promptgraph-mcp 2.8.1 → 2.8.3

package/db.js

@@ -1,157 +1,157 @@
-import Database from 'better-sqlite3';
-import path from 'path';
-import fs from 'fs';
-import { PROMPTGRAPH_DIR } from './config.js';
-
-const DB_PATH = path.join(PROMPTGRAPH_DIR, 'promptgraph.db');
-
-let _db = null;
-
-const MIGRATIONS = [
-  {
-    version: 1, description: 'initial schema',
-    up: db => db.exec(`
-      CREATE TABLE IF NOT EXISTS skills (
-        id TEXT PRIMARY KEY,
-        name TEXT NOT NULL,
-        description TEXT,
-        path TEXT NOT NULL,
-        source TEXT NOT NULL,
-        content TEXT NOT NULL,
-        hash TEXT
-      );
-      CREATE TABLE IF NOT EXISTS chunks (
-        id INTEGER PRIMARY KEY AUTOINCREMENT,
-        skill_id TEXT NOT NULL,
-        chunk_index INTEGER NOT NULL,
-        text TEXT NOT NULL,
-        embedding BLOB NOT NULL,
-        UNIQUE(skill_id, chunk_index)
-      );
-      CREATE TABLE IF NOT EXISTS edges (
-        from_skill TEXT NOT NULL,
-        to_skill TEXT NOT NULL,
-        PRIMARY KEY (from_skill, to_skill)
-      );
-      CREATE TABLE IF NOT EXISTS ratings (
-        skill_id TEXT PRIMARY KEY,
-        uses INTEGER DEFAULT 0,
-        success INTEGER DEFAULT 0,
-        fail INTEGER DEFAULT 0
-      );
-      CREATE VIRTUAL TABLE IF NOT EXISTS skills_fts USING fts5(
-        id UNINDEXED, name, description, content,
-        content='skills', content_rowid='rowid'
-      );
-    `),
-  },
-  {
-    version: 2, description: 'add hash column',
-    up: db => {
-      const cols = db.pragma('table_info(skills)').map(c => c.name);
-      if (!cols.includes('hash')) db.exec('ALTER TABLE skills ADD COLUMN hash TEXT');
-    },
-  },
-  {
-    version: 3, description: 'add registry metadata columns',
-    up: db => {
-      const cols = db.pragma('table_info(skills)').map(c => c.name);
-      for (const [col, def] of [
-        ['version', 'TEXT'],
-        ['author', 'TEXT'],
-        ['license', 'TEXT'],
-        ['updated_at', 'TEXT'],
-        ['downloads', 'INTEGER DEFAULT 0'],
-        ['verified', 'INTEGER DEFAULT 0'],
-        ['trust_level', "TEXT DEFAULT 'unknown'"],
-        ['rating', 'REAL DEFAULT 0'],
-        ['rating_count', 'INTEGER DEFAULT 0'],
-        ['popularity', 'REAL DEFAULT 0'],
-        ['last_update', 'TEXT'],
-      ]) {
-        if (!cols.includes(col)) db.exec(`ALTER TABLE skills ADD COLUMN ${col} ${def}`);
-      }
-    },
-  },
-  {
-    version: 4, description: 'registry_entries table',
-    up: db => db.exec(`
-      CREATE TABLE IF NOT EXISTS registry_entries (
-        id TEXT PRIMARY KEY,
-        trust_level TEXT DEFAULT 'unknown',
-        downloads INTEGER DEFAULT 0,
-        rating REAL DEFAULT 0,
-        rating_count INTEGER DEFAULT 0,
-        popularity REAL DEFAULT 0,
-        last_update TEXT
-      );
-    `),
-  },
-  {
-    version: 5, description: 'convert TEXT embeddings to Float32 BLOB',
-    up: db => {
-      const textEmbeddings = db.prepare("SELECT COUNT(*) as n FROM chunks WHERE typeof(embedding) = 'text'").get();
-      if (textEmbeddings?.n > 0) {
-        const rows = db.prepare("SELECT rowid, embedding FROM chunks WHERE typeof(embedding) = 'text'").all();
-        const upd = db.prepare('UPDATE chunks SET embedding = ? WHERE rowid = ?');
