@amemhq/core 2.1.0 → 2.1.2

package/README.md

@@ -6,7 +6,7 @@ Framework-agnostic **A-MEM agentic memory engine** — memories that **evolve**,
 npm i @amemhq/core
 ```
 
-Extracted from [`openclaw-amem`](https://github.com/amemhq/amem/tree/main/packages/openclaw-amem) so any host can share one memory engine: an OpenClaw plugin, a standalone service ([`amem-api`](../amem-api)), or a game agent. Part of the [amem monorepo](../../).
+We extracted this package from [`openclaw-amem`](https://github.com/amemhq/amem/tree/main/packages/openclaw-amem) so that any host can share one memory engine: an OpenClaw plugin, a standalone service ([`amem-api`](../amem-api)), or a game agent. It is part of the [amem monorepo](../../).
 
 > Based on _A-MEM: Agentic Memory for LLM Agents_ ([arXiv:2502.12110](https://arxiv.org/abs/2502.12110), NeurIPS 2025). For the original research implementation, see [agiresearch/A-MEM](https://github.com/agiresearch/A-MEM).
 
@@ -14,30 +14,30 @@ Extracted from [`openclaw-amem`](https://github.com/amemhq/amem/tree/main/packag
 
 Unlike a flat vector store, A-MEM maintains memory as a living, self-evolving semantic graph. On every write:
 
-1. **Note Construction** — an LLM extracts keywords, tags, and a context summary; categorizes the note; and classifies it as `memory` (episodic) or `knowledge` (durable), extracting 1–5 `topics` for knowledge notes.
-2. **Link Generation** — retrieves top-6 candidates; the LLM judges whether to link bidirectionally (similarity > 0.3).
-3. **Memory Evolution** — up to 3 linked notes have their attributes evolved from the new context, possibly triggering further links.
-4. **Hybrid Retrieval** — fuses dense vectors (Transformers.js `bge-m3`, 1024-dim) and BM25 via Reciprocal Rank Fusion (RRF), boosted by retrieval heat.
-5. **2-hop BFS Graph Expansion** — after RRF top-K, BFS walks the link graph up to 2 hops, admitting up to 8 graph-connected notes that pass an embedding relevance gate (cos-sim ≥ 0.25). This is the key advantage over flat vector systems.
+1. **Note Construction** — an LLM extracts keywords, tags, and a context summary, categorizes the note, and classifies it as `memory` (episodic) or `knowledge` (durable). For knowledge notes, the LLM extracts 1–5 `topics`.
+2. **Link Generation** — the engine retrieves the top-6 candidates. The LLM judges whether to link them bidirectionally (similarity > 0.3).
+3. **Memory Evolution** — the engine evolves the attributes of up to 3 linked notes from the new context. This can trigger further links.
+4. **Hybrid Retrieval** — the engine fuses dense vectors (Transformers.js `bge-m3`, 1024-dim) and BM25 via Reciprocal Rank Fusion (RRF). Retrieval heat boosts the result.
+5. **2-hop BFS Graph Expansion** — after the RRF top-K, BFS walks the link graph up to 2 hops and admits up to 8 graph-connected notes that pass an embedding relevance gate (cos-sim ≥ 0.25). This is the key advantage over flat vector systems.
 
 ## Features
 
 - 🔄 **Dynamic memory network** (Zettelkasten-inspired) — notes are graph nodes with bidirectional links, not flat rows.
-- 🧬 **Evolution & strengthening** — linked notes update context/tags/embeddings when new details arrive; `evolution_history` audit trail.
-- 🚦 **LLM CRUD gate** — analyzes a user↔assistant exchange and decides `NEW` / `UPDATE` / `DELETE` / `NONE` to keep memory clean.
-- 🧹 **Same-day merge + daily consolidation** — merges semantic duplicates (≥ 0.80 same-day; ≥ 0.75 in the 02:30 sweep) and **cascades link references** to preserve graph topology.
-- ⏳ **Temporal soft-delete** — outdated/conflicting notes are marked `is_active: false` (zero-migration Qdrant filter) and excluded from search.
-- 🔥 **Heat tracking with time decay** — `retrieval_count` + `last_accessed` give a logarithmic boost, dampened by age so stale notes don't permanently outrank fresh ones:
+- 🧬 **Evolution & strengthening** — linked notes update context/tags/embeddings when new details arrive. The `evolution_history` is a full audit trail.
+- 🚦 **LLM CRUD gate** — the LLM analyzes a user↔assistant exchange and decides `NEW` / `UPDATE` / `DELETE` / `NONE` to keep memory clean.
+- 🧹 **Same-day merge + daily consolidation** — the engine merges semantic duplicates (≥ 0.80 same-day, ≥ 0.75 in the 02:30 sweep) and **cascades link references** to preserve graph topology.
