@remnic/core 9.3.622 → 9.3.623

package/dist/{chunk-ZNCDQZIS.js → chunk-QFQQFX2H.js}

@@ -1,6 +1,6 @@
 import {
   runPostConsolidationMaterialize
-} from "./chunk-4MHHUPNH.js";
+} from "./chunk-UGEBPVNI.js";
 import {
   discoverMemoryExtensions,
   renderExtensionsBlock,
@@ -135,7 +135,7 @@ function parseOperatorAwareConsolidationResponse(response, cluster) {
   };
   const trimmed = response.trim();
   if (trimmed.length === 0) return fallback;
-  const fenced = /^```(?:json)?\s*([\s\S]*?)```\s*$/u.exec(trimmed);
+  const fenced = /^```(?:json)?([\s\S]*?)```$/u.exec(trimmed);
   const payload = fenced ? fenced[1].trim() : trimmed;
   const parsed = findLastJsonObjectWithOperator(payload);
   if (parsed === void 0) return fallback;
@@ -217,4 +217,4 @@ export {
   buildExtensionsBlockForConsolidation,
   materializeAfterSemanticConsolidation
 };
-//# sourceMappingURL=chunk-ZNCDQZIS.js.map
+//# sourceMappingURL=chunk-QFQQFX2H.js.map

package/dist/{chunk-ZNCDQZIS.js.map → chunk-QFQQFX2H.js.map}

@@ -1 +1 @@
-{"version":3,"sources":["../src/semantic-consolidation.ts"],"sourcesContent":["/**\n * semantic-consolidation.ts — Semantic Consolidation Engine\n *\n * Finds clusters of semantically similar memories using token overlap,\n * synthesizes canonical versions via LLM, and archives the originals.\n * Reduces memory store bloat while preserving all unique information.\n */\n\nimport type { MemoryFile, PluginConfig } from \"./types.js\";\nimport { normalizeRecallTokens, countRecallTokenOverlap } from \"./recall-tokenization.js\";\nimport { runPostConsolidationMaterialize } from \"./connectors/codex-materialize-runner.js\";\nimport type { MaterializeResult, RolloutSummaryInput } from \"./connectors/codex-materialize.js\";\nimport { discoverMemoryExtensions, renderExtensionsBlock, resolveExtensionsRoot } from \"./memory-extension-host/index.js\";\nimport { log } from \"./logger.js\";\n\n// Re-export resolveExtensionsRoot for backward compatibility — existing\n// consumers that import from semantic-consolidation.ts continue to work.\nexport { resolveExtensionsRoot } from \"./memory-extension-host/index.js\";\n\n// Operator vocabulary (issue #561).  The types and validators live in a\n// standalone `consolidation-operator.ts` module so `storage.ts` can import\n// them without creating a `storage → semantic-consolidation →\n// codex-materialize-runner → storage` cycle.  Re-exported here so existing\n// consumers that reach for `./semantic-consolidation.js` keep working.\nexport {\n  CONSOLIDATION_OPERATORS,\n  isConsolidationOperator,\n  isSemanticConsolidationLlmOperator,\n  isValidDerivedFromEntry,\n  type ConsolidationOperator,\n  type SemanticConsolidationLlmOperator,\n} from \"./consolidation-operator.js\";\n\nimport {\n  CONSOLIDATION_OPERATORS as _CONSOLIDATION_OPERATORS,\n  isConsolidationOperator as _isConsolidationOperator,\n  isSemanticConsolidationLlmOperator as _isSemanticConsolidationLlmOperator,\n  type ConsolidationOperator as _ConsolidationOperator,\n  type SemanticConsolidationLlmOperator as _SemanticConsolidationLlmOperator,\n} from \"./consolidation-operator.js\";\n\nexport interface ConsolidationCluster {\n  category: string;\n  memories: MemoryFile[];\n  overlapScore: number;\n  canonicalContent?: string;\n}\n\nexport interface SemanticConsolidationResult {\n  clustersFound: number;\n  memoriesConsolidated: number;\n  memoriesArchived: number;\n  errors: number;\n  clusters: ConsolidationCluster[];\n}\n\n/**\n * Find clusters of semantically similar memories using token overlap.\n */\nexport function findSimilarClusters(\n  memories: MemoryFile[],\n  config: {\n    threshold: number;\n    minClusterSize: number;\n    excludeCategories: string[];\n    maxPerRun: number;\n  },\n): ConsolidationCluster[] {\n  const excluded = new Set(config.excludeCategories);\n\n  // Group by category first\n  const byCategory = new Map<string, MemoryFile[]>();\n  for (const m of memories) {\n    const cat = m.frontmatter.category;\n    if (excluded.has(cat)) continue;\n    if (m.frontmatter.status && m.frontmatter.status !== \"active\") continue;\n    const list = byCategory.get(cat) ?? [];\n    list.push(m);\n    byCategory.set(cat, list);\n  }\n\n  const clusters: ConsolidationCluster[] = [];\n  let totalCandidates = 0;\n\n  for (const [category, mems] of byCategory) {\n    if (totalCandidates >= config.maxPerRun) break;\n\n    // Token-normalize all memories in this category\n    const tokenized = mems.map((m) => ({\n      memory: m,\n      tokens: new Set(normalizeRecallTokens(m.content, [])),\n    }));\n\n    // Track which memories are already clustered\n    const clustered = new Set<string>();\n\n    for (let i = 0; i < tokenized.length && totalCandidates < config.maxPerRun; i++) {\n      const remainingBudget = config.maxPerRun - totalCandidates;\n      if (remainingBudget < config.minClusterSize) break;\n      if (clustered.has(tokenized[i].memory.frontmatter.id)) continue;\n\n      const cluster: MemoryFile[] = [tokenized[i].memory];\n      let totalOverlap = 0;\n      let comparisons = 0;\n\n      for (let j = i + 1; j < tokenized.length && cluster.length < remainingBudget; j++) {\n        if (clustered.has(tokenized[j].memory.frontmatter.id)) continue;\n\n        const aTokens = tokenized[i].tokens;\n        const bTokens = tokenized[j].tokens;\n        if (aTokens.size === 0 || bTokens.size === 0) continue;\n\n        // Bidirectional overlap: what fraction of tokens are shared\n        const overlap = countRecallTokenOverlap(aTokens, [...bTokens].join(\" \"));\n        const maxTokens = Math.max(aTokens.size, bTokens.size);\n        const score = maxTokens > 0 ? overlap / maxTokens : 0;\n\n        if (score >= config.threshold) {\n          cluster.push(tokenized[j].memory);\n          totalOverlap += score;\n          comparisons++;\n        }\n      }\n\n      if (cluster.length >= config.minClusterSize) {\n        for (const m of cluster) clustered.add(m.frontmatter.id);\n        clusters.push({\n          category,\n          memories: cluster,\n          overlapScore: comparisons > 0 ? totalOverlap / comparisons : 0,\n        });\n        totalCandidates += cluster.length;\n      }\n    }\n  }\n\n  return clusters;\n}\n\n/**\n * Build the LLM prompt for synthesizing a canonical memory from a cluster.\n */\nexport function buildConsolidationPrompt(cluster: ConsolidationCluster): string {\n  const memoryTexts = cluster.memories\n    .map(\n      (m, i) =>\n        `Memory ${i + 1} (${m.frontmatter.id}, created ${m.frontmatter.created}):\\n${m.content}`,\n    )\n    .join(\"\\n\\n\");\n\n  return `You are a memory consolidation system. The following ${cluster.memories.length} memories in the \"${cluster.category}\" category contain overlapping information.