@akashabot/openclaw-mem 0.3.0 → 0.5.0

package/dist/cli.js

@@ -5,7 +5,11 @@ import { v4 as uuidv4 } from 'uuid';
 import * as core from '@akashabot/openclaw-memory-offline-core';
 const { addItem, hybridSearch, hybridSearchFiltered, initSchema, openDb, runMigrations, searchItems, getMemoriesByEntity, getMemoriesBySession, getMemoriesByProcess, listEntities, listSessions, 
 // Phase 2: Facts
-insertFact, getFactsBySubject, getFactsByPredicate, searchFacts, getAllFacts, listSubjects, listPredicates, deleteFact, extractFactsSimple, } = core;
+insertFact, getFactsBySubject, getFactsByPredicate, searchFacts, getAllFacts, listSubjects, listPredicates, deleteFact, extractFactsSimple, 
+// Phase 3: Knowledge Graph
+getEntityGraph, getRelatedEntities, findPaths, getGraphStats, exportGraphJson, searchEntities, 
+// Phase 3: Embedding Optimizations
+getEmbeddingStats, quantizeF32ToF16, dequantizeF16ToF32, cosineSimilarity, benchmarkCosineSimilarity, runEmbeddingBenchmark, } = core;
 const program = new Command();
 program
     .name('openclaw-mem')
@@ -314,4 +318,118 @@ program
     const facts = extractFactsSimple(input);
     console.log(JSON.stringify({ ok: true, count: facts.length, facts }));
 });
+// ============================================================================
+// Phase 3: Knowledge Graph commands
+// ============================================================================
+program
+    .command('graph-stats')
+    .description('Get statistics about the knowledge graph')
+    .action(() => {
+    withDb((dbPath) => {
+        const db = openDb(dbPath);
+        initSchema(db);
+        const stats = getGraphStats(db);
+        console.log(JSON.stringify({ ok: true, stats }, null, 2));
+    });
+});
+program
+    .command('graph-entity <entity>')
+    .description('Get all facts connected to an entity (as subject or object)')
+    .action((entity) => {
+    withDb((dbPath) => {
+        const db = openDb(dbPath);
+        initSchema(db);
+        const edges = getEntityGraph(db, entity);
+        console.log(JSON.stringify({ ok: true, entity, count: edges.length, edges }));
+    });
+});
+program
+    .command('graph-related <entity>')
+    .description('Get all entities directly connected to an entity')
+    .action((entity) => {
+    withDb((dbPath) => {
+        const db = openDb(dbPath);
+        initSchema(db);
+        const related = getRelatedEntities(db, entity);
+        console.log(JSON.stringify({ ok: true, entity, count: related.length, related }));
+    });
+});
+program
+    .command('graph-path <fromEntity> <toEntity>')
+    .description('Find paths between two entities in the knowledge graph')
+    .option('--max-depth <n>', 'Maximum path depth (default 4)', '4')
+    .option('--max-paths <n>', 'Maximum number of paths to return (default 5)', '5')
+    .action((fromEntity, toEntity, cmdOpts) => {
+    withDb((dbPath) => {
+        const db = openDb(dbPath);
+        initSchema(db);
+        const maxDepth = Math.max(1, Math.min(10, Number(cmdOpts.maxDepth ?? 4)));
+        const maxPaths = Math.max(1, Math.min(20, Number(cmdOpts.maxPaths ?? 5)));
+        const paths = findPaths(db, fromEntity, toEntity, maxDepth, maxPaths);
+        console.log(JSON.stringify({ ok: true, from: fromEntity, to: toEntity, count: paths.length, paths }));
+    });
+});
+program
+    .command('graph-export [outputFile]')
+    .description('Export the knowledge graph as JSON (for visualization)')
+    .option('--limit <n>', 'Max edges to export (default 1000)', '1000')
+    .option('--min-confidence <n>', 'Minimum confidence threshold (default 0)', '0')
+    .option('--entity <entity>', 'Export only subgraph around this entity')
+    .action((outputFile, cmdOpts) => {
+    withDb((dbPath) => {
+        const db = openDb(dbPath);
+        initSchema(db);
+        const graph = exportGraphJson(db, {
+            limit: Number(cmdOpts.limit ?? 1000),
+            minConfidence: Number(cmdOpts.minConfidence ?? 0),
+            entity: cmdOpts.entity,
+        });
+        const output = JSON.stringify({ ok: true, graph }, null, 2);
+        if (outputFile) {
+            fs.writeFileSync(outputFile, output);
+            console.log(JSON.stringify({ ok: true, file: outputFile, nodes: graph.nodes.length, edges: graph.edges.length }));
+        }
+        else {
+            console.log(output);
+        }
+    });
+});
+program
+    .command('search-entities <pattern>')
+    .description('Search for entities matching a pattern')
+    .option('--limit <n>', 'Max results (default 50)', '50')
+    .action((pattern, cmdOpts) => {
+    withDb((dbPath) => {
+        const db = openDb(dbPath);
+        initSchema(db);
+        const limit = Math.max(1, Math.min(500, Number(cmdOpts.limit ?? 50)));
+        const entities = searchEntities(db, pattern, limit);
+        console.log(JSON.stringify({ ok: true, pattern, count: entities.length, entities }));
+    });
+});
+// ============================================================================
+// Phase 3: Embedding Optimization commands
+// ============================================================================
+program
+    .command('embedding-stats')
+    .description('Get statistics about stored embeddings')
+    .action(() => {
+    withDb((dbPath) => {
+        const db = openDb(dbPath);
+        initSchema(db);
+        const stats = getEmbeddingStats(db);
+        const sizeMB = Math.round((stats.totalSizeBytes / (1024 * 1024)) * 100) / 100;
+        console.log(JSON.stringify({ ok: true, ...stats, sizeMB }, null, 2));
+    });
+});
+program
+    .command('benchmark')
+    .description('Run embedding operation benchmarks')
+    .option('--dims <n>', 'Vector dimensions (default 1024)', '1024')
+    .option('--iterations <n>', 'Iterations (default 5000)', '5000')
+    .action((cmdOpts) => {
+    console.log('Running embedding benchmarks...\n');
+    const results = runEmbeddingBenchmark();
+    console.log(JSON.stringify({ ok: true, benchmark: results }, null, 2));
+});
 program.parse();