-        db.transaction(() => {
-          for (const row of rows) {
-            const vec = JSON.parse(row.embedding);
-            upd.run(Buffer.from(new Float32Array(vec).buffer), row.rowid);
-          }
-        })();
-        console.error(`[PromptGraph] Migrated ${textEmbeddings.n} embeddings TEXT→BLOB`);
-      }
-    },
-  },
-]
-
-function getSchemaVersion(db) {
-  try {
-    return db.prepare('SELECT MAX(version) as v FROM _schema_version').get().v || 0
-  } catch {
-    return 0
-  }
-}
-
-function runMigrations(db) {
-  const current = getSchemaVersion(db)
-  const pending = MIGRATIONS.filter(m => m.version > current).sort((a, b) => a.version - b.version)
-  if (!pending.length) return
-
-  db.exec(`CREATE TABLE IF NOT EXISTS _schema_version (version INTEGER PRIMARY KEY, description TEXT, applied_at TEXT)`)
-
-  for (const migration of pending) {
-    db.transaction(() => {
-      migration.up(db)
-      db.prepare('INSERT INTO _schema_version (version, description, applied_at) VALUES (?, ?, ?)').run(
-        migration.version, migration.description, new Date().toISOString()
-      )
-    })()
-    console.error(`[PromptGraph] DB migrated to v${migration.version}: ${migration.description}`)
-  }
-}
-
-export function getDb() {
-  if (_db) return _db;
-  fs.mkdirSync(path.dirname(DB_PATH), { recursive: true });
-  const db = new Database(DB_PATH);
-  _db = db;
-  db.pragma('journal_mode = WAL');
-  runMigrations(db);
-  return db;
-}
-
-export function skillId(source, name) {
-  return `${source}::${name}`;
-}
-
-export function vecToBlob(vec) {
-  return Buffer.from(new Float32Array(vec).buffer);
-}
-
-export function blobToVec(blob) {
-  if (typeof blob === 'string') return JSON.parse(blob);
-  const buf = Buffer.isBuffer(blob) ? blob : Buffer.from(blob);
-  return Array.from(new Float32Array(buf.buffer, buf.byteOffset, buf.length / 4));
-}
+import Database from 'better-sqlite3';
+import path from 'path';
+import fs from 'fs';
+import { PROMPTGRAPH_DIR } from './config.js';
+
+const DB_PATH = path.join(PROMPTGRAPH_DIR, 'promptgraph.db');
+
+let _db = null;
+
+const MIGRATIONS = [
+  {
+    version: 1, description: 'initial schema',
+    up: db => db.exec(`
+      CREATE TABLE IF NOT EXISTS skills (
+        id TEXT PRIMARY KEY,
+        name TEXT NOT NULL,
+        description TEXT,
+        path TEXT NOT NULL,
+        source TEXT NOT NULL,
+        content TEXT NOT NULL,
+        hash TEXT
+      );
+      CREATE TABLE IF NOT EXISTS chunks (
+        id INTEGER PRIMARY KEY AUTOINCREMENT,
+        skill_id TEXT NOT NULL,
+        chunk_index INTEGER NOT NULL,
+        text TEXT NOT NULL,
+        embedding BLOB NOT NULL,
+        UNIQUE(skill_id, chunk_index)
+      );
+      CREATE TABLE IF NOT EXISTS edges (
+        from_skill TEXT NOT NULL,
+        to_skill TEXT NOT NULL,
+        PRIMARY KEY (from_skill, to_skill)
+      );
+      CREATE TABLE IF NOT EXISTS ratings (
+        skill_id TEXT PRIMARY KEY,
+        uses INTEGER DEFAULT 0,
+        success INTEGER DEFAULT 0,
+        fail INTEGER DEFAULT 0
+      );
+      CREATE VIRTUAL TABLE IF NOT EXISTS skills_fts USING fts5(
+        id UNINDEXED, name, description, content,
+        content='skills', content_rowid='rowid'
+      );
+    `),
+  },
+  {
+    version: 2, description: 'add hash column',