\n\nSynthesize them into ONE canonical memory that:\n1. Preserves ALL unique information from every source memory\n2. Removes redundancy and repetition\n3. Uses clear, concise language\n4. Maintains the same category and tone\n5. Does NOT add information that isn't in the sources\n\n${memoryTexts}\n\nWrite ONLY the consolidated memory content (no metadata, no explanation, no preamble):`;\n}\n\n/**\n * Parse the LLM response to extract the canonical content.\n */\nexport function parseConsolidationResponse(response: string): string {\n  return response.trim();\n}\n\n// ─── Operator-aware prompt / parse (issue #561 PR 3) ─────────────────────────\n\n/**\n * Structured result from an operator-aware consolidation LLM call.\n *\n * - `operator` — the consolidation operator the LLM chose for this cluster.\n *   Falls back to the heuristic default when the LLM omits or returns an\n *   unknown value (the parser never surfaces an invalid operator; see\n *   `parseOperatorAwareConsolidationResponse`).\n * - `output` — the canonical content (same format the legacy prompt\n *   returns).  Callers persist this as the body of the new memory.\n */\nexport interface OperatorAwareConsolidationResult {\n  // Restricted to the LLM-allowed subset (Cursor Bugbot, PR #730):\n  // `pattern-reinforcement` is reserved for the maintenance job and\n  // must never reach this struct from a consolidation LLM response.\n  operator: _SemanticConsolidationLlmOperator;\n  output: string;\n}\n\n/**\n * Heuristic default operator for a cluster.  Used as the fallback when the\n * LLM does not return a parseable operator, and as the \"floor\" decision in\n * `parseOperatorAwareConsolidationResponse`.\n *\n * Current heuristic (kept deliberately conservative — PR 3 only):\n *\n *   - Two or more memories being collapsed into one canonical blob is a\n *     MERGE by definition.  This is the path the current clustering\n *     pipeline exercises.\n *   - A cluster of size 1 that still reaches consolidation (future path,\n *     e.g. supersession of a single older memory by a newer value) is an\n *     UPDATE.\n *   - SPLIT is never selected by the heuristic because the current write\n *     path emits exactly one canonical memory per cluster.  The prompt\n *     reserves SPLIT for future cluster shapes where the LLM decides one\n *     logical source actually encodes several distinct facts — at which\n *     point the orchestrator would need to write multiple outputs.\n */\nexport function chooseConsolidationOperator(\n  cluster: ConsolidationCluster,\n): _SemanticConsolidationLlmOperator {\n  if (cluster.memories.length <= 1) return \"update\";\n  return \"merge\";\n}\n\n/**\n * Build the operator-aware LLM prompt.  The LLM is asked to return a\n * JSON object `{ \"operator\": <split|merge|update>, \"output\": <content> }`.\n *\n * The prompt is additive: it still asks for a single canonical blob under\n * `output`, so the upstream write path does not change.  Future expansions\n * (SPLIT emitting multiple outputs) are explicitly documented as\n * out-of-scope for the parser — `parseOperatorAwareConsolidationResponse`\n * collapses any SPLIT response into a single canonical output for now.\n */\nexport function buildOperatorAwareConsolidationPrompt(\n  cluster: ConsolidationCluster,\n): string {\n  const memoryTexts = cluster.memories\n    .map(\n      (m, i) =>\n        `Memory ${i + 1} (${m.frontmatter.id}, created ${m.frontmatter.created}):\\n${m.content}`,\n    )\n    .join(\"\\n\\n\");\n\n  return `You are a memory consolidation system.  The following ${cluster.memories.length} memories in the \"${cluster.category}\" category contain overlapping information.\n\nPick exactly ONE consolidation operator for this cluster and return a JSON object.\n\nOperator vocabulary:\n  - \"merge\"  — multiple distinct source memories overlap and should be collapsed into one canonical memory (most common).\n  - \"update\" — one source memory carries a stale value that a newer source supersedes within the same logical fact.\n  - \"split\"  — a single logical source really encodes multiple distinct facts that should be separated (rare; if you pick split, still emit ONE canonical body — the write path will chunk it later).\n\nOutput JSON ONLY, no prose before or after.  The \"operator\" key MUST be set to exactly one of the three strings \"merge\", \"update\", or \"split\" — never a pipe-separated placeholder like \"merge|update|split\".  Example shape:\n  {\n    \"operator\": \"merge\",\n    \"output\": \"<the canonical memory text>\"\n  }\n\nThe \"output\" value must:\n1. Preserve ALL unique information from every source memory\n2. Remove redundancy and repetition\n3. Use clear, concise language\n4. Match the \"${cluster.category}\" category and tone\n5. NOT add information that isn't in the sources\n\n${memoryTexts}\n\nReturn ONLY the JSON object:`;\n}\n\n/**\n * Parse an operator-aware consolidation response.\n *\n * Contract:\n *   - Accepts strict JSON `{ \"operator\": \"...\", \"output\": \"...\" }`.\n *   - Tolerates a JSON payload wrapped in a fenced code block (```json ...```).\n *   - Falls back to the heuristic operator when the JSON is malformed,\n *     the `operator` field is missing / unknown, or the raw response is\n *     a plain blob with no JSON at all.  This keeps PR 3 backwards\n *     compatible with older models that ignore the JSON instruction.\n *   - Never throws.  A missing / empty `output` field falls back to the\n *     trimmed raw response so the caller still writes something rather\n *     than dropping the cluster.\n */\nexport function parseOperatorAwareConsolidationResponse(\n  response: string,\n  cluster: ConsolidationCluster,\n): OperatorAwareConsolidationResult {\n  const fallback: OperatorAwareConsolidationResult = {\n    operator: chooseConsolidationOperator(cluster),\n    output: response.trim(),\n  };\n\n  const trimmed = response.trim();\n  if (trimmed.length === 0) return fallback;\n\n  // Strip a fenced code block if present.\n  const fenced = /^```(?:json)?\\s*([\\s\\S]*?)```\\s*$/u.exec(trimmed);\n  const payload = fenced ? fenced[1].trim() : trimmed;\n\n  // Find a balanced brace-delimited JSON object that has an `operator`\n  // key (PR #632 round-4 review, codex P1).  A first/last-brace slice\n  // breaks when the model prepends an earlier brace block (e.g.\n  // `Example: {\"note\":\"...\"} ... {\"operator\":\"merge\",...}`).  