agentfootprint 10.3.0 → 10.4.0

package/dist/esm/lib/context-bisect/ablation.js

@@ -1,233 +0,0 @@
-/**
- * Ablation — the counterfactual seam (RFC-003 Part B, D8 stage 4 + the
- * D9 stats engine).
- *
- * Three pieces:
- *
- *   1. **Adapters** — `ablationForSuspect` maps a classified suspect to
- *      the spec that removes it (tool → drop from catalog; injection /
- *      fact / skill → exclude the `Injection.id`; memory → filter the
- *      `MemoryEntry.id`; arg → consumer-override note).
- *
- *   2. **The seam** — `applyAblations` filters the inputs an agent is
- *      BUILT from. Documented here because the seam did not previously
- *      exist: `AgentOptions` has no `ignoredTools` runtime kill-switch, so
- *      tool ablation happens at construction (the consumer's
- *      `AblationRunner` rebuilds the agent from filtered inputs). Same for
- *      injections and memory entries.
- *
- *   3. **The probe engine** — `runAblationProbe` calls the consumer's
- *      runner N seeded times, measures embedding similarity to the
- *      original output, counts outcome flips, and returns variance —
- *      never a single-run verdict (D9 discipline).
- *
- * §B2: only `runAblationProbe`-derived verdicts are causal claims; every
- * score elsewhere is a correlational proxy.
- */
-import { cosineSimilarity } from '../../memory/embedding/cosine.js';
-import { CONTEXT_BISECT_DEFAULTS } from './types.js';
-// ─── Adapters: suspect → spec ────────────────────────────────────────
-/**
- * The spec that removes one suspect — or `undefined` for kind `'stage'`
- * (plain pipeline stages have no removable input; re-rank or refactor).
- */
-export function ablationForSuspect(suspect) {
-    switch (suspect.kind) {
-        case 'tool':
-            return suspect.detail?.toolName !== undefined
-                ? { kind: 'tool', ignoredTools: [suspect.detail.toolName] }
-                : undefined;
-        case 'injection':
-            return suspect.detail?.injectionId !== undefined
-                ? { kind: 'injection', excludeInjectionIds: [suspect.detail.injectionId] }
-                : undefined;
-        case 'memory':
-            return suspect.detail?.injectionId !== undefined
-                ? { kind: 'memory', excludeMemoryIds: [suspect.detail.injectionId] }
-                : undefined;
-        case 'arg':
-            return {
-                kind: 'arg',
-                source: suspect.source,
-                note: `step ${suspect.source} consumed untracked run input ($getArgs()/env) — ` +
-                    `the runner must override the input itself; the library cannot filter it.`,
-            };
-        case 'stage':
-            return undefined;
-    }
-}
-/**
- * Apply ablation specs to the inputs an agent is constructed from —
- * THE documented seam (see module docs). Generic over the concrete tool /
- * injection / memory-entry types so it filters without importing them.
- *
- * `'arg'` specs are deliberately NOT handled here: run input belongs to
- * the consumer's runner (`spec.note` says so).
- *
- * @example inside an AblationRunner
- * ```ts
- * const { tools, injections } = applyAblations(specs, {
- *   tools: ALL_TOOLS, injections: ALL_FACTS,
- * });
- * const agent = Agent.create({ provider: freshProvider(), model })
- *   .tools([...tools]);
- * for (const inj of injections) agent.fact(inj);
- * ```
- */
-export function applyAblations(specs, targets) {
-    const ignoredTools = new Set();
-    const excludedInjections = new Set();
-    const excludedMemory = new Set();
-    for (const spec of specs) {
-        if (spec.kind === 'tool')
-            for (const name of spec.ignoredTools)
-                ignoredTools.add(name);
-        if (spec.kind === 'injection')
-            for (const id of spec.excludeInjectionIds)
-                excludedInjections.add(id);
-        if (spec.kind === 'memory')
-            for (const id of spec.excludeMemoryIds)
-                excludedMemory.add(id);
-    }
-    return {
-        tools: (targets.tools ?? []).filter((tool) => !ignoredTools.has(tool.schema.name)),
-        injections: (targets.injections ?? []).filter((injection) => !excludedInjections.has(injection.id)),
-        memoryEntries: (targets.memoryEntries ?? []).filter((entry) => !excludedMemory.has(entry.id)),
-    };
-}
-// ─── The probe engine (D9 stats) ─────────────────────────────────────
-/** Resolve the seeded-rerun count: default on non-finite, floor, clamp to >= 2
- *  (no single-run verdicts — D9). Shared by the ablation + restoration probes. */
-export function resolveSamples(samples) {
-    const raw = samples ?? CONTEXT_BISECT_DEFAULTS.samples;
-    return Math.max(2, Number.isFinite(raw) ? Math.floor(raw) : CONTEXT_BISECT_DEFAULTS.samples);
-}
-/** Median of a numeric sample (mean of the two middles for even length). */
-export function median(values) {
-    if (values.length === 0)
-        return 0;
-    const sorted = [...values].sort((a, b) => a - b);
-    const mid = sorted.length >> 1;
-    return sorted.length % 2 ? sorted[mid] : (sorted[mid - 1] + sorted[mid]) / 2;
-}
-function costRange(values) {
-    return { median: median(values), min: Math.min(...values), max: Math.max(...values) };
-}
-/** Build a probe's CostStats from per-seed loop/token samples — undefined when
- *  the runner reported no cost (keeps quality-only behavior byte-identical). */
-export function costStatsFrom(samples, loops, tokens) {
-    if (loops.length === 0 && tokens.length === 0)
-        return undefined;
-    return {
-        samples,
-        ...(loops.length > 0 ? { loops: costRange(loops) } : {}),
-        ...(tokens.length > 0 ? { tokens: costRange(tokens) } : {}),
-    };
-}
-export function similarityStats(values) {
-    if (values.length === 0)
-        return { mean: 0, min: 0, max: 0, stdev: 0 };
-    const mean = values.reduce((sum, v) => sum + v, 0) / values.length;
-    const variance = values.reduce((sum, v) => sum + (v - mean) ** 2, 0) / values.length;
-    return {
-        mean,
-        min: Math.min(...values),
-        max: Math.max(...values),
-        stdev: Math.sqrt(variance),
-    };
-}
-/** The default comparator: embedding similarity below the threshold. */
-export function defaultOutcomeComparator(embedder, flipThreshold) {
-    return async (original, ablated) => {
-        const [a, b] = await Promise.all([
-            embedder.embed({ text: original }),
-            embedder.embed({ text: ablated }),
-        ]);
-        return cosineSimilarity(a, b) < flipThreshold;
-    };
-}
-/**
- * Run ONE probe: call the consumer's runner with `specs` once per seed
- * (0..samples-1), measure each output's embedding similarity to the
- * original, and count outcome flips. Variance is always reported.
- *
- * `samples` is clamped to ≥ 2 — D9: never single-run verdicts.
- */
-export async function runAblationProbe(config, specs) {
-    const samples = resolveSamples(config.rerun.samples);
-    const flipThreshold = config.rerun.flipThreshold ?? CONTEXT_BISECT_DEFAULTS.flipThreshold;
-    const outcomeChanged = config.rerun.outcomeChanged ?? defaultOutcomeComparator(config.embedder, flipThreshold);
-    const similarities = [];
-    const loopsPerSeed = [];
-    const tokensPerSeed = [];
-    let flips = 0;
-    const originalVec = await config.embedder.embed({ text: config.rerun.originalOutput });
-    for (let seed = 0; seed < samples; seed++) {
-        // Normalize string | { output, cost } — backward-compatible (one ablation,
-        // two readouts: output for the flip, cost for the second score).
-        const raw = await config.rerun.runner(specs, { seed });
-        const output = typeof raw === 'string' ? raw : raw.output;
-        const cost = typeof raw === 'string' ? undefined : raw.cost;
-        if (cost?.loops !== undefined)
-            loopsPerSeed.push(cost.loops);
-        if (cost?.tokens !== undefined)
-            tokensPerSeed.push(cost.tokens);
-        const outputVec = await config.embedder.embed({ text: output });
-        similarities.push(cosineSimilarity(originalVec, outputVec));
-        if (await outcomeChanged(config.rerun.originalOutput, output))
-            flips++;
-    }
-    const cost = costStatsFrom(samples, loopsPerSeed, tokensPerSeed);
-    return {
-        samples,
-        flips,
-        similarity: similarityStats(similarities),
-        ...(cost !== undefined ? { cost } : {}),
-    };
-}
-/** Majority-flip rule shared by D8 verdicts and D9 probes. */
-export function probeFlipped(stats) {
-    return stats.flips * 2 > stats.samples;
-}
-/**
- * Translate probe evidence into the verdict — the ONLY causal claim tier
- * (§B2). `baselineStable=false` (the un-ablated scenario itself flipped)
- * forces `'inconclusive'`: no ablation verdict is trustworthy on an
- * unstable baseline.
- */
-export function verdictFor(label, stats, baselineStable, 
-/** The counterfactual intervention. `'ablating'` (default) for present
- *  suspects; `'restoring'` for missing-context candidates (interface #3).
- *  Default keeps every claim string byte-identical to before. */
-action = 'ablating') {
-    const baselineWord = action === 'ablating' ? 'un-ablated' : 'un-restored';
-    const tierWord = action === 'ablating' ? 'ablation' : 'restoration';
-    if (!baselineStable) {
-        return {
-            verdict: 'inconclusive',
-            claim: `INCONCLUSIVE: the ${baselineWord} baseline itself changed outcome across seeded reruns — ` +
-                `no ${tierWord} verdict for ${label} is trustworthy on an unstable scenario.`,
-        };
-    }
-    if (probeFlipped(stats)) {
-        return {
-            verdict: 'confirmed',
-            claim: `CAUSAL: ${action} ${label} flipped the outcome in ${stats.flips}/${stats.samples} ` +
-                `seeded reruns (mean similarity to original ${stats.similarity.mean.toFixed(3)} ` +
-                `± ${stats.similarity.stdev.toFixed(3)}).`,
-        };
-    }
-    if (stats.flips > 0) {
-        return {
-            verdict: 'inconclusive',
-            claim: `INCONCLUSIVE: ${action} ${label} flipped only ${stats.flips}/${stats.samples} seeded ` +
-                `reruns — below majority; raise samples or check scenario stability.`,
-        };
-    }
-    return {
-        verdict: 'not-confirmed',
-        claim: `NOT CONFIRMED: ${action} ${label} did not change the outcome in ${stats.samples} seeded ` +
-            `reruns — its ranking remains a correlational proxy only.`,
-    };
-}
-//# sourceMappingURL=ablation.js.map

