mcpscraper-memory-db 1.0.0 → 1.0.1

package/dist/rag/query-tool.d.ts

@@ -1,4 +1,8 @@
-import { RagTool } from './vector-search.js';
+import { Tool } from '@mastra/core/tools';
+
+type RagTool<TInput, TOutput> = Tool<TInput, TOutput, unknown, unknown> & {
+    execute: NonNullable<Tool<TInput, TOutput, unknown, unknown>['execute']>;
+};
 
 declare const vectorQueryTool: RagTool<{
     queryText: string;

package/dist/rag/query-tool.js

@@ -1,7 +1,7 @@
 import { createVectorQueryTool, createGraphRAGTool } from "@mastra/rag";
 import { embed } from "ai";
-import { jinaEmbedder } from "./embedder";
-import { memoryVectors, INDEX_NAME } from "./vector-store";
+import { jinaEmbedder } from "./embedder.js";
+import { memoryVectors, INDEX_NAME } from "./vector-store.js";
 const vectorQueryTool = createVectorQueryTool({
   vectorStoreName: "memoryVectors",
   indexName: INDEX_NAME,

package/dist/rag/query-tool.js.map

@@ -1 +1 @@
-{"version":3,"sources":["../../src/rag/query-tool.ts"],"sourcesContent":["import { createVectorQueryTool, createGraphRAGTool } from '@mastra/rag'\nimport { embed } from 'ai'\nimport { jinaEmbedder } from './embedder'\nimport { memoryVectors, INDEX_NAME } from './vector-store'\n\nexport const vectorQueryTool = createVectorQueryTool({\n  vectorStoreName: 'memoryVectors',\n  indexName: INDEX_NAME,\n  model: jinaEmbedder,\n})\n\nexport const graphQueryTool = createGraphRAGTool({\n  vectorStoreName: 'memoryVectors',\n  indexName: INDEX_NAME,\n  model: jinaEmbedder,\n  graphOptions: { dimension: 1024 },\n})\n\nexport async function embedQuery(query: string): Promise<number[]> {\n  const { embedding } = await embed({ model: jinaEmbedder, value: query })\n  return embedding\n}\n\nexport async function queryWithEmbedding(args: {\n  embedding: number[]\n  vault: string | string[]\n  topK?: number\n  docId?: string\n}): Promise<{ results: { text: string; source: string; score: number; vault?: string }[] }> {\n  const vaultFilter = Array.isArray(args.vault) ? { $in: args.vault } : args.vault\n  const hits = await memoryVectors.query({\n    indexName: INDEX_NAME,\n    queryVector: args.embedding,\n    topK: args.topK ?? 8,\n    filter: args.docId ? { vault: vaultFilter, docId: args.docId } : { vault: vaultFilter },\n  })\n  const results = hits.map((h) => ({\n    text: String(h.metadata?.text ?? ''),\n    source: String(h.metadata?.source ?? ''),\n    score: Number(h.score ?? 