Walk the\n  // payload tracking nesting + string/escape state and skip past\n  // objects that don't look like our target shape.\n  const parsed = findLastJsonObjectWithOperator(payload);\n  if (parsed === undefined) return fallback;\n  if (typeof parsed !== \"object\" || parsed === null) return fallback;\n\n  const obj = parsed as Record<string, unknown>;\n  const rawOperator = typeof obj.operator === \"string\" ? obj.operator.trim().toLowerCase() : \"\";\n  const rawOutput = typeof obj.output === \"string\" ? obj.output : \"\";\n\n  // Narrow gate (Cursor Bugbot review on PR #730 head `aa1c2a8`):\n  // accept ONLY the legacy split/merge/update LLM vocabulary here.\n  // `pattern-reinforcement` joined the broader `ConsolidationOperator`\n  // type in #687 PR 2/4 but is reserved for the maintenance job — if\n  // an LLM hallucinates that operator we must NOT promote it onto\n  // `derived_via`.\n  const operator = _isSemanticConsolidationLlmOperator(rawOperator)\n    ? rawOperator\n    : chooseConsolidationOperator(cluster);\n  const output = rawOutput.trim().length > 0 ? rawOutput.trim() : response.trim();\n\n  return { operator, output };\n}\n\n/**\n * Walk `text`, find all balanced top-level `{ ... }` blocks whose\n * JSON.parse result is an object with an `operator` key, and return\n * the LAST one (function name reflects this — PR #632 review,\n * cursor Low).  Returns `undefined` when nothing matches.  Tracks\n * string state + escape sequences so braces inside string values\n * don't throw off the depth counter.\n *\n * Used by `parseOperatorAwareConsolidationResponse` to tolerate models\n * that prepend an explanatory JSON example block before the real\n * payload (PR #632 round-4 + round-5 review, codex P1).  We take the\n * LAST candidate so that an instructional example with an `operator`\n * key ahead of the real answer doesn't steal precedence — models\n * typically write the example first and the real answer last.\n */\nfunction findLastJsonObjectWithOperator(text: string): unknown {\n  let searchFrom = 0;\n  let last: unknown = undefined;\n  while (searchFrom < text.length) {\n    const start = text.indexOf(\"{\", searchFrom);\n    if (start < 0) return last;\n    let depth = 0;\n    let inString = false;\n    let escape = false;\n    let closed = false;\n    let endIdx = -1;\n    for (let i = start; i < text.length; i++) {\n      const ch = text[i];\n      if (inString) {\n        if (escape) {\n          escape = false;\n        } else if (ch === \"\\\\\") {\n          escape = true;\n        } else if (ch === '\"') {\n          inString = false;\n        }\n        continue;\n      }\n      if (ch === '\"') {\n        inString = true;\n      } else if (ch === \"{\") {\n        depth += 1;\n      } else if (ch === \"}\") {\n        depth -= 1;\n        if (depth === 0) {\n          closed = true;\n          endIdx = i;\n          break;\n        }\n      }\n    }\n    if (!closed) return last;\n    const slice = text.slice(start, endIdx + 1);\n    try {\n      const parsed = JSON.parse(slice);\n      if (\n        typeof parsed === \"object\" &&\n        parsed !== null &&\n        \"operator\" in (parsed as Record<string, unknown>)\n      ) {\n        last = parsed;\n      }\n    } catch {\n      // Fall through to the next candidate.\n    }\n    searchFrom = endIdx + 1;\n  }\n  return last;\n}\n\n// Silence unused-import warnings when tsup tree-shakes: these are used\n// above in chooseConsolidationOperator + parse helpers.\nvoid _CONSOLIDATION_OPERATORS;\n\n/**\n * Discover extensions and build the block to append to a consolidation prompt.\n * Returns \"\" when extensions are disabled or none are found.\n */\nexport async function buildExtensionsBlockForConsolidation(\n  config: PluginConfig,\n): Promise<string> {\n  if (!config.memoryExtensionsEnabled) return \"\";\n  const root = resolveExtensionsRoot(config);\n  const extensions = await discoverMemoryExtensions(root, log);\n  if (extensions.length === 0) return \"\";\n  return renderExtensionsBlock(extensions);\n}\n\n/**\n * Optional post-consolidation hook — materializes the namespace into Codex's\n * native memory layout when the consolidation run finishes. Kept here (rather\n * than in orchestrator.ts) so #378 doesn't conflict with Wave 1 edits.\n *\n * Safe to call regardless of config state: honors `codexMaterializeMemories`\n * and `codexMaterializeOnConsolidation` and silently becomes a no-op when\n * either is disabled.\n */\nexport async function materializeAfterSemanticConsolidation(options: {\n  config: PluginConfig;\n  namespace?: string;\n  memories?: MemoryFile[];\n  memoryDir?: string;\n  codexHome?: string;\n  rolloutSummaries?: RolloutSummaryInput[];\n  now?: Date;\n}): Promise<MaterializeResult | null> {\n  // Delegates to the shared post-consolidation helper so semantic and causal\n  // flows stay in lock-step — any guard/logging change happens in one place.\n  return runPostConsolidationMaterialize(\"[semantic-consolidation]\", options);\n}\n"],"mappings":";;;;;;;;;;;;;;;;;;;;AA2DO,SAAS,oBACd,UACA,QAMwB;AACxB,QAAM,WAAW,IAAI,IAAI,OAAO,iBAAiB;AAGjD,QAAM,aAAa,oBAAI,IAA0B;AACjD,aAAW,KAAK,UAAU;AACxB,UAAM,MAAM,EAAE,YAAY;AAC1B,QAAI,SAAS,IAAI,GAAG,EAAG;AACvB,QAAI,EAAE,YAAY,UAAU,EAAE,YAAY,WAAW,SAAU;AAC/D,UAAM,OAAO,WAAW,IAAI,GAAG,KAAK,CAAC;AACrC,SAAK,KAAK,CAAC;AACX,eAAW,IAAI,KAAK,IAAI;AAAA,EAC1B;AAEA,QAAM,WAAmC,CAAC;AAC1C,MAAI,kBAAkB;AAEtB,aAAW,CAAC,UAAU,IAAI,KAAK,YAAY;AACzC,QAAI,mBAAmB,OAAO,UAAW;AAGzC,UAAM,YAAY,KAAK,IAAI,CAAC,OAAO;AAAA,MACjC,QAAQ;AAAA,MACR,QAAQ,IAAI,IAAI,sBAAsB,EAAE,SAAS,CAAC,CAAC,CAAC;AAAA,IACtD,EAAE;AAGF,UAAM,YAAY,oBAAI,IAAY;AAElC,aAAS,IAAI,GAAG,IAAI,UAAU,UAAU,kBAAkB,OAAO,WAAW,KAAK;AAC/E,YAAM,kBAAkB,OAAO,YAAY;AAC3C,UAAI,kBAAkB,OAAO,eAAgB;AAC7C,UAAI,UAAU,IAAI,UAAU,CAAC,EAAE,OAAO,YAAY,EAAE,EAAG;AAEvD,YAAM,UAAwB,CAAC,UAAU,CAAC,EAAE,MAAM;AAClD,UAAI,eAAe;AACnB,UAAI,cAAc;AAElB,eAAS,IAAI,IAAI,GAAG,IAAI,UAAU,UAAU,QAAQ,SAAS,iBAAiB,KAAK;AACjF,YAAI,UAAU,IAAI,UAAU,CAAC,EAAE,OAAO,YAAY,EAAE,EAAG;AAEvD,cAAM,UAAU,UAAU,CAAC,EAAE;AAC7B,cAAM,UAAU,UAAU,CAAC,EAAE;AAC7B,YAAI,QAAQ,SAAS,KAAK,QAAQ,SAAS,EAAG;AAG9C,cAAM,UAAU,wBAAwB,SAAS,CAAC,GAAG,OAAO,EAAE,KAAK,GAAG,CAAC;AACvE,cAAM,YAAY,KAAK,IAAI,QAAQ,MAAM,QAAQ,IAAI;AACrD,cAAM,QAAQ,YAAY,IAAI,UAAU,YAAY;AAEpD,YAAI,SAAS,OAAO,WAAW;AAC7B,kBAAQ,KAAK,UAAU,CAAC,EAAE,MAAM;AAChC,0BAAgB;AAChB;AAAA,QACF;AAAA,MACF;AAEA,UAAI,QAAQ,UAAU,OAAO,gBAAgB;AAC3C,mBAAW,KAAK,QAAS,WAAU,IAAI,EAAE,YAAY,EAAE;AACvD,iBAAS,KAAK;AAAA,UACZ;AAAA,UACA,UAAU;AAAA,UACV,cAAc,cAAc,IAAI,eAAe,cAAc;AAAA,QAC/D,CAAC;AACD,2BAAmB,QAAQ;AAAA,MAC7B;AAAA,IACF;AAAA,EACF;AAEA,SAAO;AACT;AAKO,SAAS,yBAAyB,SAAuC;AAC9E,QAAM,cAAc,QAAQ,SACzB;AAAA,IACC,CAAC,GAAG,MACF,UAAU,IAAI,CAAC,KAAK,EAAE,YAAY,EAAE,aAAa