+- ⏳ **Temporal soft-delete** — the engine marks outdated and conflicting notes `is_active: false` (zero-migration Qdrant filter) and excludes them from search.
+- 🔥 **Heat tracking with time decay** — `retrieval_count` + `last_accessed` give a logarithmic boost. Age dampens the boost so stale notes do not permanently outrank fresh ones:
 
   ```
   Final Score = RRF Score × (1 + 0.05 × ln(1 + retrieval_count) / (age_days + 1))
   ```
 
-- 🔍 **2-hop graph traversal with relevance gate** — BFS from anchors, admitting only nodes with cos-sim ≥ 0.25 to the query.
-- 🀄 **Chinese-optimized BM25** — [Jieba](https://github.com/fxsjy/jieba) (`@node-rs/jieba`) word segmentation for CJK; whitespace fallback for other languages.
-- 🧠 **Knowledge vs episodic** — `note_type` separates durable `knowledge` (skips consolidation-merge + time-decay) from `memory`; `topics` tags + `topicsFilter` enable subject-level recall.
-- 🔐 **Multi-agent isolation** — explicit `owner` / `readers` / `writers` on every note; Mode A (shared collection filtered by `agent_id`) or Mode B (dedicated collection).
-- 📊 **Quality controls** — write-time gate rejects < 10-char content and flags ephemeral notes; `scanLowQuality` finds too-short/expired/conflicting notes.
+- 🔍 **2-hop graph traversal with relevance gate** — BFS walks from anchors and admits only nodes with cos-sim ≥ 0.25 to the query.
+- 🀄 **Chinese-optimized BM25** — [Jieba](https://github.com/fxsjy/jieba) (`@node-rs/jieba`) provides word segmentation for Chinese. Whitespace segmentation is the fallback elsewhere. Japanese and Korean get no lexical tokens, but dense retrieval still covers them.
+- 🧠 **Knowledge vs episodic** — `note_type` separates durable `knowledge` (skips consolidation-merge + time-decay) from `memory`. The `topics` tags and `topicsFilter` enable subject-level recall.
+- 🔐 **Multi-agent isolation** — every note has explicit `owner` / `readers` / `writers` fields. Mode A uses a shared collection filtered by `agent_id`. Mode B uses a dedicated collection.
+- 📊 **Quality controls** — the write-time gate rejects content shorter than 10 characters and flags ephemeral notes. `scanLowQuality` finds too-short, expired, and conflicting notes.
 
 ## Architecture
 
@@ -53,7 +53,7 @@ host (OpenClaw plugin / amem-api / game agent)
 
 ## Memory Evolution
 
-When a new note is borderline-similar to an existing one (cosine 0.72–0.85), @amemhq/core routes it through an **LLM evolution judgment** instead of naive dedup, classifying the relationship:
+When a new note is borderline-similar to an existing one (cosine 0.72–0.85), @amemhq/core routes it through an **LLM evolution judgment** instead of naive dedup. The LLM classifies the relationship:
 
 | Type | Meaning | Action |
 | --- | --- | --- |
@@ -62,12 +62,12 @@ When a new note is borderline-similar to an existing one (cosine 0.72–0.85), @
 | **EXPAND** | New info complements the old | Content merged into old note, history appended, new note absorbed |
 | **NEW** | Unrelated | Both kept as-is |
 
-Memories **evolve** rather than being silently overwritten; `evolution_history` is a full audit trail. (Taxonomy per the SSGM framework, arXiv:2603.11768.)
+Memories **evolve** rather than being silently overwritten. The `evolution_history` is a full audit trail. (Taxonomy per the SSGM framework, arXiv:2603.11768.)
 
 ## Quality Scoring
 
-- **Write-time gate** (`checkQuality`) — content < 10 chars is rejected; temporal signal words (`待跑`, `等确认`, `昨日`, `明天完成`) flag the note `ephemeral: true`.
-- **Periodic scan** (`scanLowQuality` / `generateReviewBatch`) — flags `too_short`, `expired_ephemeral` (> 7 days), and `pending_conflict`, patching `low_quality: true` and emitting an Obsidian-compatible review batch.
+- **Write-time gate** (`checkQuality`) — the gate rejects content shorter than 10 chars. Temporal signal words (`待跑`, `等确认`, `昨日`, `明天完成`) flag the note `ephemeral: true`.
+- **Periodic scan** (`scanLowQuality` / `generateReviewBatch`) — flags `too_short`, `expired_ephemeral` (> 7 days), and `pending_conflict`. It patches `low_quality: true` and emits an Obsidian-compatible review batch.
 
 ## Multi-Agent Isolation
 
@@ -77,10 +77,10 @@ Every `MemoryNote` carries access fields:
 { owner: 'main', readers: ['main'], writers: ['main'] } // readers: ['*'] = shared with all agents
 ```
 