package/dist/esm/lib/context-bisect/ablation.js.map

@@ -1 +0,0 @@
-{"version":3,"file":"ablation.js","sourceRoot":"","sources":["../../../../src/lib/context-bisect/ablation.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;;;;;;;;GAyBG;AAEH,OAAO,EAAE,gBAAgB,EAAE,MAAM,kCAAkC,CAAC;AAapE,OAAO,EAAE,uBAAuB,EAAE,MAAM,YAAY,CAAC;AAErD,wEAAwE;AAExE;;;GAGG;AACH,MAAM,UAAU,kBAAkB,CAAC,OAAgB;IACjD,QAAQ,OAAO,CAAC,IAAI,EAAE,CAAC;QACrB,KAAK,MAAM;YACT,OAAO,OAAO,CAAC,MAAM,EAAE,QAAQ,KAAK,SAAS;gBAC3C,CAAC,CAAC,EAAE,IAAI,EAAE,MAAM,EAAE,YAAY,EAAE,CAAC,OAAO,CAAC,MAAM,CAAC,QAAQ,CAAC,EAAE;gBAC3D,CAAC,CAAC,SAAS,CAAC;QAChB,KAAK,WAAW;YACd,OAAO,OAAO,CAAC,MAAM,EAAE,WAAW,KAAK,SAAS;gBAC9C,CAAC,CAAC,EAAE,IAAI,EAAE,WAAW,EAAE,mBAAmB,EAAE,CAAC,OAAO,CAAC,MAAM,CAAC,WAAW,CAAC,EAAE;gBAC1E,CAAC,CAAC,SAAS,CAAC;QAChB,KAAK,QAAQ;YACX,OAAO,OAAO,CAAC,MAAM,EAAE,WAAW,KAAK,SAAS;gBAC9C,CAAC,CAAC,EAAE,IAAI,EAAE,QAAQ,EAAE,gBAAgB,EAAE,CAAC,OAAO,CAAC,MAAM,CAAC,WAAW,CAAC,EAAE;gBACpE,CAAC,CAAC,SAAS,CAAC;QAChB,KAAK,KAAK;YACR,OAAO;gBACL,IAAI,EAAE,KAAK;gBACX,MAAM,EAAE,OAAO,CAAC,MAAM;gBACtB,IAAI,EACF,QAAQ,OAAO,CAAC,MAAM,mDAAmD;oBACzE,0EAA0E;aAC7E,CAAC;QACJ,KAAK,OAAO;YACV,OAAO,SAAS,CAAC;IACrB,CAAC;AACH,CAAC;AAwBD;;;;;;;;;;;;;;;;;GAiBG;AACH,MAAM,UAAU,cAAc,CAK5B,KAA8B,EAC9B,OAAyD;IAMzD,MAAM,YAAY,GAAG,IAAI,GAAG,EAAU,CAAC;IACvC,MAAM,kBAAkB,GAAG,IAAI,GAAG,EAAU,CAAC;IAC7C,MAAM,cAAc,GAAG,IAAI,GAAG,EAAU,CAAC;IACzC,KAAK,MAAM,IAAI,IAAI,KAAK,EAAE,CAAC;QACzB,IAAI,IAAI,CAAC,IAAI,KAAK,MAAM;YAAE,KAAK,MAAM,IAAI,IAAI,IAAI,CAAC,YAAY;gBAAE,YAAY,CAAC,GAAG,CAAC,IAAI,CAAC,CAAC;QACvF,IAAI,IAAI,CAAC,IAAI,KAAK,WAAW;YAC3B,KAAK,MAAM,EAAE,IAAI,IAAI,CAAC,mBAAmB;gBAAE,kBAAkB,CAAC,GAAG,CAAC,EAAE,CAAC,CAAC;QACxE,IAAI,IAAI,CAAC,IAAI,KAAK,QAAQ;YAAE,KAAK,MAAM,EAAE,IAAI,IAAI,CAAC,gBAAgB;gBAAE,cAAc,CAAC,GAAG,CAAC,EAAE,CAAC,CAAC;IAC7F,CAAC;IACD,OAAO;QACL,KAAK,EAAE,CAAC,OAAO,CAAC,KAAK,IAAI,EAAE,CAAC,CAAC,MAAM,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,CAAC,YAAY,CAAC,GAAG,CAAC,IAAI,CAAC,MAAM,CAAC,IAAI,CAAC,CAAC;QAClF,UAAU,EAAE,CAAC,OAAO,CAAC,UAAU,IAAI,EAAE,CAAC,CAAC,MAAM,CAC3C,CAAC,SAAS,EAAE,EAAE,CAAC,CAAC,kBAAkB,CAAC,GAAG,CAAC,SAAS,CAAC,EAAE,CAAC,CACrD;QACD,aAAa,EAAE,CAAC,OAAO,CAAC,aAAa,IAAI,EAAE,CAAC,CAAC,MAAM,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,CAAC,cAAc,CAAC,GAAG,CAAC,KAAK,CAAC,EAAE,CAAC,CAAC;KAC9F,CAAC;AACJ,CAAC;AAED,wEAAwE;AAExE;kFACkF;AAClF,MAAM,UAAU,cAAc,CAAC,OAA2B;IACxD,MAAM,GAAG,GAAG,OAAO,IAAI,uBAAuB,CAAC,OAAO,CAAC;IACvD,OAAO,IAAI,CAAC,GAAG,CAAC,CAAC,EAAE,MAAM,CAAC,QAAQ,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,uBAAuB,CAAC,OAAO,CAAC,CAAC;AAC/F,CAAC;AAED,4EAA4E;AAC5E,MAAM,UAAU,MAAM,CAAC,MAAyB;IAC9C,IAAI,MAAM,CAAC,MAAM,KAAK,CAAC;QAAE,OAAO,CAAC,CAAC;IAClC,MAAM,MAAM,GAAG,CAAC,GAAG,MAAM,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC;IACjD,MAAM,GAAG,GAAG,MAAM,CAAC,MAAM,IAAI,CAAC,CAAC;IAC/B,OAAO,MAAM,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC,CAAC,MAAM,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,CAAC,MAAM,CAAC,GAAG,GAAG,CAAC,CAAC,GAAG,MAAM,CAAC,GAAG,CAAC,CAAC,GAAG,CAAC,CAAC;AAC/E,CAAC;AAED,SAAS,SAAS,CAAC,MAAyB;IAC1C,OAAO,EAAE,MAAM,EAAE,MAAM,CAAC,MAAM,CAAC,EAAE,GAAG,EAAE,IAAI,CAAC,GAAG,CAAC,GAAG,MAAM,CAAC,EAAE,GAAG,EAAE,IAAI,CAAC,GAAG,CAAC,GAAG,MAAM,CAAC,EAAE,CAAC;AACxF,CAAC;AAED;gFACgF;AAChF,MAAM,UAAU,aAAa,CAC3B,OAAe,EACf,KAAwB,EACxB,MAAyB;IAEzB,IAAI,KAAK,CAAC,MAAM,KAAK,CAAC,IAAI,MAAM,CAAC,MAAM,KAAK,CAAC;QAAE,OAAO,SAAS,CAAC;IAChE,OAAO;QACL,OAAO;QACP,GAAG,CAAC,KAAK,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC,CAAC,EAAE,KAAK,EAAE,SAAS,CAAC,KAAK,CAAC,EAAE,CAAC,CAAC,CAAC,EAAE,CAAC;QACxD,GAAG,CAAC,MAAM,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC,CAAC,EAAE,MAAM,EAAE,SAAS,CAAC,MAAM,CAAC,EAAE,CAAC,CAAC,CAAC,EAAE,CAAC;KAC5D,CAAC;AACJ,CAAC;AAED,MAAM,UAAU,eAAe,CAAC,MAAyB;IACvD,IAAI,MAAM,CAAC,MAAM,KAAK,CAAC;QAAE