,EAAE,YAAY,OAAO;AAAA,EAAO,EAAE,OAAO;AAAA,EAC1F,EACC,KAAK,MAAM;AAEd,SAAO,wDAAwD,QAAQ,SAAS,MAAM,qBAAqB,QAAQ,QAAQ;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA,EAS3H,WAAW;AAAA;AAAA;AAGb;AAKO,SAAS,2BAA2B,UAA0B;AACnE,SAAO,SAAS,KAAK;AACvB;AAyCO,SAAS,4BACd,SACmC;AACnC,MAAI,QAAQ,SAAS,UAAU,EAAG,QAAO;AACzC,SAAO;AACT;AAYO,SAAS,sCACd,SACQ;AACR,QAAM,cAAc,QAAQ,SACzB;AAAA,IACC,CAAC,GAAG,MACF,UAAU,IAAI,CAAC,KAAK,EAAE,YAAY,EAAE,aAAa,EAAE,YAAY,OAAO;AAAA,EAAO,EAAE,OAAO;AAAA,EAC1F,EACC,KAAK,MAAM;AAEd,SAAO,yDAAyD,QAAQ,SAAS,MAAM,qBAAqB,QAAQ,QAAQ;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA,gBAmB9G,QAAQ,QAAQ;AAAA;AAAA;AAAA,EAG9B,WAAW;AAAA;AAAA;AAGb;AAgBO,SAAS,wCACd,UACA,SACkC;AAClC,QAAM,WAA6C;AAAA,IACjD,UAAU,4BAA4B,OAAO;AAAA,IAC7C,QAAQ,SAAS,KAAK;AAAA,EACxB;AAEA,QAAM,UAAU,SAAS,KAAK;AAC9B,MAAI,QAAQ,WAAW,EAAG,QAAO;AAGjC,QAAM,SAAS,qCAAqC,KAAK,OAAO;AAChE,QAAM,UAAU,SAAS,OAAO,CAAC,EAAE,KAAK,IAAI;AAQ5C,QAAM,SAAS,+BAA+B,OAAO;AACrD,MAAI,WAAW,OAAW,QAAO;AACjC,MAAI,OAAO,WAAW,YAAY,WAAW,KAAM,QAAO;AAE1D,QAAM,MAAM;AACZ,QAAM,cAAc,OAAO,IAAI,aAAa,WAAW,IAAI,SAAS,KAAK,EAAE,YAAY,IAAI;AAC3F,QAAM,YAAY,OAAO,IAAI,WAAW,WAAW,IAAI,SAAS;AAQhE,QAAM,WAAW,mCAAoC,WAAW,IAC5D,cACA,4BAA4B,OAAO;AACvC,QAAM,SAAS,UAAU,KAAK,EAAE,SAAS,IAAI,UAAU,KAAK,IAAI,SAAS,KAAK;AAE9E,SAAO,EAAE,UAAU,OAAO;AAC5B;AAiBA,SAAS,+BAA+B,MAAuB;AAC7D,MAAI,aAAa;AACjB,MAAI,OAAgB;AACpB,SAAO,aAAa,KAAK,QAAQ;AAC/B,UAAM,QAAQ,KAAK,QAAQ,KAAK,UAAU;AAC1C,QAAI,QAAQ,EAAG,QAAO;AACtB,QAAI,QAAQ;AACZ,QAAI,WAAW;AACf,QAAI,SAAS;AACb,QAAI,SAAS;AACb,QAAI,SAAS;AACb,aAAS,IAAI,OAAO,IAAI,KAAK,QAAQ,KAAK;AACxC,YAAM,KAAK,KAAK,CAAC;AACjB,UAAI,UAAU;AACZ,YAAI,QAAQ;AACV,mBAAS;AAAA,QACX,WAAW,OAAO,MAAM;AACtB,mBAAS;AAAA,QACX,WAAW,OAAO,KAAK;AACrB,qBAAW;AAAA,QACb;AACA;AAAA,MACF;AACA,UAAI,OAAO,KAAK;AACd,mBAAW;AAAA,MACb,WAAW,OAAO,KAAK;AACrB,iBAAS;AAAA,MACX,WAAW,OAAO,KAAK;AACrB,iBAAS;AACT,YAAI,UAAU,GAAG;AACf,mBAAS;AACT,mBAAS;AACT;AAAA,QACF;AAAA,MACF;AAAA,IACF;AACA,QAAI,CAAC,OAAQ,QAAO;AACpB,UAAM,QAAQ,KAAK,MAAM,OAAO,SAAS,CAAC;AAC1C,QAAI;AACF,YAAM,SAAS,KAAK,MAAM,KAAK;AAC/B,UACE,OAAO,WAAW,YAClB,WAAW,QACX,cAAe,QACf;AACA,eAAO;AAAA,MACT;AAAA,IACF,QAAQ;AAAA,IAER;AACA,iBAAa,SAAS;AAAA,EACxB;AACA,SAAO;AACT;AAUA,eAAsB,qCACpB,QACiB;AACjB,MAAI,CAAC,OAAO,wBAAyB,QAAO;AAC5C,QAAM,OAAO,sBAAsB,MAAM;AACzC,QAAM,aAAa,MAAM,yBAAyB,MAAM,GAAG;AAC3D,MAAI,WAAW,WAAW,EAAG,QAAO;AACpC,SAAO,sBAAsB,UAAU;AACzC;AAWA,eAAsB,sCAAsC,SAQtB;AAGpC,SAAO,gCAAgC,4BAA4B,OAAO;AAC5E;","names":[]}
+{"version":3,"sources":["../src/semantic-consolidation.ts"],"sourcesContent":["/**\n * semantic-consolidation.ts — Semantic Consolidation Engine\n *\n * Finds clusters of semantically similar memories using token overlap,\n * synthesizes canonical versions via LLM, and archives the originals.\n * Reduces memory store bloat while preserving all unique information.\n */\n\nimport type { MemoryFile, PluginConfig } from \"./types.js\";\nimport { normalizeRecallTokens, countRecallTokenOverlap } from \"./recall-tokenization.js\";\nimport { runPostConsolidationMaterialize } from \"./connectors/codex-materialize-runner.js\";\nimport type { MaterializeResult, RolloutSummaryInput } from \"./connectors/codex-materialize.js\";\nimport { discoverMemoryExtensions, renderExtensionsBlock, resolveExtensionsRoot } from \"./memory-extension-host/index.js\";\nimport { log } from \"./logger.js\";\n\n// Re-export resolveExtensionsRoot for backward compatibility — existing\n// consumers that import from semantic-consolidation.ts continue to work.\nexport { resolveExtensionsRoot } from \"./memory-extension-host/index.js\";\n\n// Operator vocabulary (issue #561).  The types and validators live in a\n// standalone `consolidation-operator.ts` module so `storage.ts` can import\n// them without creating a `storage → semantic-consolidation →\n// codex-materialize-runner → storage` cycle.  Re-exported here so existing\n// consumers that reach for `./semantic-consolidation.js` keep working.\nexport {\n  CONSOLIDATION_OPERATORS,\n  isConsolidationOperator,\n  isSemanticConsolidationLlmOperator,\n  isValidDerivedFromEntry,\n  type ConsolidationOperator,\n  type SemanticConsolidationLlmOperator,\n} from \"./consolidation-operator.js\";\n\nimport {\n  CONSOLIDATION_OPERATORS as _CONSOLIDATION_OPERATORS,\n  isConsolidationOperator as _isConsolidationOperator,\n  isSemanticConsolidationLlmOperator as _isSemanticConsolidationLlmOperator,\n  type ConsolidationOperator as _ConsolidationOperator,\n  type SemanticConsolidationLlmOperator as _SemanticConsolidationLlmOperator,\n} from \"./consolidation-operator.js\";\n\nexport interface ConsolidationCluster {\n  category: string;\n  memories: MemoryFile[];\n  overlapScore: number;\n  canonicalContent?: string;\n}\n\nexport interface SemanticConsolidationResult {\n  clustersFound: number;\n  memoriesConsolidated: number;\n  memoriesArchived: number;\n  errors: number;\n  clusters: ConsolidationCluster[];\n}\n\n/**\n * Find clusters of semantically similar memories using token overlap.\n */\nexport function findSimilarClusters(\n  memories: MemoryFile[],\n  config: {\n    threshold: number;\n    minClusterSize: number;\n    excludeCategories: string[];\n    maxPerRun: number;\n  },\n): ConsolidationCluster[] {\n  const excluded = new Set(config.excludeCategories);\n\n  // Group by category first\n  const byCategory = new Map<string, MemoryFile[]>();\n  for (const m of memories) {\n    const cat = m.frontmatter.category;\n    if (excluded.has(cat)) continue;\n    if (m.frontmatter.status && m.frontmatter.status !== \"active\") continue;\n    const list = byCategory.get(cat) ?? [];\n    list.push(m);\n    byCategory.set(cat, list);\n  }\n\n  const clusters: ConsolidationCluster[] = [];\n  let totalCandidates = 0;\n\n  for (const [category, mems] of byCategory) {\n    if (totalCandidates >= config.maxPerRun) break;\n\n    // Token-normalize all memories in this category\n    const tokenized = mems.map((m) => ({\n      memory: m,\n      tokens: new Set(normalizeRecallTokens(m.content, [])),\n    }));\n\n    // Track which memories are already clustered\n    const clustered = new Set<string>();\n\n    for (let i = 0; i < tokenized.length && totalCandidates < config.maxPerRun; i++) {\n      const remainingBudget = config.maxPerRun - totalCandidates;\n      if (remainingBudget < config.minClusterSize) break;\n      if (clustered.has(tokenized[i].memory.frontmatter.id)) continue;\n\n      const cluster: MemoryFile[] = [tokenized[i].memory];\n      let totalOverlap = 0;\n      let comparisons = 0;\n\n      for (let j = i + 1; j < tokenized.length && cluster.length < remainingBudget; j++) {\n        if (clustered.has(tokenized[j].memory.frontmatter.id)) continue;\n\n        const aTokens = tokenized[i].tokens;\n        const bTokens = tokenized[j].tokens;\n        if (aTokens.size === 0 || bTokens.size === 0) continue;\n\n        // Bidirectional overlap: what fraction of tokens are shared\n        const overlap = countRecallTokenOverlap(aTokens, [...bTokens].join(\" \"));\n        const maxTokens = Math.max(aTokens.size, bTokens.size);\n        const score = maxTokens > 0 ? overlap / maxTokens : 0;\n\n        if (score >= config.threshold) {\n          cluster.push(tokenized[j].memory);\n          totalOverlap += score;\n          comparisons++;\n        }\n      }\n\n      if (cluster.length >= config.minClusterSize) {\n        for (const m of cluster) clustered.add(m.frontmatter.id);\n        clusters.push({\n          category,\n          memories: cluster,\n          overlapScore: comparisons > 0 ? totalOverlap / comparisons : 0,\n        });\n        totalCandidates += cluster.length;\n      }\n    }\n  }\n\n  return clusters;\n}\n\n/**\n * Build the LLM prompt for synthesizing a canonical memory from a cluster.\n */\nexport function buildConsolidationPrompt(cluster: ConsolidationCluster): string {\n  const memoryTexts = cluster.memories\n    .map(\n      (m, i) =>\n        `Memory ${i + 1} (${m.frontmatter.id}, created ${m.frontmatter.created}):\\n${m.content}`,\n    )\n    .join(\"\\n\\n\");\n\n  return `You are a memory consolidation system. The following ${cluster.memories.length} memories in the \"${cluster.category}\" category contain overlapping information.\n\nSynthesize them into ONE canonical memory that:\n1. Preserves ALL unique information from every source memory\n2. Removes redundancy and repetition\n3. Uses clear, concise language\n4. Maintains the same category and tone\n5. Does NOT add information that isn't in the sources\n\n${memoryTexts}\n\nWrite ONLY the consolidated memory content (no metadata, no explanation, no preamble):`;\n}\n\n/**\n * Parse the LLM response to extract the canonical content.\n */\nexport function parseConsolidationResponse(response: string): string {\n  return response.trim();\n}\n\n// ─── Operator-aware prompt / parse (issue #561 PR 3) ─────────────────────────\n\n/**\n * Structured result from an operator-aware consolidation LLM call.\n *\n * - `operator` — the consolidation operator the LLM chose for this cluster.\n *   Falls back to the heuristic default when the LLM omits or returns an\n *   unknown value (the parser never surfaces an invalid operator; see\n *   `parseOperatorAwareConsolidationResponse`).\n * - `output` — the canonical content (same format the legacy prompt\n *   returns).  Callers persist this as the body of the new memory.\n */\nexport interface OperatorAwareConsolidationResult {\n  // Restricted to the LLM-allowed subset (Cursor Bugbot, PR #730):\n  // `pattern-reinforcement` is reserved for the maintenance job and\n  // must never reach this struct from a consolidation LLM response.\n  operator: _SemanticConsolidationLlmOperator;\n  output: string;\n}\n\n/**\n * Heuristic default operator for a cluster.  Used as the fallback when the\n * LLM does not return a parseable operator, and as the \"floor\" decision in\n * `parseOperatorAwareConsolidationResponse`.\n *\n * Current heuristic (kept deliberately conservative — PR 3 only):\n *\n *   - Two or more memories being collapsed into one canonical blob is a\n *     MERGE by definition.  This is the path the current clustering\n *     pipeline exercises.\n *   - A cluster of size 1 that still reaches consolidation (future path,\n *     e.g. supersession of a single older memory by a newer value) is an\n *     UPDATE.\n *   - SPLIT is never selected by the heuristic because the current write\n *     path emits exactly one canonical memory per cluster.  The prompt\n *     reserves SPLIT for future cluster shapes where the LLM decides one\n *     logical source actually encodes several distinct facts — at which\n *     point the orchestrator would need to write multiple outputs.\n */\nexport function chooseConsolidationOperator(\n  cluster: ConsolidationCluster,\n): _SemanticConsolidationLlmOperator {\n  if (cluster.memories.length <= 1) return \"update\";\n  return \"merge\";\n}\n\n/**\n * Build the operator-aware LLM prompt.  The LLM is asked to return a\n * JSON object `{ \"operator\": <split|merge|update>, \"output\": <content> }`.\n *\n * The prompt is additive: it still asks for a single canonical blob under\n * `output`, so the upstream write path does not change.  Future expansions\n * (SPLIT emitting multiple outputs) are explicitly documented as\n * out-of-scope for the parser — `parseOperatorAwareConsolidationResponse`\n * collapses any SPLIT response into a single canonical output for now.\n */\nexport function buildOperatorAwareConsolidationPrompt(\n  cluster: ConsolidationCluster,\n): string {\n  const memoryTexts = cluster.memories\n    .map(\n      (m, i) =>\n        `Memory ${i + 1} (${m.frontmatter.id}, created ${m.frontmatter.created}):\\n${m.content}`,\n    )\n    .join(\"\\n\\n\");\n\n  return `You are a memory consolidation system.  The following ${cluster.memories.length} memories in the \"${cluster.category}\" category contain overlapping information.\n\nPick exactly ONE consolidation operator for this cluster and return a JSON object.\n\nOperator vocabulary:\n  - \"merge\"  — multiple distinct source memories overlap and should be collapsed into one canonical memory (most common).\n  - \"update\" — one source memory carries a stale value that a newer source supersedes within the same logical fact.\n  - \"split\"  — a single logical source really encodes multiple distinct facts that should be separated (rare; if you pick split, still emit ONE canonical body — the write path will chunk it later).\n\nOutput JSON ONLY, no prose before or after.  The \"operator\" key MUST be set to exactly one of the three strings \"merge\", \"update\", or \"split\" — never a pipe-separated placeholder like \"merge|update|split\".  Example shape:\n  {\n    \"operator\": \"merge\",\n    \"output\": \"<the canonical memory text>\"\n  }\n\nThe \"output\" value must:\n1. Preserve ALL unique information from every source memory\n2. Remove redundancy and repetition\n3. Use clear, concise language\n4. Match the \"${cluster.category}\" category and tone\n5. NOT add information that isn't in the sources\n\n${memoryTexts}\n\nReturn ONLY the JSON object:`;\n}\n\n/**\n * Parse an operator-aware consolidation response.\n *\n * Contract:\n *   - Accepts strict JSON `{ \"operator\": \"...\", \"output\": \"...\" }`.\n *   - Tolerates a JSON payload wrapped in a fenced code block (```json ...```).