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@akashabot/openclaw-mem",
-  "version": "0.3.0",
+  "version": "0.5.0",
   "type": "module",
   "description": "CLI for OpenClaw Offline Memory (SQLite FTS + optional embeddings)",
   "bin": {
@@ -17,7 +17,7 @@
   "dependencies": {
     "commander": "^12.1.0",
     "uuid": "^10.0.0",
-    "@akashabot/openclaw-memory-offline-core": "^0.3.0"
+    "@akashabot/openclaw-memory-offline-core": "^0.5.0"
   },
   "devDependencies": {
     "@types/uuid": "^10.0.0"

package/src/cli.ts

@@ -27,6 +27,20 @@ const {
   listPredicates,
   deleteFact,
   extractFactsSimple,
+  // Phase 3: Knowledge Graph
+  getEntityGraph,
+  getRelatedEntities,
+  findPaths,
+  getGraphStats,
+  exportGraphJson,
+  searchEntities,
+  // Phase 3: Embedding Optimizations
+  getEmbeddingStats,
+  quantizeF32ToF16,
+  dequantizeF16ToF32,
+  cosineSimilarity,
+  benchmarkCosineSimilarity,
+  runEmbeddingBenchmark,
 } = core;
 
 const program = new Command();
@@ -381,4 +395,128 @@ program
     console.log(JSON.stringify({ ok: true, count: facts.length, facts }));
   });
 