,OAAO,EAAE,IAAI,EAAE,CAAC,EAAE,GAAG,EAAE,CAAC,EAAE,GAAG,EAAE,CAAC,EAAE,KAAK,EAAE,CAAC,EAAE,CAAC;IACtE,MAAM,IAAI,GAAG,MAAM,CAAC,MAAM,CAAC,CAAC,GAAG,EAAE,CAAC,EAAE,EAAE,CAAC,GAAG,GAAG,CAAC,EAAE,CAAC,CAAC,GAAG,MAAM,CAAC,MAAM,CAAC;IACnE,MAAM,QAAQ,GAAG,MAAM,CAAC,MAAM,CAAC,CAAC,GAAG,EAAE,CAAC,EAAE,EAAE,CAAC,GAAG,GAAG,CAAC,CAAC,GAAG,IAAI,CAAC,IAAI,CAAC,EAAE,CAAC,CAAC,GAAG,MAAM,CAAC,MAAM,CAAC;IACrF,OAAO;QACL,IAAI;QACJ,GAAG,EAAE,IAAI,CAAC,GAAG,CAAC,GAAG,MAAM,CAAC;QACxB,GAAG,EAAE,IAAI,CAAC,GAAG,CAAC,GAAG,MAAM,CAAC;QACxB,KAAK,EAAE,IAAI,CAAC,IAAI,CAAC,QAAQ,CAAC;KAC3B,CAAC;AACJ,CAAC;AAED,wEAAwE;AACxE,MAAM,UAAU,wBAAwB,CACtC,QAAkB,EAClB,aAAqB;IAErB,OAAO,KAAK,EAAE,QAAQ,EAAE,OAAO,EAAE,EAAE;QACjC,MAAM,CAAC,CAAC,EAAE,CAAC,CAAC,GAAG,MAAM,OAAO,CAAC,GAAG,CAAC;YAC/B,QAAQ,CAAC,KAAK,CAAC,EAAE,IAAI,EAAE,QAAQ,EAAE,CAAC;YAClC,QAAQ,CAAC,KAAK,CAAC,EAAE,IAAI,EAAE,OAAO,EAAE,CAAC;SAClC,CAAC,CAAC;QACH,OAAO,gBAAgB,CAAC,CAAC,EAAE,CAAC,CAAC,GAAG,aAAa,CAAC;IAChD,CAAC,CAAC;AACJ,CAAC;AAQD;;;;;;GAMG;AACH,MAAM,CAAC,KAAK,UAAU,gBAAgB,CACpC,MAAmB,EACnB,KAA8B;IAE9B,MAAM,OAAO,GAAG,cAAc,CAAC,MAAM,CAAC,KAAK,CAAC,OAAO,CAAC,CAAC;IACrD,MAAM,aAAa,GAAG,MAAM,CAAC,KAAK,CAAC,aAAa,IAAI,uBAAuB,CAAC,aAAa,CAAC;IAC1F,MAAM,cAAc,GAClB,MAAM,CAAC,KAAK,CAAC,cAAc,IAAI,wBAAwB,CAAC,MAAM,CAAC,QAAQ,EAAE,aAAa,CAAC,CAAC;IAE1F,MAAM,YAAY,GAAa,EAAE,CAAC;IAClC,MAAM,YAAY,GAAa,EAAE,CAAC;IAClC,MAAM,aAAa,GAAa,EAAE,CAAC;IACnC,IAAI,KAAK,GAAG,CAAC,CAAC;IACd,MAAM,WAAW,GAAG,MAAM,MAAM,CAAC,QAAQ,CAAC,KAAK,CAAC,EAAE,IAAI,EAAE,MAAM,CAAC,KAAK,CAAC,cAAc,EAAE,CAAC,CAAC;IACvF,KAAK,IAAI,IAAI,GAAG,CAAC,EAAE,IAAI,GAAG,OAAO,EAAE,IAAI,EAAE,EAAE,CAAC;QAC1C,2EAA2E;QAC3E,iEAAiE;QACjE,MAAM,GAAG,GAAG,MAAM,MAAM,CAAC,KAAK,CAAC,MAAM,CAAC,KAAK,EAAE,EAAE,IAAI,EAAE,CAAC,CAAC;QACvD,MAAM,MAAM,GAAG,OAAO,GAAG,KAAK,QAAQ,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,GAAG,CAAC,MAAM,CAAC;QAC1D,MAAM,IAAI,GAAG,OAAO,GAAG,KAAK,QAAQ,CAAC,CAAC,CAAC,SAAS,CAAC,CAAC,CAAC,GAAG,CAAC,IAAI,CAAC;QAC5D,IAAI,IAAI,EAAE,KAAK,KAAK,SAAS;YAAE,YAAY,CAAC,IAAI,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC;QAC7D,IAAI,IAAI,EAAE,MAAM,KAAK,SAAS;YAAE,aAAa,CAAC,IAAI,CAAC,IAAI,CAAC,MAAM,CAAC,CAAC;QAChE,MAAM,SAAS,GAAG,MAAM,MAAM,CAAC,QAAQ,CAAC,KAAK,CAAC,EAAE,IAAI,EAAE,MAAM,EAAE,CAAC,CAAC;QAChE,YAAY,CAAC,IAAI,CAAC,gBAAgB,CAAC,WAAW,EAAE,SAAS,CAAC,CAAC,CAAC;QAC5D,IAAI,MAAM,cAAc,CAAC,MAAM,CAAC,KAAK,CAAC,cAAc,EAAE,MAAM,CAAC;YAAE,KAAK,EAAE,CAAC;IACzE,CAAC;IACD,MAAM,IAAI,GAAG,aAAa,CAAC,OAAO,EAAE,YAAY,EAAE,aAAa,CAAC,CAAC;IACjE,OAAO;QACL,OAAO;QACP,KAAK;QACL,UAAU,EAAE,eAAe,CAAC,YAAY,CAAC;QACzC,GAAG,CAAC,IAAI,KAAK,SAAS,CAAC,CAAC,CAAC,EAAE,IAAI,EAAE,CAAC,CAAC,CAAC,EAAE,CAAC;KACxC,CAAC;AACJ,CAAC;AAED,8DAA8D;AAC9D,MAAM,UAAU,YAAY,CAAC,KAAuB;IAClD,OAAO,KAAK,CAAC,KAAK,GAAG,CAAC,GAAG,KAAK,CAAC,OAAO,CAAC;AACzC,CAAC;AAED;;;;;GAKG;AACH,MAAM,UAAU,UAAU,CACxB,KAAa,EACb,KAAuB,EACvB,cAAuB;AACvB;;iEAEiE;AACjE,SAAmC,UAAU;IAE7C,MAAM,YAAY,GAAG,MAAM,KAAK,UAAU,CAAC,CAAC,CAAC,YAAY,CAAC,CAAC,CAAC,aAAa,CAAC;IAC1E,MAAM,QAAQ,GAAG,MAAM,KAAK,UAAU,CAAC,CAAC,CAAC,UAAU,CAAC,CAAC,CAAC,aAAa,CAAC;IACpE,IAAI,CAAC,cAAc,EAAE,CAAC;QACpB,OAAO;YACL,OAAO,EAAE,cAAc;YACvB,KAAK,EACH,qBAAqB,YAAY,0DAA0D;gBAC3F,MAAM,QAAQ,gBAAgB,KAAK,0CAA0C;SAChF,CAAC;IACJ,CAAC;IACD,IAAI,YAAY,CAAC,KAAK,CAAC,EAAE,CAAC;QACxB,OAAO;YACL,OAAO,EAAE,WAAW;YACpB,KAAK,EACH,WAAW,MAAM,IAAI,KAAK,2BAA2B,KAAK,CAAC,KAAK,IAAI,KAAK,CAAC,OAAO,GAAG;gBACpF,8CAA8C,KAAK,CAAC,UAAU,CAAC,IAAI,CAAC,OAAO,CAAC,CAAC,CAAC,GAAG;gBACjF,KAAK,KAAK,CAAC,UAAU,CAAC,KAAK,CAAC,OAAO,CAAC,CAAC,CAAC,IAAI;SAC7C,CAAC;IACJ,CAAC;IACD,IAAI,KAAK,CAAC,KAAK,GAAG,CAAC,EAAE,CAAC;QACpB,OAAO;YACL,OAAO,EAAE,cAAc;YACvB,KAAK,EACH,iBAAiB,MAAM,IAAI,KAAK,iBAAiB,KAAK,CAAC,KAAK,IAAI,KAAK,CAAC,OAAO,UAAU;gBACvF,qEAAqE;SACxE,CAAC;IACJ,CAAC;IACD,OAAO;QACL,OAAO,EAAE,eAAe;QACxB,KAAK,EACH,kBAAkB,MAAM,IAAI,KAAK,kCAAkC,KAAK,CAAC,OAAO,UAAU;YAC1F,0DAA0D;KAC7D,CAAC;AACJ,CAAC"}