\n *   - Falls back to the heuristic operator when the JSON is malformed,\n *     the `operator` field is missing / unknown, or the raw response is\n *     a plain blob with no JSON at all.  This keeps PR 3 backwards\n *     compatible with older models that ignore the JSON instruction.\n *   - Never throws.  A missing / empty `output` field falls back to the\n *     trimmed raw response so the caller still writes something rather\n *     than dropping the cluster.\n */\nexport function parseOperatorAwareConsolidationResponse(\n  response: string,\n  cluster: ConsolidationCluster,\n): OperatorAwareConsolidationResult {\n  const fallback: OperatorAwareConsolidationResult = {\n    operator: chooseConsolidationOperator(cluster),\n    output: response.trim(),\n  };\n\n  const trimmed = response.trim();\n  if (trimmed.length === 0) return fallback;\n\n  // Strip a fenced code block if present.\n  // Dropped the \\s* groups around the lazy body (they overlapped it and\n  // backtracked polynomially — CodeQL js/polynomial-redos). Input is already\n  // trimmed and fenced[1] is trimmed below, so matches are identical.\n  const fenced = /^```(?:json)?([\\s\\S]*?)```$/u.exec(trimmed);\n  const payload = fenced ? fenced[1].trim() : trimmed;\n\n  // Find a balanced brace-delimited JSON object that has an `operator`\n  // key (PR #632 round-4 review, codex P1).  A first/last-brace slice\n  // breaks when the model prepends an earlier brace block (e.g.\n  // `Example: {\"note\":\"...\"} ... {\"operator\":\"merge\",...}`).  Walk the\n  // payload tracking nesting + string/escape state and skip past\n  // objects that don't look like our target shape.\n  const parsed = findLastJsonObjectWithOperator(payload);\n  if (parsed === undefined) return fallback;\n  if (typeof parsed !== \"object\" || parsed === null) return fallback;\n\n  const obj = parsed as Record<string, unknown>;\n  const rawOperator = typeof obj.operator === \"string\" ? obj.operator.trim().toLowerCase() : \"\";\n  const rawOutput = typeof obj.output === \"string\" ? obj.output : \"\";\n\n  // Narrow gate (Cursor Bugbot review on PR #730 head `aa1c2a8`):\n  // accept ONLY the legacy split/merge/update LLM vocabulary here.\n  // `pattern-reinforcement` joined the broader `ConsolidationOperator`\n  // type in #687 PR 2/4 but is reserved for the maintenance job — if\n  // an LLM hallucinates that operator we must NOT promote it onto\n  // `derived_via`.\n  const operator = _isSemanticConsolidationLlmOperator(rawOperator)\n    ? rawOperator\n    : chooseConsolidationOperator(cluster);\n  const output = rawOutput.trim().length > 0 ? rawOutput.trim() : response.trim();\n\n  return { operator, output };\n}\n\n/**\n * Walk `text`, find all balanced top-level `{ ... }` blocks whose\n * JSON.parse result is an object with an `operator` key, and return\n * the LAST one (function name reflects this — PR #632 review,\n * cursor Low).  Returns `undefined` when nothing matches.  Tracks\n * string state + escape sequences so braces inside string values\n * don't throw off the depth counter.\n *\n * Used by `parseOperatorAwareConsolidationResponse` to tolerate models\n * that prepend an explanatory JSON example block before the real\n * payload (PR #632 round-4 + round-5 review, codex P1).  We take the\n * LAST candidate so that an instructional example with an `operator`\n * key ahead of the real answer doesn't steal precedence — models\n * typically write the example first and the real answer last.\n */\nfunction findLastJsonObjectWithOperator(text: string): unknown {\n  let searchFrom = 0;\n  let last: unknown = undefined;\n  while (searchFrom < text.length) {\n    const start = text.indexOf(\"{\", searchFrom);\n    if (start < 0) return last;\n    let depth = 0;\n    let inString = false;\n    let escape = false;\n    let closed = false;\n    let endIdx = -1;\n    for (let i = start; i < text.length; i++) {\n      const ch = text[i];\n      if (inString) {\n        if (escape) {\n          escape = false;\n        } else if (ch === \"\\\\\") {\n          escape = true;\n        } else if (ch === '\"') {\n          inString = false;\n        }\n        continue;\n      }\n      if (ch === '\"') {\n        inString = true;\n      } else if (ch === \"{\") {\n        depth += 1;\n      } else if (ch === \"}\") {\n        depth -= 1;\n        if (depth === 0) {\n          closed = true;\n          endIdx = i;\n          break;\n        }\n      }\n    }\n    if (!closed) return last;\n    const slice = text.slice(start, endIdx + 1);\n    try {\n      const parsed = JSON.parse(slice);\n      if (\n        typeof parsed === \"object\" &&\n        parsed !== null &&\n        \"operator\" in (parsed as Record<string, unknown>)\n      ) {\n        last = parsed;\n      }\n    } catch {\n      // Fall through to the next candidate.\n    }\n    searchFrom = endIdx + 1;\n  }\n  return last;\n}\n\n// Silence unused-import warnings when tsup tree-shakes: these are used\n// above in chooseConsolidationOperator + parse helpers.\nvoid _CONSOLIDATION_OPERATORS;\n\n/**\n * Discover extensions and build the block to append to a consolidation prompt.\n * Returns \"\" when extensions are disabled or none are found.\n */\nexport async function buildExtensionsBlockForConsolidation(\n  config: PluginConfig,\n): Promise<string> {\n  if (!config.memoryExtensionsEnabled) return \"\";\n  const root = resolveExtensionsRoot(config);\n  const extensions = await discoverMemoryExtensions(root, log);\n  if (extensions.length === 0) return \"\";\n  return renderExtensionsBlock(extensions);\n}\n\n/**\n * Optional post-consolidation hook — materializes the namespace into Codex's\n * native memory layout when the consolidation run finishes. Kept here (rather\n * than in orchestrator.ts) so #378 doesn't conflict with Wave 1 edits.\n *\n * Safe to call regardless of config state: honors `codexMaterializeMemories`\n * and `codexMaterializeOnConsolidation` and silently becomes a no-op when\n * either is disabled.\n */\nexport async function materializeAfterSemanticConsolidation(options: {\n  config: PluginConfig;\n  namespace?: string;\n  memories?: MemoryFile[];\n  memoryDir?: string;\n  codexHome?: string;\n  rolloutSummaries?: RolloutSummaryInput[];\n  now?: Date;\n}): Promise<MaterializeResult | null> {\n  // Delegates to the shared post-consolidation helper so semantic and causal\n  // flows stay in lock-step — any guard/logging change happens in one place.