,MAAM;AACtB,gBAAQ,MAAM,KAAK,QAAQ,CAAC;AAAA,MAC9B,CAAC;AAAA,IACH,CAAC;AAAA,EACH;AAAA,EAEQ,UAAgB;AACtB,UAAM,OAAO,KAAK,QAAQ,MAAM;AAChC,QAAI,MAAM;AACR,WAAK;AAAA,IACP,OAAO;AACL,WAAK,SAAS;AAAA,IAChB;AAAA,EACF;AAAA,EAEA,MAAM,aAAgB,IAAsC;AAC1D,UAAM,UAAU,MAAM,KAAK,QAAQ;AACnC,QAAI;AACF,aAAO,MAAM,GAAG;AAAA,IAClB,UAAE;AACA,cAAQ;AAAA,IACV;AAAA,EACF;AAAA;AAAA,EAGA,IAAI,WAAoB;AACtB,WAAO,KAAK;AAAA,EACd;AAAA;AAAA,EAGA,IAAI,UAAkB;AACpB,WAAO,KAAK,QAAQ;AAAA,EACtB;AACF;AAOA,eAAsB,cACpB,OACA,aACA,KACc;AACd,MAAI,CAAC,OAAO,UAAU,WAAW,KAAK,cAAc,GAAG;AACrD,UAAM,IAAI,MAAM,8DAA8D,WAAW,EAAE;AAAA,EAC7F;AACA,MAAI,MAAM,WAAW,EAAG,QAAO,CAAC;AAEhC,QAAM,UAAU,IAAI,MAAS,MAAM,MAAM;AACzC,MAAI,YAAY;AAChB,MAAI,UAAU;AACd,MAAI,SAAS;AACb,MAAI;AAEJ,QAAM,SAAS,YAA2B;AACxC,WAAO,CAAC,SAAS;AACf,YAAM,QAAQ;AACd,mBAAa;AACb,UAAI,SAAS,MAAM,OAAQ;AAE3B,UAAI;AACF,gBAAQ,KAAK,IAAI,MAAM,IAAI,MAAM,KAAK,GAAI,KAAK;AAAA,MACjD,SAAS,OAAO;AACd,kBAAU;AACV,YAAI,CAAC,QAAQ;AACX,mBAAS;AACT,oBAAU;AAAA,QACZ;AAAA,MACF;AAAA,IACF;AAAA,EACF;AAEA,QAAM,QAAQ,IAAI,MAAM,KAAK,EAAE,QAAQ,KAAK,IAAI,aAAa,MAAM,MAAM,EAAE,GAAG,MAAM,OAAO,CAAC,CAAC;AAC7F,MAAI,OAAQ,OAAM;AAClB,SAAO;AACT;;;ACtEO,IAAM,4BAA6C;AAAA,EACxD,SAAS;AAAA,EACT,cAAc;AAAA,EACd,kBAAkB;AAAA,EAClB,cAAc;AAAA,EACd,gBAAgB;AAAA,EAChB,cAAc;AAAA,EACd,aAAa;AAAA,EACb,WAAW;AAAA,EACX,kBAAkB;AAAA,EAClB,SAAS;AAAA,EACT,aAAa;AACf;AAEO,SAAS,kBAAkB,OAAiB,UAAoC,CAAC,GAAW;AACjG,QAAM,IAAI,EAAE,GAAG,2BAA2B,GAAG,QAAQ;AACrD,SACE,EAAE,UAAU,QAAQ,MAAM,OAAO,IACjC,EAAE,eAAe,QAAQ,MAAM,YAAY,IAC3C,EAAE,mBAAmB,QAAQ,MAAM,gBAAgB,IACnD,EAAE,eAAe,QAAQ,MAAM,YAAY,IAC3C,EAAE,iBAAiB,QAAQ,MAAM,cAAc,IAC/C,EAAE,eAAe,QAAQ,MAAM,YAAY,IAC3C,EAAE,cAAc,QAAQ,MAAM,WAAW,IACzC,EAAE,YAAY,QAAQ,MAAM,SAAS,IACrC,EAAE,mBAAmB,QAAQ,MAAM,gBAAgB,IACnD,EAAE,UAAU,KAAK,IAAI,GAAG,aAAa,MAAM,OAAO,CAAC,IACnD,EAAE,cAAc,KAAK,IAAI,GAAG,aAAa,MAAM,WAAW,IAAI,EAAE;AAEpE;AAEO,SAAS,QAAQ,OAAuB;AAC7C,MAAI,CAAC,OAAO,SAAS,KAAK,EAAG,QAAO;AACpC,SAAO,KAAK,IAAI,GAAG,KAAK,IAAI,GAAG,KAAK,CAAC;AACvC;AAEA,SAAS,aAAa,OAAuB;AAC3C,SAAO,OAAO,SAAS,KAAK,IAAI,QAAQ;AAC1C;;;AClEO,SAAS,uBACX,SACsB;AACzB,QAAM,SAAS;AAAA,IACb,GAAG,IAAI,IAAI,QAAQ,OAAO,CAAC,WAAkC,WAAW,MAAS,CAAC;AAAA,EACpF;AACA,MAAI,OAAO,WAAW,EAAG,QAAO;AAChC,MAAI,OAAO,WAAW,EAAG,QAAO,OAAO,CAAC;AACxC,SAAO,YAAY,IAAI,MAAM;AAC/B;","names":[]}