package/dist/esm/lib/context-bisect/arms/apply.js.map

@@ -1 +0,0 @@
-{"version":3,"file":"apply.js","sourceRoot":"","sources":["../../../../../src/lib/context-bisect/arms/apply.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;;GAmBG;AAEH,OAAO,EAAE,cAAc,EAAwB,MAAM,gBAAgB,CAAC;AAatE;;;;;;;;;;;;;;;GAeG;AACH,MAAM,UAAU,QAAQ,CAKtB,GAAgB,EAChB,OAAyD;IAMzD,OAAO,cAAc,CAAC,GAAG,CAAC,SAAS,IAAI,EAAE,EAAE,OAAO,CAAC,CAAC;AACtD,CAAC"}

package/dist/esm/lib/context-bisect/arms/compare.js.map

@@ -1 +0,0 @@
-{"version":3,"file":"compare.js","sourceRoot":"","sources":["../../../../../src/lib/context-bisect/arms/compare.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;;;;;;;GAwBG;AAEH,OAAO,EAAE,wBAAwB,EAAE,cAAc,EAAE,MAAM,gBAAgB,CAAC;AAC1E,OAAO,EAAE,uBAAuB,EAAE,MAAM,aAAa,CAAC;AAEtD,OAAO,EACL,mBAAmB,EACnB,cAAc,EACd,YAAY,EACZ,YAAY,GAEb,MAAM,YAAY,CAAC;AAOpB,OAAO,EAAE,oBAAoB,EAAE,MAAM,eAAe,CAAC;AACrD,OAAO,EAAE,aAAa,EAAE,MAAM,cAAc,CAAC;AAE7C,MAAM,CAAC,KAAK,UAAU,mBAAmB,CACvC,OAAmC;IAEnC,MAAM,aAAa,GAAG,oBAAoB,CAAC,OAAO,CAAC,IAAI,EAAE,OAAO,CAAC,aAAa,CAAC,CAAC;IAChF,MAAM,OAAO,GAAG,cAAc,CAAC,OAAO,CAAC,OAAO,CAAC,CAAC;IAChD,MAAM,aAAa,GAAG,OAAO,CAAC,aAAa,IAAI,uBAAuB,CAAC,aAAa,CAAC;IACrF,6EAA6E;IAC7E,qEAAqE;IACrE,MAAM,qBAAqB,GAAG,OAAO,CAAC,aAAa,KAAK,SAAS,CAAC;IAClE,MAAM,cAAc,GAClB,OAAO,CAAC,aAAa,IAAI,wBAAwB,CAAC,OAAO,CAAC,QAAQ,EAAE,aAAa,CAAC,CAAC;IAErF,MAAM,WAAW,GAAG,OAAO,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC,GAAG,EAAE,EAAE,CAAC,GAAG,CAAC,EAAE,KAAK,aAAa,CAAgB,CAAC;IACxF,MAAM,YAAY,GAAG,MAAM,cAAc,CAAC,WAAW,EAAE,OAAO,CAAC,MAAM,EAAE,OAAO,CAAC,CAAC;IAEhF,8EAA8E;IAC9E,6EAA6E;IAC7E,4EAA4E;IAC5E,8DAA8D;IAC9D,MAAM,aAAa,GAAG,OAAO,CAAC,cAAc,KAAK,SAAS,CAAC,CAAC,CAAC,WAAW,CAAC,CAAC,CAAC,cAAc,CAAC;IAC1F,MAAM,SAAS,GAAG,OAAO,CAAC,cAAc,IAAI,YAAY,CAAC,OAAO,CAAC,CAAC,CAAC,IAAI,EAAE,CAAC;IAE1E,MAAM,aAAa,GAAG,MAAM,YAAY,CACtC,YAAY,EACZ,SAAS,EACT,OAAO,CAAC,QAAQ,EAChB,cAAc,CACf,CAAC;IACF,qEAAqE;IACrE,8EAA8E;IAC9E,uDAAuD;IACvD,MAAM,cAAc,GAAG,aAAa,CAAC,KAAK,KAAK,CAAC,CAAC;IACjD,MAAM,IAAI,GAAG,YAAY,CAAC,aAAa,EAAE,aAAa,EAAE,qBAAqB,CAAC,CAAC;IAE/E,MAAM,IAAI,GAAiB,EAAE,CAAC;IAC9B,KAAK,MAAM,GAAG,IAAI,OAAO,CAAC,IAAI,EAAE,CAAC;QAC/B,MAAM,UAAU,GAAG,GAAG,CAAC,EAAE,KAAK,aAAa,CAAC;QAC5C,MAAM,IAAI,GAAY,UAAU;YAC9B,CAAC,CAAC,YAAY;YACd,CAAC,CAAC,MAAM,cAAc,CAAC,GAAG,EAAE,OAAO,CAAC,MAAM,EAAE,OAAO,CAAC,CAAC;QACvD,MAAM,KAAK,GAAqB,UAAU;YACxC,CAAC,CAAC,aAAa;YACf,CAAC,CAAC,MAAM,YAAY,CAAC,IAAI,EAAE,SAAS,EAAE,OAAO,CAAC,QAAQ,EAAE,cAAc,CAAC,CAAC;QAC1E,wEAAwE;QACxE,sDAAsD;QACtD,MAAM,eAAe,GAAG,CAAC,UAAU,IAAI,KAAK,CAAC,UAAU,CAAC,IAAI,GAAG,IAAI,CAAC,KAAK,CAAC;QAC1E,MAAM,WAAW,GAAG,mBAAmB,CAAC,GAAG,EAAE,IAAI,CAAC,SAAS,CAAC,CAAC;QAC7D,IAAI,CAAC,IAAI,CAAC;YACR,KAAK,EAAE,GAAG,CAAC,EAAE;YACb,UAAU;YACV,IAAI,EAAE,KAAK;YACX,OAAO,EAAE,IAAI,CAAC,OAAO;YACrB,eAAe;YACf,WAAW;YACX,GAAG,CAAC,UAAU;gBACZ,CAAC,CAAC,EAAE;gBACJ,CAAC,CAAC;oBACE,OAAO,EAAE,aAAa,CAAC,GAAG,CAAC,EAAE,EAAE,KAAK,EAAE;wBACpC,aAAa;wBACb,cAAc;wBACd,aAAa;wBACb,IAAI;wBACJ,eAAe;wBACf,WAAW;qBACZ,CAAC;iBACH,CAAC;SACP,CAAC,CAAC;IACL,CAAC;IAED,OAAO;QACL,SAAS,EAAE,EAAE,IAAI,EAAE,aAAa,EAAE,IAAI,EAAE,SAAS,EAAE;QACnD,aAAa;QACb,cAAc;QACd,QAAQ,EAAE,IAAI;QACd,IAAI;QACJ,QAAQ,EAAE,OAAO,GAAG,OAAO,CAAC,IAAI,CAAC,MAAM;QACvC,OAAO,EAAE,SAAS,CAAC,IAAI,EAAE,aAAa,EAAE,cAAc,EAAE,OAAO,CAAC;KACjE,CAAC;AACJ,CAAC;AAED,+EAA+E;AAC/E,SAAS,SAAS,CAChB,IAA2B,EAC3B,aAAqB,EACrB,cAAuB,EACvB,OAAe;IAEf,MAAM,WAAW,GAAG,IAAI,CAAC,MAAM,CAAC,CAAC,GAAG,EAAE,EAAE,CAAC,CAAC,GAAG,CAAC,UAAU,CAAC,CAAC;IAC1D,MAAM,IAAI,GAAG,GAAG,IAAI,CAAC,MAAM,WAAW,OAAO,cAAc,CAAC;IAC5D,IAAI,CAAC,cAAc,EAAE,CAAC;QACpB,OAAO,CACL,qBAAqB,aAAa,0DAA0D;YAC5F,6DAA6D,IAAI,yBAAyB;YAC1F,WAAW,CACZ,CAAC;IACJ,CAAC;IACD,MAAM,KAAK,GAAG,CAAC,IAAY,EAAU,EAAE,CACrC,WAAW;SACR,MAAM,CAAC,CAAC,GAAG,EAAE,EAAE,CAAC,GAAG,CAAC,OAAO,EAAE,OAAO,KAAK,IAAI,CAAC;SAC9C,GAAG,CAAC,CAAC,GAAG,EAAE,EAAE,CAAC,IAAI,GAAG,CAAC,KAAK,GAAG,CAAC;SAC9B,IAAI,CAAC,IAAI,CAAC,IAAI,MAAM,CAAC;IAC1B,OAAO,GAAG,IAAI,mBAAmB,aAAa,uCAAuC,KAAK,CACxF,WAAW,CACZ,0BAA0B,KAAK,CAAC,eAAe,CAAC,mBAAmB,KAAK,CAAC,cAAc,CAAC,GAAG,CAAC;AAC/F,CAAC"}