\n  return runPostConsolidationMaterialize(\"[semantic-consolidation]\", options);\n}\n"],"mappings":";;;;;;;;;;;;;;;;;;;;AA2DO,SAAS,oBACd,UACA,QAMwB;AACxB,QAAM,WAAW,IAAI,IAAI,OAAO,iBAAiB;AAGjD,QAAM,aAAa,oBAAI,IAA0B;AACjD,aAAW,KAAK,UAAU;AACxB,UAAM,MAAM,EAAE,YAAY;AAC1B,QAAI,SAAS,IAAI,GAAG,EAAG;AACvB,QAAI,EAAE,YAAY,UAAU,EAAE,YAAY,WAAW,SAAU;AAC/D,UAAM,OAAO,WAAW,IAAI,GAAG,KAAK,CAAC;AACrC,SAAK,KAAK,CAAC;AACX,eAAW,IAAI,KAAK,IAAI;AAAA,EAC1B;AAEA,QAAM,WAAmC,CAAC;AAC1C,MAAI,kBAAkB;AAEtB,aAAW,CAAC,UAAU,IAAI,KAAK,YAAY;AACzC,QAAI,mBAAmB,OAAO,UAAW;AAGzC,UAAM,YAAY,KAAK,IAAI,CAAC,OAAO;AAAA,MACjC,QAAQ;AAAA,MACR,QAAQ,IAAI,IAAI,sBAAsB,EAAE,SAAS,CAAC,CAAC,CAAC;AAAA,IACtD,EAAE;AAGF,UAAM,YAAY,oBAAI,IAAY;AAElC,aAAS,IAAI,GAAG,IAAI,UAAU,UAAU,kBAAkB,OAAO,WAAW,KAAK;AAC/E,YAAM,kBAAkB,OAAO,YAAY;AAC3C,UAAI,kBAAkB,OAAO,eAAgB;AAC7C,UAAI,UAAU,IAAI,UAAU,CAAC,EAAE,OAAO,YAAY,EAAE,EAAG;AAEvD,YAAM,UAAwB,CAAC,UAAU,CAAC,EAAE,MAAM;AAClD,UAAI,eAAe;AACnB,UAAI,cAAc;AAElB,eAAS,IAAI,IAAI,GAAG,IAAI,UAAU,UAAU,QAAQ,SAAS,iBAAiB,KAAK;AACjF,YAAI,UAAU,IAAI,UAAU,CAAC,EAAE,OAAO,YAAY,EAAE,EAAG;AAEvD,cAAM,UAAU,UAAU,CAAC,EAAE;AAC7B,cAAM,UAAU,UAAU,CAAC,EAAE;AAC7B,YAAI,QAAQ,SAAS,KAAK,QAAQ,SAAS,EAAG;AAG9C,cAAM,UAAU,wBAAwB,SAAS,CAAC,GAAG,OAAO,EAAE,KAAK,GAAG,CAAC;AACvE,cAAM,YAAY,KAAK,IAAI,QAAQ,MAAM,QAAQ,IAAI;AACrD,cAAM,QAAQ,YAAY,IAAI,UAAU,YAAY;AAEpD,YAAI,SAAS,OAAO,WAAW;AAC7B,kBAAQ,KAAK,UAAU,CAAC,EAAE,MAAM;AAChC,0BAAgB;AAChB;AAAA,QACF;AAAA,MACF;AAEA,UAAI,QAAQ,UAAU,OAAO,gBAAgB;AAC3C,mBAAW,KAAK,QAAS,WAAU,IAAI,EAAE,YAAY,EAAE;AACvD,iBAAS,KAAK;AAAA,UACZ;AAAA,UACA,UAAU;AAAA,UACV,cAAc,cAAc,IAAI,eAAe,cAAc;AAAA,QAC/D,CAAC;AACD,2BAAmB,QAAQ;AAAA,MAC7B;AAAA,IACF;AAAA,EACF;AAEA,SAAO;AACT;AAKO,SAAS,yBAAyB,SAAuC;AAC9E,QAAM,cAAc,QAAQ,SACzB;AAAA,IACC,CAAC,GAAG,MACF,UAAU,IAAI,CAAC,KAAK,EAAE,YAAY,EAAE,aAAa,EAAE,YAAY,OAAO;AAAA,EAAO,EAAE,OAAO;AAAA,EAC1F,EACC,KAAK,MAAM;AAEd,SAAO,wDAAwD,QAAQ,SAAS,MAAM,qBAAqB,QAAQ,QAAQ;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA,EAS3H,WAAW;AAAA;AAAA;AAGb;AAKO,SAAS,2BAA2B,UAA0B;AACnE,SAAO,SAAS,KAAK;AACvB;AAyCO,SAAS,4BACd,SACmC;AACnC,MAAI,QAAQ,SAAS,UAAU,EAAG,QAAO;AACzC,SAAO;AACT;AAYO,SAAS,sCACd,SACQ;AACR,QAAM,cAAc,QAAQ,SACzB;AAAA,IACC,CAAC,GAAG,MACF,UAAU,IAAI,CAAC,KAAK,EAAE,YAAY,EAAE,aAAa,EAAE,YAAY,OAAO;AAAA,EAAO,EAAE,OAAO;AAAA,EAC1F,EACC,KAAK,MAAM;AAEd,SAAO,yDAAyD,QAAQ,SAAS,MAAM,qBAAqB,QAAQ,QAAQ;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA,gBAmB9G,QAAQ,QAAQ;AAAA;AAAA;AAAA,EAG9B,WAAW;AAAA;AAAA;AAGb;AAgBO,SAAS,wCACd,UACA,SACkC;AAClC,QAAM,WAA6C;AAAA,IACjD,UAAU,4BAA4B,OAAO;AAAA,IAC7C,QAAQ,SAAS,KAAK;AAAA,EACxB;AAEA,QAAM,UAAU,SAAS,KAAK;AAC9B,MAAI,QAAQ,WAAW,EAAG,QAAO;AAMjC,QAAM,SAAS,+BAA+B,KAAK,OAAO;AAC1D,QAAM,UAAU,SAAS,OAAO,CAAC,EAAE,KAAK,IAAI;AAQ5C,QAAM,SAAS,+BAA+B,OAAO;AACrD,MAAI,WAAW,OAAW,QAAO;AACjC,MAAI,OAAO,WAAW,YAAY,WAAW,KAAM,QAAO;AAE1D,QAAM,MAAM;AACZ,QAAM,cAAc,OAAO,IAAI,aAAa,WAAW,IAAI,SAAS,KAAK,EAAE,YAAY,IAAI;AAC3F,QAAM,YAAY,OAAO,IAAI,WAAW,WAAW,IAAI,SAAS;AAQhE,QAAM,WAAW,mCAAoC,WAAW,IAC5D,cACA,4BAA4B,OAAO;AACvC,QAAM,SAAS,UAAU,KAAK,EAAE,SAAS,IAAI,UAAU,KAAK,IAAI,SAAS,KAAK;AAE9E,SAAO,EAAE,UAAU,OAAO;AAC5B;AAiBA,SAAS,+BAA+B,MAAuB;AAC7D,MAAI,aAAa;AACjB,MAAI,OAAgB;AACpB,SAAO,aAAa,KAAK,QAAQ;AAC/B,UAAM,QAAQ,KAAK,QAAQ,KAAK,UAAU;AAC1C,QAAI,QAAQ,EAAG,QAAO;AACtB,QAAI,QAAQ;AACZ,QAAI,WAAW;AACf,QAAI,SAAS;AACb,QAAI,SAAS;AACb,QAAI,SAAS;AACb,aAAS,IAAI,OAAO,IAAI,KAAK,QAAQ,KAAK;AACxC,YAAM,KAAK,KAAK,CAAC;AACjB,UAAI,UAAU;AACZ,YAAI,QAAQ;AACV,mBAAS;AAAA,QACX,WAAW,OAAO,MAAM;AACtB,mBAAS;AAAA,QACX,WAAW,OAAO,KAAK;AACrB,qBAAW;AAAA,QACb;AACA;AAAA,MACF;AACA,UAAI,OAAO,KAAK;AACd,mBAAW;AAAA,MACb,WAAW,OAAO,KAAK;AACrB,iBAAS;AAAA,MACX,WAAW,OAAO,KAAK;AACrB,iBAAS;AACT,YAAI,UAAU,GAAG;AACf,mBAAS;AACT,mBAAS;AACT;AAAA,QACF;AAAA,MACF;AAAA,IACF;AACA,QAAI,CAAC,OAAQ,QAAO;AACpB,UAAM,QAAQ,KAAK,MAAM,OAAO,SAAS,CAAC;AAC1C,QAAI;AACF,YAAM,SAAS,KAAK,MAAM,KAAK;AAC/B,UACE,OAAO,WAAW,YAClB,WAAW,QACX,cAAe,QACf;AACA,eAAO;AAAA,MACT;AAAA,IACF,QAAQ;AAAA,IAER;AACA,iBAAa,SAAS;AAAA,EACxB;AACA,SAAO;AACT;AAUA,eAAsB,qCACpB,QACiB;AACjB,MAAI,CAAC,OAAO,wBAAyB,QAAO;AAC5C,QAAM,OAAO,sBAAsB,MAAM;AACzC,QAAM,aAAa,MAAM,yBAAyB,MAAM,GAAG;AAC3D,MAAI,WAAW,WAAW,EAAG,QAAO;AACpC,SAAO,sBAAsB,UAAU;AACzC;AAWA,eAAsB,sCAAsC,SAQtB;AAGpC,SAAO,gCAAgC,4BAA4B,OAAO;AAC5E;","names":[]}

package/dist/{chunk-CCNZM5UM.js → chunk-R3OQGYOU.js}

@@ -3,7 +3,7 @@ import {
 } from "./chunk-HQ6NIBL6.js";
 import {
   StorageManager
-} from "./chunk-DXBCNDVD.js";
+} from "./chunk-PJGB7XRR.js";
 import {
   getCachedEpisodeMap,
   setCachedEpisodeMap
@@ -105,4 +105,4 @@ async function searchVerifiedEpisodes(options) {
 export {
   searchVerifiedEpisodes
 };
-//# sourceMappingURL=chunk-CCNZM5UM.js.map
+//# sourceMappingURL=chunk-R3OQGYOU.js.map

package/dist/{chunk-UZB5KHKX.js → chunk-RGMVMVMF.js}

@@ -1,6 +1,6 @@
 // src/json-extract.ts
 function stripCodeFences(text) {
-  return text.replace(/```(?:json)?\s*([\s\S]*?)```/gi, (_m, inner) => String(inner).trim());
+  return text.replace(/```(?:json)?([\s\S]*?)```/gi, (_m, inner) => String(inner).trim());
 }
 function extractJsonCandidates(text) {
   const trimmed = text.trim();
@@ -60,4 +60,4 @@ export {
   stripCodeFences,
   extractJsonCandidates
 };
-//# sourceMappingURL=chunk-UZB5KHKX.js.map
+//# sourceMappingURL=chunk-RGMVMVMF.js.map

package/dist/chunk-RGMVMVMF.js.map

@@ -0,0 +1 @@
+{"version":3,"sources":["../src/json-extract.ts"],"sourcesContent":["/**\n * Utilities for extracting JSON payloads from LLM outputs.\n *\n * We see common failure modes:\n * - \"Here's an example: {..}\\nHere's the real answer: {..}\" (multiple JSON blocks)\n * - fenced ```json blocks\n * - leading/trailing prose around JSON\n *\n * These helpers attempt multiple candidates and let callers validate with schemas.\n */\n\nexport function stripCodeFences(text: string): string {\n  // Drop the leading \\s* before the lazy body: it overlapped the body and caused\n  // polynomial backtracking on unterminated fences (CodeQL js/polynomial-redos).\n  // inner is trimmed, so captured content is identical.\n  return text.replace(/```(?:json)?([\\s\\S]*?)