0),\n    vault: h.metadata?.vault ? String(h.metadata.vault) : undefined,\n  }))\n  return { results }\n}\n\nexport async function semanticSearch(args: {\n  vault: string | string[]\n  query: string\n  topK?: number\n  docId?: string\n}): Promise<{ results: { text: string; source: string; score: number; vault?: string }[] }> {\n  const embedding = await embedQuery(args.query)\n  return queryWithEmbedding({ embedding, vault: args.vault, topK: args.topK, docId: args.docId })\n}\n"],"mappings":"AAAA,SAAS,uBAAuB,0BAA0B;AAC1D,SAAS,aAAa;AACtB,SAAS,oBAAoB;AAC7B,SAAS,eAAe,kBAAkB;AAEnC,MAAM,kBAAkB,sBAAsB;AAAA,EACnD,iBAAiB;AAAA,EACjB,WAAW;AAAA,EACX,OAAO;AACT,CAAC;AAEM,MAAM,iBAAiB,mBAAmB;AAAA,EAC/C,iBAAiB;AAAA,EACjB,WAAW;AAAA,EACX,OAAO;AAAA,EACP,cAAc,EAAE,WAAW,KAAK;AAClC,CAAC;AAED,eAAsB,WAAW,OAAkC;AACjE,QAAM,EAAE,UAAU,IAAI,MAAM,MAAM,EAAE,OAAO,cAAc,OAAO,MAAM,CAAC;AACvE,SAAO;AACT;AAEA,eAAsB,mBAAmB,MAKmD;AAC1F,QAAM,cAAc,MAAM,QAAQ,KAAK,KAAK,IAAI,EAAE,KAAK,KAAK,MAAM,IAAI,KAAK;AAC3E,QAAM,OAAO,MAAM,cAAc,MAAM;AAAA,IACrC,WAAW;AAAA,IACX,aAAa,KAAK;AAAA,IAClB,MAAM,KAAK,QAAQ;AAAA,IACnB,QAAQ,KAAK,QAAQ,EAAE,OAAO,aAAa,OAAO,KAAK,MAAM,IAAI,EAAE,OAAO,YAAY;AAAA,EACxF,CAAC;AACD,QAAM,UAAU,KAAK,IAAI,CAAC,OAAO;AAAA,IAC/B,MAAM,OAAO,EAAE,UAAU,QAAQ,EAAE;AAAA,IACnC,QAAQ,OAAO,EAAE,UAAU,UAAU,EAAE;AAAA,IACvC,OAAO,OAAO,EAAE,SAAS,CAAC;AAAA,IAC1B,OAAO,EAAE,UAAU,QAAQ,OAAO,EAAE,SAAS,KAAK,IAAI;AAAA,EACxD,EAAE;AACF,SAAO,EAAE,QAAQ;AACnB;AAEA,eAAsB,eAAe,MAKuD;AAC1F,QAAM,YAAY,MAAM,WAAW,KAAK,KAAK;AAC7C,SAAO,mBAAmB,EAAE,WAAW,OAAO,KAAK,OAAO,MAAM,KAAK,MAAM,OAAO,KAAK,MAAM,CAAC;AAChG;","names":[]}
+{"version":3,"sources":["../../src/rag/query-tool.ts"],"sourcesContent":["import { createVectorQueryTool, createGraphRAGTool } from '@mastra/rag'\nimport { embed } from 'ai'\nimport { jinaEmbedder } from './embedder.js'\nimport { memoryVectors, INDEX_NAME } from './vector-store.js'\n\nexport const vectorQueryTool = createVectorQueryTool({\n  vectorStoreName: 'memoryVectors',\n  indexName: INDEX_NAME,\n  model: jinaEmbedder,\n})\n\nexport const graphQueryTool = createGraphRAGTool({\n  vectorStoreName: 'memoryVectors',\n  indexName: INDEX_NAME,\n  model: jinaEmbedder,\n  graphOptions: { dimension: 1024 },\n})\n\nexport async function embedQuery(query: string): Promise<number[]> {\n  const { embedding } = await embed({ model: jinaEmbedder, value: query })\n  return embedding\n}\n\nexport async function queryWithEmbedding(args: {\n  embedding: number[]\n  vault: string | string[]\n  topK?: number\n  docId?: string\n}): Promise<{ results: { text: string; source: string; score: number; vault?: string }[] }> {\n  const vaultFilter = Array.isArray(args.vault) ? { $in: args.vault } : args.vault\n  const hits = await memoryVectors.query({\n    indexName: INDEX_NAME,\n    queryVector: args.embedding,\n    topK: args.topK ?? 8,\n    filter: args.docId ? { vault: vaultFilter, docId: args.docId } : { vault: vaultFilter },\n  })\n  const results = hits.map((h) => ({\n    text: String(h.metadata?.text ?? ''),\n    source: String(h.metadata?.source ?? ''),\n    score: Number(h.score ?? 0),\n    vault: h.metadata?.vault ? String(h.metadata.vault) : undefined,\n  }))\n  return { results }\n}\n\nexport async function semanticSearch(args: {\n  vault: string | string[]\n  query: string\n  topK?: number\n  docId?: string\n}): Promise<{ results: { text: string; source: string; score: number; vault?: string }[] }> {\n  const embedding = await embedQuery(args.query)\n  return queryWithEmbedding({ embedding, vault: args.vault, topK: args.topK, docId: args.docId })\n}\n"],"mappings":"AAAA,SAAS,uBAAuB,0BAA0B;AAC1D,SAAS,aAAa;AACtB,SAAS,oBAAoB;AAC7B,SAAS,eAAe,kBAAkB;AAEnC,MAAM,kBAAkB,sBAAsB;AAAA,EACnD,iBAAiB;AAAA,EACjB,WAAW;AAAA,EACX,OAAO;AACT,CAAC;AAEM,MAAM,iBAAiB,mBAAmB;AAAA,EAC/C,iBAAiB;AAAA,EACjB,WAAW;AAAA,EACX,OAAO;AAAA,EACP,cAAc,EAAE,WAAW,KAAK;AAClC,CAAC;AAED,eAAsB,WAAW,OAAkC;AACjE,QAAM,EAAE,UAAU,IAAI,MAAM,MAAM,EAAE,OAAO,cAAc,OAAO,MAAM,CAAC;AACvE,SAAO;AACT;AAEA,eAAsB,mBAAmB,MAKmD;AAC1F,QAAM,cAAc,MAAM,QAAQ,KAAK,KAAK,IAAI,EAAE,KAAK,KAAK,MAAM,IAAI,KAAK;AAC3E,QAAM,OAAO,MAAM,cAAc,MAAM;AAAA,IACrC,WAAW;AAAA,IACX,aAAa,KAAK;AAAA,IAClB,MAAM,KAAK,QAAQ;AAAA,IACnB,QAAQ,KAAK,QAAQ,EAAE,OAAO,aAAa,OAAO,KAAK,MAAM,IAAI,EAAE,OAAO,YAAY;AAAA,EACxF,CAAC;AACD,QAAM,UAAU,KAAK,IAAI,CAAC,OAAO;AAAA,IAC/B,MAAM,OAAO,EAAE,UAAU,QAAQ,EAAE;AAAA,IACnC,QAAQ,OAAO,EAAE,UAAU,UAAU,EAAE;AAAA,IACvC,OAAO,OAAO,EAAE,SAAS,CAAC;AAAA,IAC1B,OAAO,EAAE,UAAU,QAAQ,OAAO,EAAE,SAAS,KAAK,IAAI;AAAA,EACxD,EAAE;AACF,SAAO,EAAE,QAAQ;AACnB;AAEA,eAAsB,eAAe,MAKuD;AAC1F,QAAM,YAAY,MAAM,WAAW,KAAK,KAAK;AAC7C,SAAO,mBAAmB,EAAE,WAAW,OAAO,KAAK,OAAO,MAAM,KAAK,MAAM,OAAO,KAAK,MAAM,CAAC;AAChG;","names":[]}