package/dist/esm/lib/context-bisect/arms/index.d.ts

@@ -1,16 +0,0 @@
-/**
- * Strategy arms — SUBSTITUTION counterfactuals, kept distinct from ablation's
- * removals and sharing its statistics. See README.md in this folder for the
- * design argument, and `types.ts` for why this is not a fifth `AblationSpec`
- * arm.
- *
- * @beta Beta feature (RFC-003 Part B). The API works and is tested, but may
- * change before GA.
- */
-export { applyArm } from './apply.js';
-export { compareStrategyArms } from './compare.js';
-export { armFacetsFromManifest, armLabel, checkArmApplied, manifestFromEvents, matchArm, RUN_CONFIGURED_EVENT, } from './manifest.js';
-export { checkArmApplication, collectArmRuns, nullBandFrom, scoreArmRuns, type ArmRuns, } from './probe.js';
-export { armConfigKey, declaredFacetCount, validateStrategyArms } from './validate.js';
-export { verdictForArm, type ArmVerdictContext } from './verdict.js';
-export type { ArmApplication, ArmFacetMismatch, ArmFacets, ArmMemoryFacet, ArmOutcome, ArmRunResult, ArmRunner, CompareStrategyArmsOptions, ManifestMemoryLike, NullBand, RunManifestLike, StrategyArm, StrategyComparison, } from './types.js';

package/dist/esm/lib/context-bisect/arms/index.js

@@ -1,16 +0,0 @@
-/**
- * Strategy arms — SUBSTITUTION counterfactuals, kept distinct from ablation's
- * removals and sharing its statistics. See README.md in this folder for the
- * design argument, and `types.ts` for why this is not a fifth `AblationSpec`
- * arm.
- *
- * @beta Beta feature (RFC-003 Part B). The API works and is tested, but may
- * change before GA.
- */
-export { applyArm } from './apply.js';
-export { compareStrategyArms } from './compare.js';
-export { armFacetsFromManifest, armLabel, checkArmApplied, manifestFromEvents, matchArm, RUN_CONFIGURED_EVENT, } from './manifest.js';
-export { checkArmApplication, collectArmRuns, nullBandFrom, scoreArmRuns, } from './probe.js';
-export { armConfigKey, declaredFacetCount, validateStrategyArms } from './validate.js';
-export { verdictForArm } from './verdict.js';
-//# sourceMappingURL=index.js.map

package/dist/esm/lib/context-bisect/arms/index.js.map

@@ -1 +0,0 @@
-{"version":3,"file":"index.js","sourceRoot":"","sources":["../../../../../src/lib/context-bisect/arms/index.ts"],"names":[],"mappings":"AAAA;;;;;;;;GAQG;AAEH,OAAO,EAAE,QAAQ,EAAE,MAAM,YAAY,CAAC;AACtC,OAAO,EAAE,mBAAmB,EAAE,MAAM,cAAc,CAAC;AACnD,OAAO,EACL,qBAAqB,EACrB,QAAQ,EACR,eAAe,EACf,kBAAkB,EAClB,QAAQ,EACR,oBAAoB,GACrB,MAAM,eAAe,CAAC;AACvB,OAAO,EACL,mBAAmB,EACnB,cAAc,EACd,YAAY,EACZ,YAAY,GAEb,MAAM,YAAY,CAAC;AACpB,OAAO,EAAE,YAAY,EAAE,kBAAkB,EAAE,oBAAoB,EAAE,MAAM,eAAe,CAAC;AACvF,OAAO,EAAE,aAAa,EAA0B,MAAM,cAAc,CAAC"}

package/dist/esm/lib/context-bisect/arms/manifest.d.ts

@@ -1,79 +0,0 @@
-/**
- * The manifest bridge — which arm did this run belong to, according to the RUN
- * rather than according to the experimenter.
- *
- * Before 9.41.0 an arm label was bookkeeping: the consumer set a variable to
- * `'rerank'`, built an agent it believed re-ranked, and nothing in the library
- * could contradict it. `agentfootprint.agent.run_configured` names the adapters
- * and strategies a run is about to use, so the belief became checkable — and
- * {@link ArmFacets} is deliberately spelled in that event's own vocabulary so
- * the check is a field-by-field comparison and not a translation layer.
- *
- * Two directions, both here:
- *  - **verify** — `checkArmApplied(arm, manifest)`: the arm said `retrieval:
- *    'rerank'`; did the run agree? A contradiction costs the arm its verdict
- *    (`compare.ts`), because a difference measured between two arms that were
- *    secretly one configuration is not evidence about either.
- *  - **classify** — `matchArm(arms, manifest)`: hand it a recorded run's
- *    manifest and it says which declared arm that run belongs to. This is the
- *    door for grouping runs a study already made, offline, out of saved
- *    recordings.
- *
- * ## Absence is a contradiction, not a wildcard
- *
- * The manifest's own rule is that an absent field means "not configured", never
- * a guessed default. So an arm declaring `window: 'tokenBudget'` against a
- * manifest with no `window` is a MISMATCH, reported with `observed` absent. The
- * alternative — treating absence as "could be anything" — would let a run with
- * no window stage pass as the token-budget arm, which is exactly the mis-wiring
- * this file exists to catch.
- */
-import type { CapturedEventLike } from '../types.js';
-import type { ArmFacetMismatch, ArmFacets, RunManifestLike, StrategyArm } from './types.js';
-/** The event type the run manifest rides on (9.41.0). */
-export declare const RUN_CONFIGURED_EVENT = "agentfootprint.agent.run_configured";
-/**
- * Pull the run manifest out of a run's captured typed events — the same
- * `CapturedEventLike[]` bag `localizeContextBug` already accepts, so a consumer
- * who collected events with `agent.on('*', …)` needs no second capture.
- *
- * One manifest per run by design; the FIRST is returned. A bag holding events
- * from several runs is the caller's own mixing — filter by `meta.runId` first.
- */
-export declare function manifestFromEvents(events: readonly CapturedEventLike[] | undefined): RunManifestLike | undefined;
-/**
- * Project a manifest onto the arm-facet vocabulary — the inverse of what an arm
- * declares. Memory rows collapse to the FIRST one; a study varying retrieval
- * across several mounted memories should declare `memory.id` and read the rows
- * itself rather than trust a projection to pick.
- */
-export declare function armFacetsFromManifest(manifest: RunManifestLike): ArmFacets;
-/**
- * A stable label for a set of facets — sorted `key=value`, so two runs of one
- * configuration produce the same string whatever order the fields were written
- * in. Useful as a grouping key for N recorded runs.
- */
-export declare function armLabel(facets: ArmFacets): string;
-/**
- * Compare one arm's DECLARED facets against a run's manifest. Empty result =
- * the run agreed with everything the arm claimed.
- *
- * Only declared facets are compared. An arm is a claim about the facets it
- * names and says nothing about the rest — checking undeclared fields would fail
- * every arm for differing from the manifest in ways it never claimed.
- */
-export declare function checkArmApplied(arm: StrategyArm, manifest: RunManifestLike): readonly ArmFacetMismatch[];
-/**
- * Which declared arm does this run belong to? The offline door: group N
- * recorded runs into arms by what each run's own manifest says.
- *
- * Rules, all conservative:
- * - only arms that DECLARE facets are candidates (an arm naming nothing matches
- *   every run and would swallow the whole study);
- * - a candidate matches when the manifest contradicts none of its facets;
- * - the MOST SPECIFIC match wins (most declared facets), and a tie returns
- *   `undefined` — two arms fitting one run equally well is an ambiguity, and
- *   guessing which the experimenter meant is exactly the bookkeeping this
- *   function replaces.
- */
-export declare function matchArm(arms: readonly StrategyArm[], manifest: RunManifestLike): StrategyArm | undefined;