```/gi, (_m, inner) => String(inner).trim());\n}\n\nexport function extractJsonCandidates(text: string): string[] {\n  const trimmed = text.trim();\n  const cleaned = stripCodeFences(trimmed);\n  const candidates: string[] = [];\n\n  if (cleaned.length > 0) candidates.push(cleaned);\n  candidates.push(...scanBalancedJsonBlocks(cleaned));\n\n  // Legacy regex fallback (single object)\n  const objMatch = cleaned.match(/\\{[\\s\\S]*\\}/);\n  if (objMatch) candidates.push(objMatch[0]);\n\n  const seen = new Set<string>();\n  return candidates\n    .map((c) => c.trim())\n    .filter((c) => c.length > 0)\n    .filter((c) => {\n      if (seen.has(c)) return false;\n      seen.add(c);\n      return true;\n    });\n}\n\nfunction scanBalancedJsonBlocks(text: string): string[] {\n  const out: string[] = [];\n  const opens = new Set([\"{\", \"[\"]);\n  const closes: Record<string, string> = { \"{\": \"}\", \"[\": \"]\" };\n\n  for (let i = 0; i < text.length; i++) {\n    const start = text[i];\n    if (!opens.has(start)) continue;\n\n    const expectedClose = closes[start];\n    let depth = 0;\n    let inString = false;\n    let escape = false;\n\n    for (let j = i; j < text.length; j++) {\n      const ch = text[j];\n\n      if (inString) {\n        if (escape) {\n          escape = false;\n        } else if (ch === \"\\\\\") {\n          escape = true;\n        } else if (ch === \"\\\"\") {\n          inString = false;\n        }\n        continue;\n      }\n\n      if (ch === \"\\\"\") {\n        inString = true;\n        continue;\n      }\n\n      if (ch === start) depth++;\n      if (ch === expectedClose) depth--;\n\n      if (depth === 0) {\n        out.push(text.slice(i, j + 1).trim());\n        i = j;\n        break;\n      }\n    }\n  }\n\n  return out;\n}\n\n"],"mappings":";AAWO,SAAS,gBAAgB,MAAsB;AAIpD,SAAO,KAAK,QAAQ,+BAA+B,CAAC,IAAI,UAAU,OAAO,KAAK,EAAE,KAAK,CAAC;AACxF;AAEO,SAAS,sBAAsB,MAAwB;AAC5D,QAAM,UAAU,KAAK,KAAK;AAC1B,QAAM,UAAU,gBAAgB,OAAO;AACvC,QAAM,aAAuB,CAAC;AAE9B,MAAI,QAAQ,SAAS,EAAG,YAAW,KAAK,OAAO;AAC/C,aAAW,KAAK,GAAG,uBAAuB,OAAO,CAAC;AAGlD,QAAM,WAAW,QAAQ,MAAM,aAAa;AAC5C,MAAI,SAAU,YAAW,KAAK,SAAS,CAAC,CAAC;AAEzC,QAAM,OAAO,oBAAI,IAAY;AAC7B,SAAO,WACJ,IAAI,CAAC,MAAM,EAAE,KAAK,CAAC,EACnB,OAAO,CAAC,MAAM,EAAE,SAAS,CAAC,EAC1B,OAAO,CAAC,MAAM;AACb,QAAI,KAAK,IAAI,CAAC,EAAG,QAAO;AACxB,SAAK,IAAI,CAAC;AACV,WAAO;AAAA,EACT,CAAC;AACL;AAEA,SAAS,uBAAuB,MAAwB;AACtD,QAAM,MAAgB,CAAC;AACvB,QAAM,QAAQ,oBAAI,IAAI,CAAC,KAAK,GAAG,CAAC;AAChC,QAAM,SAAiC,EAAE,KAAK,KAAK,KAAK,IAAI;AAE5D,WAAS,IAAI,GAAG,IAAI,KAAK,QAAQ,KAAK;AACpC,UAAM,QAAQ,KAAK,CAAC;AACpB,QAAI,CAAC,MAAM,IAAI,KAAK,EAAG;AAEvB,UAAM,gBAAgB,OAAO,KAAK;AAClC,QAAI,QAAQ;AACZ,QAAI,WAAW;AACf,QAAI,SAAS;AAEb,aAAS,IAAI,GAAG,IAAI,KAAK,QAAQ,KAAK;AACpC,YAAM,KAAK,KAAK,CAAC;AAEjB,UAAI,UAAU;AACZ,YAAI,QAAQ;AACV,mBAAS;AAAA,QACX,WAAW,OAAO,MAAM;AACtB,mBAAS;AAAA,QACX,WAAW,OAAO,KAAM;AACtB,qBAAW;AAAA,QACb;AACA;AAAA,MACF;AAEA,UAAI,OAAO,KAAM;AACf,mBAAW;AACX;AAAA,MACF;AAEA,UAAI,OAAO,MAAO;AAClB,UAAI,OAAO,cAAe;AAE1B,UAAI,UAAU,GAAG;AACf,YAAI,KAAK,KAAK,MAAM,GAAG,IAAI,CAAC,EAAE,KAAK,CAAC;AACpC,YAAI;AACJ;AAAA,MACF;AAAA,IACF;AAAA,EACF;AAEA,SAAO;AACT;","names":[]}

package/dist/{chunk-EDP57PFC.js → chunk-RKW6QR7W.js}

@@ -22,7 +22,7 @@ import {
 import {
   CompoundingEngine,
   defaultTierMigrationCycleBudget
-} from "./chunk-UK727RHF.js";
+} from "./chunk-2L54V4ZO.js";
 import {
   SharedContextManager
 } from "./chunk-DRD2Q7HQ.js";
@@ -48,7 +48,7 @@ import {
   applyTemporalSupersession,
   normalizeSupersessionKey,
   shouldFilterSupersededFromRecall
-} from "./chunk-K5O2QY6T.js";
+} from "./chunk-YTWNKQ2G.js";
 import {
   isValidAsOf
 } from "./chunk-MDYG7VI7.js";
@@ -57,7 +57,7 @@ import {
 } from "./chunk-VOUOLGIP.js";
 import {
   HourlySummarizer
-} from "./chunk-XAZOWLW4.js";
+} from "./chunk-3VONWEQB.js";
 import {
   SessionObserverState
 } from "./chunk-MHQC2WU2.js";
@@ -147,10 +147,10 @@ import {
   getVerdictKind,
   judgeFactDurability,
   validateProcedureExtraction
-} from "./chunk-2SGJY2UY.js";
+} from "./chunk-Z3CCEP6F.js";
 import {
   ExtractionEngine
-} from "./chunk-WB3LYXC5.js";
+} from "./chunk-MON3LMO7.js";
 import {
   parseMemoryActionEligibilityContext
 } from "./chunk-ROZJACKP.js";
@@ -168,7 +168,7 @@ import {
   buildEntityRecallSection,
   entityRecentTranscriptLookbackHours,
   readRecentEntityTranscriptEntries
-} from "./chunk-LANHQ7EN.js";
+} from "./chunk-BNW5NJJH.js";
 import {
   buildEventOrderRecallSection,
   shouldRecallEventOrderEvidence
@@ -203,7 +203,7 @@ import {
   materializeAfterSemanticConsolidation,
   parseConsolidationResponse,
   parseOperatorAwareConsolidationResponse
-} from "./chunk-ZNCDQZIS.js";
+} from "./chunk-QFQQFX2H.js";
 import {
   normalizeReplaySessionKey
 } from "./chunk-2PRQG7PV.js";
@@ -212,13 +212,13 @@ import {
 } from "./chunk-X6IRLNOO.js";
 import {
   searchVerifiedEpisodes
-} from "./chunk-CCNZM5UM.js";
+} from "./chunk-R3OQGYOU.js";
 import {
   ThreadingManager
 } from "./chunk-W4RVMTHR.js";
 import {
   searchVerifiedSemanticRules
-} from "./chunk-YPNGPHNZ.js";
+} from "./chunk-2UFQYU5F.js";
 import {
   searchWorkProductLedgerEntries
 } from "./chunk-ZRWB5D4H.js";
@@ -237,7 +237,7 @@ import {
 } from "./chunk-CYEPCZN5.js";
 import {
   NamespaceStorageRouter
-} from "./chunk-ZUNNG6PC.js";
+} from "./chunk-W3BKVM64.js";
 import {
   namespaceIdentityFromToken
 } from "./chunk-ZFXCQPNO.js";
@@ -274,17 +274,17 @@ import {
 } from "./chunk-LMDRGRJ2.js";
 import {
   semanticChunkContent
-} from "./chunk-WCYKT2DE.js";
+} from "./chunk-P4BC54KI.js";
 import {
   chunkContent
-} from "./chunk-4WMCPJWX.js";
+} from "./chunk-UQ7RN5HK.js";
 import {
   SmartBuffer
 } from "./chunk-IBTZEBUD.js";
 import {
   FallbackLlmClient,
   fallbackLlmRuntimeContextFromConfig
-} from "./chunk-DEVUWMME.js";
+} from "./chunk-KGIGRNR6.js";
 import {
   buildRecallQueryPolicy
 } from "./chunk-LBJBNWS2.js";
@@ -330,7 +330,7 @@ import {
 } from "./chunk-QDW3E4RD.js";
 import {
   shouldSkipImplicitExtraction
-} from "./chunk-IMA6GU4Y.js";
+} from "./chunk-H3PHZLMF.js";
 import {
   GraphIndex
 } from "./chunk-XL7FK7PJ.js";
@@ -342,12 +342,12 @@ import {
   normalizeAttributePairs,
   normalizeEntityName,
   parseEntityFile
-} from "./chunk-DXBCNDVD.js";
+} from "./chunk-PJGB7XRR.js";
 import {
   attachCitation,
   hasCitationForTemplate,
   stripCitationForTemplate
-} from "./chunk-4H5ZJHEN.js";
+} from "./chunk-J6A3CX5N.js";
 import {
   confidenceTier
 } from "./chunk-4R4KTDIE.js";
@@ -1200,7 +1200,11 @@ function buildCompressionGuidelinesMarkdown2(events, generatedAtIso = (/* @__PUR
 function formatCompressionGuidelinesForRecall(raw, maxLines = 5) {
   if (typeof raw !== "string" || raw.trim().length === 0) return null;
   const sectionMatch = raw.match(
-    /## Suggested Guidelines\s*\n([\s\S]*?)(?:\n##\s+|\s*$)/i
+    // End the section at `\n## ` or end-of-string. Plain $ (not \s*$): the
+    // \s* branch overlapped the lazy body and backtracked polynomially
+    // (CodeQL js/polynomial-redos). Captured lines are trimmed/filtered below,
+    // so trailing whitespace handling is unchanged.
+    /## Suggested Guidelines\s*\n([\s\S]*?)(?:\n##\s+|$)/i
   );
   if (!sectionMatch) return null;
   const lines = sectionMatch[1].split("\n").map((line) => line.trim()).filter((line) => line.startsWith("- ")).slice(0, Math.max(1, Math.floor(maxLines)));
@@ -12498,4 +12502,4 @@ export {
   resolvePersistedMemoryRelativePath,
   Orchestrator
 };
-//# sourceMappingURL=chunk-EDP57PFC.js.map
+//# sourceMappingURL=chunk-RKW6QR7W.js.map