-- **Mode A** (default) — one shared Qdrant collection, isolated by `agent_id` at query time; `agent_id="shared"` (explicit, auditable) publishes a note to all agents.
+- **Mode A** (default) — one shared Qdrant collection, isolated by `agent_id` at query time. Setting `agent_id="shared"` (explicit, auditable) publishes a note to all agents.
 - **Mode B** — a dedicated collection per agent for full physical isolation.
 
-Isolation is the default; sharing is an explicit exception (per arXiv:2604.16548). Consolidation runs per-agent scope.
+Isolation is the default. Sharing is an explicit exception (per arXiv:2604.16548). Consolidation runs per-agent scope.
 
 ## Usage
 
@@ -98,7 +98,7 @@ const hits = await searchMemory('what does the player like to build with?', 5, '
 
 - Node.js 24 (18+ works)
 - [Qdrant](https://qdrant.tech) on `:6333`
-- An LLM for note/link/evolution calls: `ANTHROPIC_API_KEY` by default, or set `AMEM_LLM_PROVIDER=openai` with `AMEM_LLM_BASE_URL` + `AMEM_LLM_API_KEY` to use any OpenAI-compatible endpoint (OpenAI, DeepSeek, OpenRouter, Ollama, vLLM…)
+- An LLM for note, link, and evolution calls — `ANTHROPIC_API_KEY` is the default. To use any OpenAI-compatible endpoint (OpenAI, DeepSeek, OpenRouter, Ollama, vLLM…), set `AMEM_LLM_PROVIDER=openai` with `AMEM_LLM_BASE_URL` and `AMEM_LLM_API_KEY`.
 
 ## References & Citation
 

package/dist/{chunk-XEMQZNLD.js → chunk-Z7G7KOJK.js}

@@ -7,6 +7,22 @@ function canRead(note, callerAgentId) {
   return note.owner === callerAgentId || note.readers.includes(callerAgentId) || note.readers.includes("*");
 }
 