package/dist/esm/lib/context-bisect/arms/manifest.js

@@ -1,199 +0,0 @@
-/**
- * The manifest bridge — which arm did this run belong to, according to the RUN
- * rather than according to the experimenter.
- *
- * Before 9.41.0 an arm label was bookkeeping: the consumer set a variable to
- * `'rerank'`, built an agent it believed re-ranked, and nothing in the library
- * could contradict it. `agentfootprint.agent.run_configured` names the adapters
- * and strategies a run is about to use, so the belief became checkable — and
- * {@link ArmFacets} is deliberately spelled in that event's own vocabulary so
- * the check is a field-by-field comparison and not a translation layer.
- *
- * Two directions, both here:
- *  - **verify** — `checkArmApplied(arm, manifest)`: the arm said `retrieval:
- *    'rerank'`; did the run agree? A contradiction costs the arm its verdict
- *    (`compare.ts`), because a difference measured between two arms that were
- *    secretly one configuration is not evidence about either.
- *  - **classify** — `matchArm(arms, manifest)`: hand it a recorded run's
- *    manifest and it says which declared arm that run belongs to. This is the
- *    door for grouping runs a study already made, offline, out of saved
- *    recordings.
- *
- * ## Absence is a contradiction, not a wildcard
- *
- * The manifest's own rule is that an absent field means "not configured", never
- * a guessed default. So an arm declaring `window: 'tokenBudget'` against a
- * manifest with no `window` is a MISMATCH, reported with `observed` absent. The
- * alternative — treating absence as "could be anything" — would let a run with
- * no window stage pass as the token-budget arm, which is exactly the mis-wiring
- * this file exists to catch.
- */
-import { declaredFacetCount } from './validate.js';
-/** The event type the run manifest rides on (9.41.0). */
-export const RUN_CONFIGURED_EVENT = 'agentfootprint.agent.run_configured';
-/**
- * Pull the run manifest out of a run's captured typed events — the same
- * `CapturedEventLike[]` bag `localizeContextBug` already accepts, so a consumer
- * who collected events with `agent.on('*', …)` needs no second capture.
- *
- * One manifest per run by design; the FIRST is returned. A bag holding events
- * from several runs is the caller's own mixing — filter by `meta.runId` first.
- */
-export function manifestFromEvents(events) {
-    const event = events?.find((e) => e.type === RUN_CONFIGURED_EVENT);
-    if (event === undefined)
-        return undefined;
-    const payload = event.payload;
-    // Structural, not a cast-and-pray: a payload that is not an object at all is
-    // no manifest, and reporting `undefined` keeps `checked: false` honest.
-    return payload !== null && typeof payload === 'object' ? payload : undefined;
-}
-/**
- * Project a manifest onto the arm-facet vocabulary — the inverse of what an arm
- * declares. Memory rows collapse to the FIRST one; a study varying retrieval
- * across several mounted memories should declare `memory.id` and read the rows
- * itself rather than trust a projection to pick.
- */
-export function armFacetsFromManifest(manifest) {
-    const memory = manifest.memories?.[0];
-    const provider = manifest.llm?.provider;
-    const model = manifest.llm?.model;
-    const reactMode = manifest.reactMode;
-    const scorer = manifest.skillGraph?.scorer;
-    const routing = manifest.skillGraph?.routing;
-    const continuity = manifest.skillGraph?.continuity;
-    const evidenceGate = manifest.evidenceGate;
-    return {
-        ...(provider !== undefined && { provider }),
-        ...(model !== undefined && { model }),
-        ...(isReactMode(reactMode) && { reactMode }),
-        ...(manifest.window !== undefined && { window: manifest.window }),
-        ...(scorer !== undefined && { scorer }),
-        ...(isPosture(routing) && { routing }),
-        ...(isContinuity(continuity) && { continuity }),
-        ...(isPosture(evidenceGate) && { evidenceGate }),
-        ...(memory !== undefined && { memory: { ...memory } }),
-    };
-}
-/**
- * A stable label for a set of facets — sorted `key=value`, so two runs of one
- * configuration produce the same string whatever order the fields were written
- * in. Useful as a grouping key for N recorded runs.
- */
-export function armLabel(facets) {
-    const parts = [];
-    for (const [key, value] of Object.entries(facets)) {
-        if (value === undefined)
-            continue;
-        if (key === 'memory') {
-            for (const [mk, mv] of Object.entries(value)) {
-                if (mv !== undefined)
-                    parts.push(`memory.${mk}=${String(mv)}`);
-            }
-            continue;
-        }
-        parts.push(`${key}=${String(value)}`);
-    }
-    return parts.sort().join(',');
-}
-/**
- * Compare one arm's DECLARED facets against a run's manifest. Empty result =
- * the run agreed with everything the arm claimed.
- *
- * Only declared facets are compared. An arm is a claim about the facets it
- * names and says nothing about the rest — checking undeclared fields would fail
- * every arm for differing from the manifest in ways it never claimed.
- */
-export function checkArmApplied(arm, manifest) {
-    const facets = arm.facets;
-    if (facets === undefined)
-        return [];
-    const out = [];
-    compare(out, 'provider', facets.provider, manifest.llm?.provider);
-    compare(out, 'model', facets.model, manifest.llm?.model);
-    compare(out, 'reactMode', facets.reactMode, manifest.reactMode);
-    compare(out, 'window', facets.window, manifest.window);
-    compare(out, 'skillGraph.scorer', facets.scorer, manifest.skillGraph?.scorer);
-    compare(out, 'skillGraph.routing', facets.routing, manifest.skillGraph?.routing);
-    compare(out, 'skillGraph.continuity', facets.continuity, manifest.skillGraph?.continuity);
-    compare(out, 'evidenceGate', facets.evidenceGate, manifest.evidenceGate);
-    const memory = facets.memory;
-    if (memory !== undefined) {
-        const rows = manifest.memories ?? [];
-        if (memory.id !== undefined) {
-            // Narrowed to ONE row: a missing row is a mismatch on the id itself, and
-            // the remaining fields are then compared against that row alone.
-            const row = rows.find((r) => r.id === memory.id);
-            if (row === undefined) {
-                out.push({
-                    facet: 'memory.id',
-                    declared: memory.id,
-                    ...(rows.length > 0 && { observed: rows.map((r) => r.id ?? '?').join(', ') }),
-                });
-            }
-            else {
-                for (const field of MEMORY_FIELDS) {
-                    compare(out, `memory.${field}`, memory[field], row[field]);
-                }
-            }
-        }
-        else if (!rows.some((row) => MEMORY_FIELDS.every((f) => matches(memory[f], row[f])))) {
-            // Un-narrowed: ANY mounted memory satisfying every declared field counts.
-            // Reported as ONE mismatch — naming a per-field winner across rows would
-            // invent a memory that does not exist.
-            out.push({
-                facet: 'memory',
-                declared: armLabel({ memory }),
-                ...(rows.length > 0 && {
-                    observed: rows.map((row) => armLabel({ memory: row })).join(' | '),
-                }),
-            });
-        }
-    }
-    return out;
-}
-/**
- * Which declared arm does this run belong to? The offline door: group N
- * recorded runs into arms by what each run's own manifest says.
- *
- * Rules, all conservative:
- * - only arms that DECLARE facets are candidates (an arm naming nothing matches
- *   every run and would swallow the whole study);
- * - a candidate matches when the manifest contradicts none of its facets;
- * - the MOST SPECIFIC match wins (most declared facets), and a tie returns
- *   `undefined` — two arms fitting one run equally well is an ambiguity, and
- *   guessing which the experimenter meant is exactly the bookkeeping this
- *   function replaces.
- */
-export function matchArm(arms, manifest) {
-    const fits = arms
-        .filter((arm) => declaredFacetCount(arm.facets) > 0)
-        .filter((arm) => checkArmApplied(arm, manifest).length === 0)
-        .map((arm) => ({ arm, specificity: declaredFacetCount(arm.facets) }))
-        .sort((a, b) => b.specificity - a.specificity);
-    if (fits.length === 0)
-        return undefined;
-    if (fits.length > 1 && fits[0].specificity === fits[1].specificity)
-        return undefined;
-    return fits[0].arm;
-}
-// ─── internals ───────────────────────────────────────────────────────
-const MEMORY_FIELDS = ['type', 'strategy', 'retrieval', 'embedderId', 'flavor'];
-function matches(declared, observed) {
-    return declared === undefined || declared === observed;
-}
-function compare(out, facet, declared, observed) {
-    if (matches(declared, observed))
-        return;
-    out.push({ facet, declared: declared, ...(observed !== undefined && { observed }) });
-}
-function isReactMode(value) {
-    return value === 'classic' || value === 'dynamic' || value === 'dynamic-grouped';
-}
-function isPosture(value) {
-    return value === 'assist' || value === 'guard' || value === 'rails';
-}
-function isContinuity(value) {
-    return value === 'turn' || value === 'conversation';
-}
-//# sourceMappingURL=manifest.js.map