package/dist/rag/vector-store.js

@@ -1,5 +1,5 @@
 import { PgVector } from "@mastra/pg";
-import { EMBED_DIM } from "./embedder";
+import { EMBED_DIM } from "./embedder.js";
 const INDEX_NAME = "memory";
 const memoryVectors = new PgVector({
   id: "memoryVectors",

package/dist/rag/vector-store.js.map

@@ -1 +1 @@
-{"version":3,"sources":["../../src/rag/vector-store.ts"],"sourcesContent":["import { PgVector } from '@mastra/pg';\nimport { EMBED_DIM } from './embedder';\n\nexport const INDEX_NAME = 'memory';\n\nexport const memoryVectors = new PgVector({\n  id: 'memoryVectors',\n  connectionString: process.env.DATABASE_URL ?? '',\n});\n\nexport async function ensureIndex(): Promise<void> {\n  try {\n    await memoryVectors.createIndex({ indexName: INDEX_NAME, dimension: EMBED_DIM });\n  } catch (error) {\n    const message = error instanceof Error ? error.message : String(error);\n    const normalized = message.toLowerCase();\n    const alreadyExists =\n      normalized.includes('already exists') || normalized.includes('duplicate');\n    if (!alreadyExists) {\n      throw error;\n    }\n  }\n}\n"],"mappings":"AAAA,SAAS,gBAAgB;AACzB,SAAS,iBAAiB;AAEnB,MAAM,aAAa;AAEnB,MAAM,gBAAgB,IAAI,SAAS;AAAA,EACxC,IAAI;AAAA,EACJ,kBAAkB,QAAQ,IAAI,gBAAgB;AAChD,CAAC;AAED,eAAsB,cAA6B;AACjD,MAAI;AACF,UAAM,cAAc,YAAY,EAAE,WAAW,YAAY,WAAW,UAAU,CAAC;AAAA,EACjF,SAAS,OAAO;AACd,UAAM,UAAU,iBAAiB,QAAQ,MAAM,UAAU,OAAO,KAAK;AACrE,UAAM,aAAa,QAAQ,YAAY;AACvC,UAAM,gBACJ,WAAW,SAAS,gBAAgB,KAAK,WAAW,SAAS,WAAW;AAC1E,QAAI,CAAC,eAAe;AAClB,YAAM;AAAA,IACR;AAAA,EACF;AACF;","names":[]}
+{"version":3,"sources":["../../src/rag/vector-store.ts"],"sourcesContent":["import { PgVector } from '@mastra/pg';\nimport { EMBED_DIM } from './embedder.js';\n\nexport const INDEX_NAME = 'memory';\n\nexport const memoryVectors = new PgVector({\n  id: 'memoryVectors',\n  connectionString: process.env.DATABASE_URL ?? '',\n});\n\nexport async function ensureIndex(): Promise<void> {\n  try {\n    await memoryVectors.createIndex({ indexName: INDEX_NAME, dimension: EMBED_DIM });\n  } catch (error) {\n    const message = error instanceof Error ? error.message : String(error);\n    const normalized = message.toLowerCase();\n    const alreadyExists =\n      normalized.includes('already exists') || normalized.includes('duplicate');\n    if (!alreadyExists) {\n      throw error;\n    }\n  }\n}\n"],"mappings":"AAAA,SAAS,gBAAgB;AACzB,SAAS,iBAAiB;AAEnB,MAAM,aAAa;AAEnB,MAAM,gBAAgB,IAAI,SAAS;AAAA,EACxC,IAAI;AAAA,EACJ,kBAAkB,QAAQ,IAAI,gBAAgB;AAChD,CAAC;AAED,eAAsB,cAA6B;AACjD,MAAI;AACF,UAAM,cAAc,YAAY,EAAE,WAAW,YAAY,WAAW,UAAU,CAAC;AAAA,EACjF,SAAS,OAAO;AACd,UAAM,UAAU,iBAAiB,QAAQ,MAAM,UAAU,OAAO,KAAK;AACrE,UAAM,aAAa,QAAQ,YAAY;AACvC,UAAM,gBACJ,WAAW,SAAS,gBAAgB,KAAK,WAAW,SAAS,WAAW;AAC1E,QAAI,CAAC,eAAe;AAClB,YAAM;AAAA,IACR;AAAA,EACF;AACF;","names":[]}

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "mcpscraper-memory-db",
-  "version": "1.0.0",
+  "version": "1.0.1",
   "description": "Shared data-access layer for the mcpscraper memory product — Postgres db/*.ts, the auth/crypto/vault lib/*.ts modules it depends on, and the Jina/pgvector rag/*.ts indexing layer. Consumed by mcp-memory and mcp-scraper-scheduler so both stay on one implementation instead of hand-copied duplicates.",
   "type": "module",
   "license": "UNLICENSED",