+    up: db => {
+      const cols = db.pragma('table_info(skills)').map(c => c.name);
+      if (!cols.includes('hash')) db.exec('ALTER TABLE skills ADD COLUMN hash TEXT');
+    },
+  },
+  {
+    version: 3, description: 'add registry metadata columns',
+    up: db => {
+      const cols = db.pragma('table_info(skills)').map(c => c.name);
+      for (const [col, def] of [
+        ['version', 'TEXT'],
+        ['author', 'TEXT'],
+        ['license', 'TEXT'],
+        ['updated_at', 'TEXT'],
+        ['downloads', 'INTEGER DEFAULT 0'],
+        ['verified', 'INTEGER DEFAULT 0'],
+        ['trust_level', "TEXT DEFAULT 'unknown'"],
+        ['rating', 'REAL DEFAULT 0'],
+        ['rating_count', 'INTEGER DEFAULT 0'],
+        ['popularity', 'REAL DEFAULT 0'],
+        ['last_update', 'TEXT'],
+      ]) {
+        if (!cols.includes(col)) db.exec(`ALTER TABLE skills ADD COLUMN ${col} ${def}`);
+      }
+    },
+  },
+  {
+    version: 4, description: 'registry_entries table',
+    up: db => db.exec(`
+      CREATE TABLE IF NOT EXISTS registry_entries (
+        id TEXT PRIMARY KEY,
+        trust_level TEXT DEFAULT 'unknown',
+        downloads INTEGER DEFAULT 0,
+        rating REAL DEFAULT 0,
+        rating_count INTEGER DEFAULT 0,
+        popularity REAL DEFAULT 0,
+        last_update TEXT
+      );
+    `),
+  },
+  {
+    version: 5, description: 'convert TEXT embeddings to Float32 BLOB',
+    up: db => {
+      const textEmbeddings = db.prepare("SELECT COUNT(*) as n FROM chunks WHERE typeof(embedding) = 'text'").get();
+      if (textEmbeddings?.n > 0) {
+        const rows = db.prepare("SELECT rowid, embedding FROM chunks WHERE typeof(embedding) = 'text'").all();
+        const upd = db.prepare('UPDATE chunks SET embedding = ? WHERE rowid = ?');
+        db.transaction(() => {
+          for (const row of rows) {
+            const vec = JSON.parse(row.embedding);
+            upd.run(Buffer.from(new Float32Array(vec).buffer), row.rowid);
+          }
+        })();
+        console.error(`[PromptGraph] Migrated ${textEmbeddings.n} embeddings TEXT→BLOB`);
+      }
+    },
+  },
+]
+
+function getSchemaVersion(db) {
+  try {
+    return db.prepare('SELECT MAX(version) as v FROM _schema_version').get().v || 0
+  } catch {
+    return 0
+  }
+}
+
+function runMigrations(db) {
+  const current = getSchemaVersion(db)
+  const pending = MIGRATIONS.filter(m => m.version > current).sort((a, b) => a.version - b.version)
+  if (!pending.length) return
+
+  db.exec(`CREATE TABLE IF NOT EXISTS _schema_version (version INTEGER PRIMARY KEY, description TEXT, applied_at TEXT)`)
+
+  for (const migration of pending) {
+    db.transaction(() => {
+      migration.up(db)
+      db.prepare('INSERT INTO _schema_version (version, description, applied_at) VALUES (?, ?, ?)').run(
+        migration.version, migration.description, new Date().toISOString()
+      )
+    })()
+    console.error(`[PromptGraph] DB migrated to v${migration.version}: ${migration.description}`)
+  }
+}
+
+export function getDb() {
+  if (_db) return _db;
+  fs.mkdirSync(path.dirname(DB_PATH), { recursive: true });
+  const db = new Database(DB_PATH);
+  _db = db;
+  db.pragma('journal_mode = WAL');
+  runMigrations(db);
+  return db;
+}
+
+export function skillId(source, name) {