+// ============================================================================
+// Phase 3: Knowledge Graph commands
+// ============================================================================
+
+program
+  .command('graph-stats')
+  .description('Get statistics about the knowledge graph')
+  .action(() => {
+    withDb((dbPath) => {
+      const db = openDb(dbPath);
+      initSchema(db);
+      const stats = getGraphStats(db);
+      console.log(JSON.stringify({ ok: true, stats }, null, 2));
+    });
+  });
+
+program
+  .command('graph-entity <entity>')
+  .description('Get all facts connected to an entity (as subject or object)')
+  .action((entity: string) => {
+    withDb((dbPath) => {
+      const db = openDb(dbPath);
+      initSchema(db);
+      const edges = getEntityGraph(db, entity);
+      console.log(JSON.stringify({ ok: true, entity, count: edges.length, edges }));
+    });
+  });
+
+program
+  .command('graph-related <entity>')
+  .description('Get all entities directly connected to an entity')
+  .action((entity: string) => {
+    withDb((dbPath) => {
+      const db = openDb(dbPath);
+      initSchema(db);
+      const related = getRelatedEntities(db, entity);
+      console.log(JSON.stringify({ ok: true, entity, count: related.length, related }));
+    });
+  });
+
+program
+  .command('graph-path <fromEntity> <toEntity>')
+  .description('Find paths between two entities in the knowledge graph')
+  .option('--max-depth <n>', 'Maximum path depth (default 4)', '4')
+  .option('--max-paths <n>', 'Maximum number of paths to return (default 5)', '5')
+  .action((fromEntity: string, toEntity: string, cmdOpts) => {
+    withDb((dbPath) => {
+      const db = openDb(dbPath);
+      initSchema(db);
+      const maxDepth = Math.max(1, Math.min(10, Number(cmdOpts.maxDepth ?? 4)));
+      const maxPaths = Math.max(1, Math.min(20, Number(cmdOpts.maxPaths ?? 5)));
+      const paths = findPaths(db, fromEntity, toEntity, maxDepth, maxPaths);
+      console.log(JSON.stringify({ ok: true, from: fromEntity, to: toEntity, count: paths.length, paths }));
+    });
+  });
+
+program
+  .command('graph-export [outputFile]')
+  .description('Export the knowledge graph as JSON (for visualization)')
+  .option('--limit <n>', 'Max edges to export (default 1000)', '1000')
+  .option('--min-confidence <n>', 'Minimum confidence threshold (default 0)', '0')
+  .option('--entity <entity>', 'Export only subgraph around this entity')
+  .action((outputFile: string | undefined, cmdOpts) => {
+    withDb((dbPath) => {
+      const db = openDb(dbPath);
+      initSchema(db);
+      const graph = exportGraphJson(db, {
+        limit: Number(cmdOpts.limit ?? 1000),
+        minConfidence: Number(cmdOpts.minConfidence ?? 0),
+        entity: cmdOpts.entity,
+      });
+
+      const output = JSON.stringify({ ok: true, graph }, null, 2);
+      if (outputFile) {
+        fs.writeFileSync(outputFile, output);
+        console.log(JSON.stringify({ ok: true, file: outputFile, nodes: graph.nodes.length, edges: graph.edges.length }));
+      } else {
+        console.log(output);
+      }
+    });
+  });
+
+program
+  .command('search-entities <pattern>')
+  .description('Search for entities matching a pattern')
+  .option('--limit <n>', 'Max results (default 50)', '50')
+  .action((pattern: string, cmdOpts) => {
+    withDb((dbPath) => {
+      const db = openDb(dbPath);
+      initSchema(db);
+      const limit = Math.max(1, Math.min(500, Number(cmdOpts.limit ?? 50)));
+      const entities = searchEntities(db, pattern, limit);
+      console.log(JSON.stringify({ ok: true, pattern, count: entities.length, entities }));
+    });
+  });
+
+// ============================================================================
+// Phase 3: Embedding Optimization commands
+// ============================================================================
+
+program
+  .command('embedding-stats')
+  .description('Get statistics about stored embeddings')
+  .action(() => {
+    withDb((dbPath) => {
+      const db = openDb(dbPath);
+      initSchema(db);
+      const stats = getEmbeddingStats(db);
+      const sizeMB = Math.round((stats.totalSizeBytes / (1024 * 1024)) * 100) / 100;
+      console.log(JSON.stringify({ ok: true, ...stats, sizeMB }, null, 2));
+    });
+  });
+
+program
+  .command('benchmark')
+  .description('Run embedding operation benchmarks')
+  .option('--dims <n>', 'Vector dimensions (default 1024)', '1024')
+  .option('--iterations <n>', 'Iterations (default 5000)', '5000')
+  .action((cmdOpts) => {
+    console.log('Running embedding benchmarks...\n');
+    const results = runEmbeddingBenchmark();
+    console.log(JSON.stringify({ ok: true, benchmark: results }, null, 2));
+  });
+
 program.parse();