package/dist/esm/lib/context-bisect/arms/manifest.js.map

@@ -1 +0,0 @@
-{"version":3,"file":"manifest.js","sourceRoot":"","sources":["../../../../../src/lib/context-bisect/arms/manifest.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;;;;;;;;;;;;GA6BG;AAGH,OAAO,EAAE,kBAAkB,EAAE,MAAM,eAAe,CAAC;AASnD,yDAAyD;AACzD,MAAM,CAAC,MAAM,oBAAoB,GAAG,qCAAqC,CAAC;AAE1E;;;;;;;GAOG;AACH,MAAM,UAAU,kBAAkB,CAChC,MAAgD;IAEhD,MAAM,KAAK,GAAG,MAAM,EAAE,IAAI,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,IAAI,KAAK,oBAAoB,CAAC,CAAC;IACnE,IAAI,KAAK,KAAK,SAAS;QAAE,OAAO,SAAS,CAAC;IAC1C,MAAM,OAAO,GAAG,KAAK,CAAC,OAAO,CAAC;IAC9B,6EAA6E;IAC7E,wEAAwE;IACxE,OAAO,OAAO,KAAK,IAAI,IAAI,OAAO,OAAO,KAAK,QAAQ,CAAC,CAAC,CAAE,OAA2B,CAAC,CAAC,CAAC,SAAS,CAAC;AACpG,CAAC;AAED;;;;;GAKG;AACH,MAAM,UAAU,qBAAqB,CAAC,QAAyB;IAC7D,MAAM,MAAM,GAAG,QAAQ,CAAC,QAAQ,EAAE,CAAC,CAAC,CAAC,CAAC;IACtC,MAAM,QAAQ,GAAG,QAAQ,CAAC,GAAG,EAAE,QAAQ,CAAC;IACxC,MAAM,KAAK,GAAG,QAAQ,CAAC,GAAG,EAAE,KAAK,CAAC;IAClC,MAAM,SAAS,GAAG,QAAQ,CAAC,SAAS,CAAC;IACrC,MAAM,MAAM,GAAG,QAAQ,CAAC,UAAU,EAAE,MAAM,CAAC;IAC3C,MAAM,OAAO,GAAG,QAAQ,CAAC,UAAU,EAAE,OAAO,CAAC;IAC7C,MAAM,UAAU,GAAG,QAAQ,CAAC,UAAU,EAAE,UAAU,CAAC;IACnD,MAAM,YAAY,GAAG,QAAQ,CAAC,YAAY,CAAC;IAC3C,OAAO;QACL,GAAG,CAAC,QAAQ,KAAK,SAAS,IAAI,EAAE,QAAQ,EAAE,CAAC;QAC3C,GAAG,CAAC,KAAK,KAAK,SAAS,IAAI,EAAE,KAAK,EAAE,CAAC;QACrC,GAAG,CAAC,WAAW,CAAC,SAAS,CAAC,IAAI,EAAE,SAAS,EAAE,CAAC;QAC5C,GAAG,CAAC,QAAQ,CAAC,MAAM,KAAK,SAAS,IAAI,EAAE,MAAM,EAAE,QAAQ,CAAC,MAAM,EAAE,CAAC;QACjE,GAAG,CAAC,MAAM,KAAK,SAAS,IAAI,EAAE,MAAM,EAAE,CAAC;QACvC,GAAG,CAAC,SAAS,CAAC,OAAO,CAAC,IAAI,EAAE,OAAO,EAAE,CAAC;QACtC,GAAG,CAAC,YAAY,CAAC,UAAU,CAAC,IAAI,EAAE,UAAU,EAAE,CAAC;QAC/C,GAAG,CAAC,SAAS,CAAC,YAAY,CAAC,IAAI,EAAE,YAAY,EAAE,CAAC;QAChD,GAAG,CAAC,MAAM,KAAK,SAAS,IAAI,EAAE,MAAM,EAAE,EAAE,GAAG,MAAM,EAAE,EAAE,CAAC;KACvD,CAAC;AACJ,CAAC;AAED;;;;GAIG;AACH,MAAM,UAAU,QAAQ,CAAC,MAAiB;IACxC,MAAM,KAAK,GAAa,EAAE,CAAC;IAC3B,KAAK,MAAM,CAAC,GAAG,EAAE,KAAK,CAAC,IAAI,MAAM,CAAC,OAAO,CAAC,MAAM,CAAC,EAAE,CAAC;QAClD,IAAI,KAAK,KAAK,SAAS;YAAE,SAAS;QAClC,IAAI,GAAG,KAAK,QAAQ,EAAE,CAAC;YACrB,KAAK,MAAM,CAAC,EAAE,EAAE,EAAE,CAAC,IAAI,MAAM,CAAC,OAAO,CAAC,KAA2B,CAAC,EAAE,CAAC;gBACnE,IAAI,EAAE,KAAK,SAAS;oBAAE,KAAK,CAAC,IAAI,CAAC,UAAU,EAAE,IAAI,MAAM,CAAC,EAAE,CAAC,EAAE,CAAC,CAAC;YACjE,CAAC;YACD,SAAS;QACX,CAAC;QACD,KAAK,CAAC,IAAI,CAAC,GAAG,GAAG,IAAI,MAAM,CAAC,KAAK,CAAC,EAAE,CAAC,CAAC;IACxC,CAAC;IACD,OAAO,KAAK,CAAC,IAAI,EAAE,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC;AAChC,CAAC;AAED;;;;;;;GAOG;AACH,MAAM,UAAU,eAAe,CAC7B,GAAgB,EAChB,QAAyB;IAEzB,MAAM,MAAM,GAAG,GAAG,CAAC,MAAM,CAAC;IAC1B,IAAI,MAAM,KAAK,SAAS;QAAE,OAAO,EAAE,CAAC;IACpC,MAAM,GAAG,GAAuB,EAAE,CAAC;IAEnC,OAAO,CAAC,GAAG,EAAE,UAAU,EAAE,MAAM,CAAC,QAAQ,EAAE,QAAQ,CAAC,GAAG,EAAE,QAAQ,CAAC,CAAC;IAClE,OAAO,CAAC,GAAG,EAAE,OAAO,EAAE,MAAM,CAAC,KAAK,EAAE,QAAQ,CAAC,GAAG,EAAE,KAAK,CAAC,CAAC;IACzD,OAAO,CAAC,GAAG,EAAE,WAAW,EAAE,MAAM,CAAC,SAAS,EAAE,QAAQ,CAAC,SAAS,CAAC,CAAC;IAChE,OAAO,CAAC,GAAG,EAAE,QAAQ,EAAE,MAAM,CAAC,MAAM,EAAE,QAAQ,CAAC,MAAM,CAAC,CAAC;IACvD,OAAO,CAAC,GAAG,EAAE,mBAAmB,EAAE,MAAM,CAAC,MAAM,EAAE,QAAQ,CAAC,UAAU,EAAE,MAAM,CAAC,CAAC;IAC9E,OAAO,CAAC,GAAG,EAAE,oBAAoB,EAAE,MAAM,CAAC,OAAO,EAAE,QAAQ,CAAC,UAAU,EAAE,OAAO,CAAC,CAAC;IACjF,OAAO,CAAC,GAAG,EAAE,uBAAuB,EAAE,MAAM,CAAC,UAAU,EAAE,QAAQ,CAAC,UAAU,EAAE,UAAU,CAAC,CAAC;IAC1F,OAAO,CAAC,GAAG,EAAE,cAAc,EAAE,MAAM,CAAC,YAAY,EAAE,QAAQ,CAAC,YAAY,CAAC,CAAC;IAEzE,MAAM,MAAM,GAAG,MAAM,CAAC,MAAM,CAAC;IAC7B,IAAI,MAAM,KAAK,SAAS,EAAE,CAAC;QACzB,MAAM,IAAI,GAAG,QAAQ,CAAC,QAAQ,IAAI,EAAE,CAAC;QACrC,IAAI,MAAM,CAAC,EAAE,KAAK,SAAS,EAAE,CAAC;YAC5B,yEAAyE;YACzE,iEAAiE;YACjE,MAAM,GAAG,GAAG,IAAI,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,EAAE,KAAK,MAAM,CAAC,EAAE,CAAC,CAAC;YACjD,IAAI,GAAG,KAAK,SAAS,EAAE,CAAC;gBACtB,GAAG,CAAC,IAAI,CAAC;oBACP,KAAK,EAAE,WAAW;oBAClB,QAAQ,EAAE,MAAM,CAAC,EAAE;oBACnB,GAAG,CAAC,IAAI,CAAC,MAAM,GAAG,CAAC,IAAI,EAAE,QAAQ,EAAE,IAAI,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,EAAE,IAAI,GAAG,CAAC,CAAC,IAAI,CAAC,IAAI,CAAC,EAAE,CAAC;iBAC9E,CAAC,CAAC;YACL,CAAC;iBAAM,CAAC;gBACN,KAAK,MAAM,KAAK,IAAI,aAAa,EAAE,CAAC;oBAClC,OAAO,CAAC,GAAG,EAAE,UAAU,KAAK,EAAE,EAAE,MAAM,CAAC,KAAK,CAAC,EAAE,GAAG,CAAC,KAAK,CAAC,CAAC,CAAC;gBAC7D,CAAC;YACH,CAAC;QACH,CAAC;aAAM,IAAI,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC,GAAG,EAAE,EAAE,CAAC,aAAa,CAAC,KAAK,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,OAAO,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,GAAG,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,EAAE,CAAC;YACvF,0EAA0E;YAC1E,yEAAyE;YACzE,uCAAuC;YACvC,GAAG,CAAC,IAAI,CAAC;gBACP,KAAK,EAAE,QAAQ;gBACf,QAAQ,EAAE,QAAQ,CAAC,EAAE,MAAM,EAAE,CAAC;gBAC9B,GAAG,CAAC,IAAI,CAAC,MAAM,GAAG,CAAC,IAAI;oBACrB,QAAQ,EAAE,IAAI,CAAC,GAAG,CAAC,CAAC,GAAG,EAAE,EAAE,CAAC,QAAQ,CAAC,EAAE,MAAM,EAAE,GAAG,EAAE,CAAC,CAAC,CAAC,IAAI,CAAC,KAAK,CAAC;iBACnE,CAAC;aACH,CAAC,CAAC;QACL,CAAC;IACH,CAAC;IACD,OAAO,GAAG,CAAC;AACb,CAAC;AAED;;;;;;;;;;;;GAYG;AACH,MAAM,UAAU,QAAQ,CACtB,IAA4B,EAC5B,QAAyB;IAEzB,MAAM,IAAI,GAAG,IAAI;SACd,MAAM,CAAC,CAAC,GAAG,EAAE,EAAE,CAAC,kBAAkB,CAAC,GAAG,CAAC,MAAM,CAAC,GAAG,CAAC,CAAC;SACnD,MAAM,CAAC,CAAC,GAAG,EAAE,EAAE,CAAC,eAAe,CAAC,GAAG,EAAE,QAAQ,CAAC,CAAC,MAAM,KAAK,CAAC,CAAC;SAC5D,GAAG,CAAC,CAAC,GAAG,EAAE,EAAE,CAAC,CAAC,EAAE,GAAG,EAAE,WAAW,EAAE,kBAAkB,CAAC,GAAG,CAAC,MAAM,CAAC,EAAE,CAAC,CAAC;SACpE,IAAI,CAAC,CAAC,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,WAAW,GAAG,CAAC,CAAC,WAAW,CAAC,CAAC;IACjD,IAAI,IAAI,CAAC,MAAM,KAAK,CAAC;QAAE,OAAO,SAAS,CAAC;IACxC,IAAI,IAAI,CAAC,MAAM,GAAG,CAAC,IAAI,IAAI,CAAC,CAAC,CAAC,CAAC,WAAW,KAAK,IAAI,CAAC,CAAC,CAAC,CAAC,WAAW;QAAE,OAAO,SAAS,CAAC;IACrF,OAAO,IAAI,CAAC,CAAC,CAAC,CAAC,GAAG,CAAC;AACrB,CAAC;AAED,wEAAwE;AAExE,MAAM,aAAa,GAAG,CAAC,MAAM,EAAE,UAAU,EAAE,WAAW,EAAE,YAAY,EAAE,QAAQ,CAAU,CAAC;AAEzF,SAAS,OAAO,CAAC,QAA4B,EAAE,QAA4B;IACzE,OAAO,QAAQ,KAAK,SAAS,IAAI,QAAQ,KAAK,QAAQ,CAAC;AACzD,CAAC;AAED,SAAS,OAAO,CACd,GAAuB,EACvB,KAAa,EACb,QAA4B,EAC5B,QAA4B;IAE5B,IAAI,OAAO,CAAC,QAAQ,EAAE,QAAQ,CAAC;QAAE,OAAO;IACxC,GAAG,CAAC,IAAI,CAAC,EAAE,KAAK,EAAE,QAAQ,EAAE,QAAkB,EAAE,GAAG,CAAC,QAAQ,KAAK,SAAS,IAAI,EAAE,QAAQ,EAAE,CAAC,EAAE,CAAC,CAAC;AACjG,CAAC;AAED,SAAS,WAAW,CAClB,KAAyB;IAEzB,OAAO,KAAK,KAAK,SAAS,IAAI,KAAK,KAAK,SAAS,IAAI,KAAK,KAAK,iBAAiB,CAAC;AACnF,CAAC;AAED,SAAS,SAAS,CAAC,KAAyB;IAC1C,OAAO,KAAK,KAAK,QAAQ,IAAI,KAAK,KAAK,OAAO,IAAI,KAAK,KAAK,OAAO,CAAC;AACtE,CAAC;AAED,SAAS,YAAY,CAAC,KAAyB;IAC7C,OAAO,KAAK,KAAK,MAAM,IAAI,KAAK,KAAK,cAAc,CAAC;AACtD,CAAC"}