+  return `${source}::${name}`;
+}
+
+export function vecToBlob(vec) {
+  return Buffer.from(new Float32Array(vec).buffer);
+}
+
+export function blobToVec(blob) {
+  if (typeof blob === 'string') return JSON.parse(blob);
+  const buf = Buffer.isBuffer(blob) ? blob : Buffer.from(blob);
+  return Array.from(new Float32Array(buf.buffer, buf.byteOffset, buf.length / 4));
+}

package/doctor.js

@@ -1,48 +1,48 @@
-import { getDb } from './db.js';
-
-export function runDoctor() {
-  const db = getDb();
-  const report = {};
-
-  // orphaned chunks (skill_id not in skills)
-  const orphanChunks = db.prepare(`
-    DELETE FROM chunks WHERE skill_id NOT IN (SELECT id FROM skills)
-  `).run();
-  report.orphanChunks = orphanChunks.changes;
-
-  // orphaned ratings
-  const orphanRatings = db.prepare(`
-    DELETE FROM ratings WHERE skill_id NOT IN (SELECT id FROM skills)
-  `).run();
-  report.orphanRatings = orphanRatings.changes;
-
-  // orphaned edges where from_skill no longer exists
-  const orphanFromEdges = db.prepare(`
-    DELETE FROM edges WHERE from_skill NOT IN (SELECT id FROM skills)
-  `).run();
-  report.orphanFromEdges = orphanFromEdges.changes;
-
-  // dangling edges where to_skill is a bare name that never resolved to a real skill
-  // (keep edges that point to real ids OR bare names that match a skill name)
-  const danglingEdges = db.prepare(`
-    DELETE FROM edges
-    WHERE to_skill NOT IN (SELECT id FROM skills)
-      AND to_skill NOT IN (SELECT name FROM skills)
-  `).run();
-  report.danglingEdges = danglingEdges.changes;
-
-  // duplicate skills by path (should not happen, but check)
-  const dupPaths = db.prepare(`
-    SELECT path, COUNT(*) as c FROM skills GROUP BY path HAVING c > 1
-  `).all();
-  report.duplicatePaths = dupPaths.length;
-
-  db.pragma('wal_checkpoint(TRUNCATE)');
-  db.exec('VACUUM');
-
-  report.totalSkills = db.prepare('SELECT COUNT(*) as c FROM skills').get().c;
-  report.totalChunks = db.prepare('SELECT COUNT(*) as c FROM chunks').get().c;
-  report.totalEdges = db.prepare('SELECT COUNT(*) as c FROM edges').get().c;
-
-  return report;
-}
+import { getDb } from './db.js';
+
+export function runDoctor() {
+  const db = getDb();
+  const report = {};
+
+  // orphaned chunks (skill_id not in skills)
+  const orphanChunks = db.prepare(`
+    DELETE FROM chunks WHERE skill_id NOT IN (SELECT id FROM skills)
+  `).run();
+  report.orphanChunks = orphanChunks.changes;
+
+  // orphaned ratings
+  const orphanRatings = db.prepare(`
+    DELETE FROM ratings WHERE skill_id NOT IN (SELECT id FROM skills)
+  `).run();
+  report.orphanRatings = orphanRatings.changes;
+
+  // orphaned edges where from_skill no longer exists
+  const orphanFromEdges = db.prepare(`
+    DELETE FROM edges WHERE from_skill NOT IN (SELECT id FROM skills)
+  `).run();
+  report.orphanFromEdges = orphanFromEdges.changes;
+
+  // dangling edges where to_skill is a bare name that never resolved to a real skill
+  // (keep edges that point to real ids OR bare names that match a skill name)
+  const danglingEdges = db.prepare(`
+    DELETE FROM edges
+    WHERE to_skill NOT IN (SELECT id FROM skills)
+      AND to_skill NOT IN (SELECT name FROM skills)
+  `).run();
+  report.danglingEdges = danglingEdges.changes;
+