+// src/config.ts
+import * as os from "os";
+import * as path from "path";
+var _dataDir = process.env.AMEM_DATA_DIR || path.join(os.homedir(), ".amem");
+var _warn = (msg) => console.warn(msg);
+function configure(opts) {
+  if (opts.dataDir) _dataDir = opts.dataDir;
+  if (opts.warn) _warn = opts.warn;
+}
+function getDataDir() {
+  return _dataDir;
+}
+function warn(msg) {
+  _warn(msg);
+}
+
 // src/embedding.ts
 var pipeline = null;
 var extractor = null;
@@ -63,6 +79,12 @@ function applyModelPaths(env) {
     env.allowLocalModels = true;
   }
 }
+function humanBytes(n) {
+  if (n >= 1e9) return `${(n / 1e9).toFixed(2)} GB`;
+  if (n >= 1e6) return `${(n / 1e6).toFixed(1)} MB`;
+  if (n >= 1e3) return `${Math.round(n / 1e3)} kB`;
+  return `${n} B`;
+}
 function makeProgressReporter() {
   const lastPct = /* @__PURE__ */ new Map();
   return (e) => {
@@ -71,7 +93,7 @@ function makeProgressReporter() {
     const pct = Math.floor(e.progress / 10) * 10;
     if (lastPct.get(e.file) === pct) return;
     lastPct.set(e.file, pct);
-    const size = e.total ? ` of ${(e.total / 1e9).toFixed(2)} GB` : "";
+    const size = e.total ? ` of ${humanBytes(e.total)}` : "";
     console.log(`[amem] downloading ${e.file}: ${pct}%${size}`);
   };
 }
@@ -283,7 +305,7 @@ async function ensureCollection(collectionName) {
       if (inUse !== null && inUse !== wanted) throw new MixedEmbeddingModelsError(col, wanted, inUse);
       pinEmbeddingModel(wanted);
       if (wanted === LEGACY_DEFAULT_EMBEDDING_MODEL) {
-        console.warn(
+        warn(
           `[amem] "${col}" is on ${LEGACY_DEFAULT_EMBEDDING_MODEL} (${LEGACY_DEFAULT_DIM}-dim).
 [amem] ${DEFAULT_EMBEDDING_MODEL} reads 8192 tokens where that one stops at 128, so anything longer is being truncated before it reaches the vector.
 [amem] To move: npx --package=@amemhq/core amem-migrate --from-collection ${col}`
@@ -687,7 +709,7 @@ function makeCrud(collectionName, modeBIsolated = false) {
             })
           )
         ]).catch((err) => {
-          console.error(`[amem] retrieval tracking patch failed: ${err.message}`);
+          warn(`[amem] retrieval tracking patch failed: ${err.message}`);
         });
         for (const r of queryResults) {
           r.note.retrieval_count = (r.note.retrieval_count || 0) + 1;
@@ -1054,7 +1076,7 @@ var _warned = /* @__PURE__ */ new Set();
 function warnOnce(key, message) {
   if (_warned.has(key)) return;
   _warned.add(key);
-  console.error(message);
+  warn(message);
 }
 function resolveProvider(role = "fast") {
   const raw = role === "strong" ? process.env.AMEM_LLM_STRONG_PROVIDER || _override.strong?.provider || void 0 : void 0;
@@ -1126,9 +1148,11 @@ async function llmCall(prompt, maxTokens = 500, role = "fast") {
   const isThinking = model.includes("gemini") || model.includes("pro-agent");
   const effectiveMaxTokens = isThinking ? Math.max(maxTokens * 8, 4e3) : maxTokens;
   try {
-    return provider === "openai" ? await openaiCall(prompt, model, effectiveMaxTokens, baseURL) : await anthropicCall(prompt, model, effectiveMaxTokens, baseURL);
+    const text = provider === "openai" ? await openaiCall(prompt, model, effectiveMaxTokens, baseURL) : await anthropicCall(prompt, model, effectiveMaxTokens, baseURL);
+    if (text === null) warn(`[amem] ${provider} answered with no text (model ${model})`);
+    return text;
   } catch (e) {
-    console.error(`[amem] LLM call failed: ${e.message}`);
+    warn(`[amem] LLM call failed: ${e.message}`);
     return null;
   }
 }
@@ -1227,7 +1251,8 @@ confidence guide (Story 27):
 