package/dist/esm/lib/context-bisect/arms/probe.d.ts

@@ -1,54 +0,0 @@
-/**
- * The arm probe — seeded runs in, shared statistics out.
- *
- * Split in two on purpose. `collectArmRuns` only RUNS (it cannot know the
- * reference yet: when no prior answer was supplied, the reference IS the
- * baseline arm's first answer). `scoreArmRuns` only SCORES. Keeping them apart
- * is what lets the reference be chosen between the two halves without a second
- * pass over the model — the same "one ablation, two readouts" thrift
- * `runAblationProbe` applies to cost.
- *
- * Everything statistical is borrowed, not re-derived: `similarityStats`,
- * `costStatsFrom` and `resolveSamples` come from `ablation.ts` unchanged, so an
- * arm's `AblationRunStats` is the same number computed the same way as an
- * ablation's — which is the point of sharing the statistics while keeping the
- * spec types apart.
- */
-import type { AblationRunStats, Embedder, OutcomeComparator } from '../types.js';
-import type { ArmApplication, ArmRunner, NullBand, RunManifestLike, StrategyArm } from './types.js';
-/** The raw harvest of one arm's seeded runs — no scoring yet. */
-export interface ArmRuns {
-    readonly answers: readonly string[];
-    readonly loops: readonly number[];
-    readonly tokens: readonly number[];
-    readonly manifests: readonly RunManifestLike[];
-}
-/**
- * Call the consumer's runner once per seed (0..samples-1) for ONE arm and
- * normalize the two return shapes (`string` | `{output, cost?, manifest?}`) —
- * the same normalization `runAblationProbe` performs, so a bare-string runner
- * costs the caller nothing in ceremony.
- */
-export declare function collectArmRuns(arm: StrategyArm, runner: ArmRunner, samples: number): Promise<ArmRuns>;
-/**
- * Score one arm's harvested answers against the reference: similarity per seed
- * (mean/min/max/stdev) plus the outcome-flip count. Shape-identical to what an
- * ablation probe reports.
- */
-export declare function scoreArmRuns(runs: ArmRuns, reference: string, embedder: Embedder, outcomeChanged: OutcomeComparator): Promise<AblationRunStats>;
-/**
- * Build the inert-intervention control from the baseline arm's OWN seeded runs.
- * See {@link NullBand} for why leave-one-out does not transfer and this does.
- */
-export declare function nullBandFrom(baselineArmId: string, baselineStats: AblationRunStats, gates: boolean): NullBand;
-/**
- * Did this arm actually take effect? Compares every manifest the runner
- * reported against the arm's declared facets.
- *
- * `checked: false` — and therefore no refusal — in exactly two cases, both of
- * which are absence of evidence rather than evidence of absence: the runner
- * reported no manifest, or the arm declares only ablations (no manifest names a
- * tool catalog). Refusing a verdict in either case would punish a consumer for
- * a capture they were never required to make.
- */
-export declare function checkArmApplication(arm: StrategyArm, manifests: readonly RunManifestLike[]): ArmApplication;