 Text: ${content}`;
   const raw = await llmCall(prompt, 400);
-  if (!raw)
+  if (!raw) {
+    warn("[amem] note construction got nothing back; storing with no keywords, tags or context");
     return {
       keywords: [],
       tags: [],
@@ -1237,6 +1262,7 @@ Text: ${content}`;
       topics: [],
       confidence: "medium"
     };
+  }
   try {
     const data = parseJsonLoose(raw);
     const rawCategory = typeof data.category === "string" ? data.category : "General";
@@ -1255,7 +1281,7 @@ Text: ${content}`;
       confidence
     };
   } catch (e) {
-    console.error(`[amem] Note construction parse failed: ${e.message}`);
+    warn(`[amem] Note construction parse failed: ${e.message}`);
     return {
       keywords: [],
       tags: [],
@@ -1284,9 +1310,15 @@ async function llmCrudDecision(userText, assistantText, existingMemories) {
     const raw = await llmCall(prompt, 400, resolveCrudRole());
     if (!raw) return [];
     const match = stripReasoning(raw).match(/\[.*\]/s);
-    if (!match) return [];
+    if (!match) {
+      warn("[amem] llmCrudDecision found no array in the response; nothing from this turn is stored");
+      return [];
+    }
     const parsed = JSON.parse(match[0]);
-    if (!Array.isArray(parsed)) return [];
+    if (!Array.isArray(parsed)) {
+      warn("[amem] llmCrudDecision parsed a non-array; nothing from this turn is stored");
+      return [];
+    }
     const ops = [];
     for (const item of parsed) {
       if (!item || typeof item !== "object") continue;
@@ -1305,7 +1337,7 @@ async function llmCrudDecision(userText, assistantText, existingMemories) {
     }
     return ops.slice(0, 3);
   } catch (e) {
-    console.error(`[amem] llmCrudDecision failed: ${e.message}`);
+    warn(`[amem] llmCrudDecision failed: ${e.message}`);
     return [];
   }
 }
@@ -1315,13 +1347,16 @@ async function llmShouldMerge(contentA, contentB) {
   if (!raw) return { shouldMerge: false };
   try {
     const data = parseJsonLoose(raw);
-    if (typeof data.shouldMerge !== "boolean") return { shouldMerge: false };
+    if (typeof data.shouldMerge !== "boolean") {
+      warn("[amem] llmShouldMerge got no boolean verdict; treating the pair as distinct");
+      return { shouldMerge: false };
+    }
     if (data.shouldMerge && typeof data.merged === "string") {
       return { shouldMerge: true, merged: data.merged };
     }
     return { shouldMerge: false };
   } catch (e) {
-    console.error(`[amem] llmShouldMerge parse failed: ${e.message}`);
+    warn(`[amem] llmShouldMerge parse failed: ${e.message}`);
     return { shouldMerge: false };
   }
 }
@@ -1329,7 +1364,10 @@ var VALID_EVOLUTION_TYPES = /* @__PURE__ */ new Set(["EVOLVE", "CONFLICT", "EXPA
 async function llmEvolutionJudge(oldContent, newContent) {
   const prompt = t.evolutionJudge(oldContent, newContent);
   const raw = await llmCall(prompt, 300, "strong");
-  if (!raw) return { type: "NEW" };
+  if (!raw) {
+    warn("[amem] evolution judge got nothing back; defaulting the pair to NEW");
+    return { type: "NEW" };
+  }
   try {
     const data = parseJsonLoose(raw);
     const type = VALID_EVOLUTION_TYPES.has(data.type) ? data.type : "NEW";
@@ -1338,7 +1376,7 @@ async function llmEvolutionJudge(oldContent, newContent) {
       mergedContent: typeof data.mergedContent === "string" ? data.mergedContent : void 0
     };
   } catch (e) {
-    console.error(`[amem] llmEvolutionJudge parse failed: ${e.message}`);
+    warn(`[amem] llmEvolutionJudge parse failed: ${e.message}`);
     return { type: "NEW" };
   }
 }
@@ -1377,7 +1415,7 @@ ${linkedStr}`;
       tagsToUpdate: Array.isArray(data.tags_to_update) ? data.tags_to_update.map(String) : []
     };
   } catch (e) {
-    console.error(`[amem] Evolution parse failed: ${e.message}`);
+    warn(`[amem] Evolution parse failed: ${e.message}`);
     return { tags: null, context: null, shouldStrengthen: false, suggestedConnections: [], tagsToUpdate: [] };
   }
 }
@@ -1389,9 +1427,15 @@ async function llmConflictScan(contents) {
     if (!raw) return [];
     const cleaned = stripReasoning(raw);
     const match = cleaned.match(/\[[\s\S]*\]/);
-    if (!match) return [];
+    if (!match) {
+      warn("[amem] llmConflictScan found no array in the response; this sweep reports no pairs");
+      return [];
+    }
     const parsed = JSON.parse(match[0]);
-    if (!Array.isArray(parsed)) return [];
+    if (!Array.isArray(parsed)) {
+      warn("[amem] llmConflictScan parsed a non-array; this sweep reports no pairs");
+      return [];
+    }
     const pairs = [];
     const seen = /* @__PURE__ */ new Set();
     for (const item of parsed) {
@@ -1415,22 +1459,11 @@ async function llmConflictScan(contents) {
     }
     return pairs;
   } catch (e) {
-    console.error(`[amem] llmConflictScan failed: ${e.message}`);
+    warn(`[amem] llmConflictScan failed: ${e.message}`);
     return [];
   }
 }
 
-// src/config.ts
-import * as os from "os";
-import * as path from "path";
-var _dataDir = process.env.AMEM_DATA_DIR || path.join(os.homedir(), ".amem");
-function configure(opts) {
-  if (opts.dataDir) _dataDir = opts.dataDir;
-}
-function getDataDir() {
-  return _dataDir;
-}
-
 // src/memory.ts
 import { v4 as uuidv4 } from "uuid";
 import { createHash } from "crypto";
@@ -1746,7 +1779,7 @@ async function addMemory(content, agentId = "main", opts) {
       }
     }
   } catch (e) {
-    console.error(`[warn] Link/Evolution phase failed: ${e.message}`);
+    warn(`[amem] link/evolution phase failed: ${e.message}`);
   }
   console.log(`[done] Note added: ${note.id}`);
   return note.id;
@@ -1871,7 +1904,12 @@ async function searchMemory(query, topK = 5, agentId = "main", opts) {
       keywords: note.keywords,
       links: note.links,
       timestamp: note.timestamp,
-      similarity: embSimMap.get(id) ?? bfsSimMap.get(id) ?? 0,
+      // Neither map covers a note that got here on BM25 alone: it was never in
+      // the dense results, and it was not expanded into. That is a real cosine
+      // nobody had measured, not a zero — and reporting 0 made a lexical match
+      // look like the least relevant row in the list. Both vectors are already
+      // in hand, so measuring it is one dot product.
+      similarity: embSimMap.get(id) ?? bfsSimMap.get(id) ?? cosineSimilarity(queryEmbedding, note.embedding),
       rrf: rrfMap.get(id) ?? 0,
       via,
       topics: note.topics ?? [],
@@ -2007,8 +2045,8 @@ async function consolidateMemories(agentId, logger, storageCtx) {
   const ctx = storageCtx ?? defaultCtx();
   const log = {
     info: (msg) => logger ? logger.info(msg) : console.log(msg),
-    warn: (msg) => logger ? logger.warn(msg) : console.warn(msg),
-    error: (msg) => logger ? logger.error(msg) : console.error(msg)
+    warn: (msg) => logger ? logger.warn(msg) : warn(msg),
+    error: (msg) => logger ? logger.error(msg) : warn(msg)
   };
   log.info(`[Consolidation] Starting consolidation for agentId: ${agentId}`);
   const rawNotes = await ctx.listNotes(agentId);
@@ -2202,7 +2240,7 @@ async function migrateCollection(opts) {
   const refreshFields = opts.refreshFields !== false;
   const dryRun = opts.dryRun !== false;
   const log = opts.logger?.info ?? ((m) => console.log(m));
-  const warn = opts.logger?.warn ?? ((m) => console.warn(m));
+  const warn2 = opts.logger?.warn ?? warn;
   if (from === to) throw new Error(`migrate: source and target are the same collection ("${from}")`);
   const model = getEmbeddingModel();
   const targetDim = await getEmbeddingDim();
@@ -2270,7 +2308,7 @@ async function migrateCollection(opts) {
         if (!note.context) note.context = built.context;
         refreshed++;
       } catch (e) {
-        warn(`[migrate] re-extract failed for ${note.id.slice(0, 8)} \u2014 keeping as-is: ${e.message}`);
+        warn2(`[migrate] re-extract failed for ${note.id.slice(0, 8)} \u2014 keeping as-is: ${e.message}`);
       }
     }
     const point = noteToPoint({ ...note, embedding: await encode(buildEmbedText(note)) });
@@ -2283,7 +2321,7 @@ async function migrateCollection(opts) {
   await flush();
   const finalCount = await countPointsRaw(to);
   if (finalCount !== notes.length) {
-    warn(`[migrate] target holds ${finalCount} point(s) but the source had ${notes.length} \u2014 check before switching`);
+    warn2(`[migrate] target holds ${finalCount} point(s) but the source had ${notes.length} \u2014 check before switching`);
   }
   log(`[migrate] done: ${migrated} written, ${alreadyDone.size} already present, ${refreshed} re-extracted.`);
   return {
@@ -2336,6 +2374,7 @@ export {
   SYSTEM_ACTOR,
   canWrite,
   canRead,
+  configure,
   DEFAULT_EMBEDDING_MODEL,
   LEGACY_DEFAULT_EMBEDDING_MODEL,
   LEGACY_DEFAULT_DIM,
@@ -2367,7 +2406,6 @@ export {
   patchNotePayload,
   configureLlm,
   llmCrudDecision,
-  configure,
   checkQuality,
   addMemory,
   addEpisodic,
@@ -2379,4 +2417,4 @@ export {
   migrateCollection,
   switchToMigrated
 };
-//# sourceMappingURL=chunk-XEMQZNLD.js.map
+//# sourceMappingURL=chunk-Z7G7KOJK.js.map