+  // duplicate skills by path (should not happen, but check)
+  const dupPaths = db.prepare(`
+    SELECT path, COUNT(*) as c FROM skills GROUP BY path HAVING c > 1
+  `).all();
+  report.duplicatePaths = dupPaths.length;
+
+  db.pragma('wal_checkpoint(TRUNCATE)');
+  db.exec('VACUUM');
+
+  report.totalSkills = db.prepare('SELECT COUNT(*) as c FROM skills').get().c;
+  report.totalChunks = db.prepare('SELECT COUNT(*) as c FROM chunks').get().c;
+  report.totalEdges = db.prepare('SELECT COUNT(*) as c FROM edges').get().c;
+
+  return report;
+}

package/embedder.js

@@ -1,54 +1,54 @@
-import { EmbeddingModel, FlagEmbedding } from 'fastembed';
-import path from 'path';
-import os from 'os';
-
-const CACHE_DIR = path.join(os.homedir(), '.claude', '.promptgraph', 'model-cache');
-const BATCH_SIZE = 256;
-const MAX_EMBEDDING_CALLS = 10000;
-let embedCallCount = 0;
-
-export function getEmbedCallCount() { return embedCallCount; }
-export function resetEmbedCallCount() { embedCallCount = 0; }
-
-let model = null;
-
-async function getModel() {
-  if (!model) {
-    model = await FlagEmbedding.init({
-      model: EmbeddingModel.BGESmallENV15,
-      cacheDir: CACHE_DIR,
-    });
-  }
-  return model;
-}
-
-export async function embed(text) {
-  embedCallCount++;
-  const m = await getModel();
-  const results = [];
-  for await (const batch of m.embed([text])) {
-    results.push(...batch);
-  }
-  return Array.from(results[0]);
-}
-
-export async function embedBatch(texts) {
-  if (embedCallCount + texts.length > MAX_EMBEDDING_CALLS) {
-    throw new Error(`Embedding queue limit exceeded (max ${MAX_EMBEDDING_CALLS} chunks per session). Use --fast or reindex incrementally.`);
-  }
-  embedCallCount += texts.length;
-  const m = await getModel();
-  const all = [];
-  for await (const batch of m.embed(texts)) {
-    all.push(...batch);
-  }
-  return all.map(v => Array.from(v));
-}
-
-export function cosineSimilarity(a, b) {
-  let dot = 0;
-  for (let i = 0; i < a.length; i++) dot += a[i] * b[i];
-  return dot;
-}
-
-export { BATCH_SIZE };
+import { EmbeddingModel, FlagEmbedding } from 'fastembed';
+import path from 'path';
+import os from 'os';
+
+const CACHE_DIR = path.join(os.homedir(), '.claude', '.promptgraph', 'model-cache');
+const BATCH_SIZE = 256;
+const MAX_EMBEDDING_CALLS = 10000;
+let embedCallCount = 0;
+
+export function getEmbedCallCount() { return embedCallCount; }
+export function resetEmbedCallCount() { embedCallCount = 0; }
+
+let model = null;
+
+async function getModel() {
+  if (!model) {
+    model = await FlagEmbedding.init({
+      model: EmbeddingModel.BGESmallENV15,
+      cacheDir: CACHE_DIR,
+    });
+  }
+  return model;
+}
+
+export async function embed(text) {
+  embedCallCount++;
+  const m = await getModel();
+  const results = [];
+  for await (const batch of m.embed([text])) {
+    results.push(...batch);
+  }
+  return Array.from(results[0]);
+}
+
+export async function embedBatch(texts) {
+  if (embedCallCount + texts.length > MAX_EMBEDDING_CALLS) {
+    throw new Error(`Embedding queue limit exceeded (max ${MAX_EMBEDDING_CALLS} chunks per session). Use --fast or reindex incrementally.`);
+  }
+  embedCallCount += texts.length;
+  const m = await getModel();
+  const all = [];
+  for await (const batch of m.embed(texts)) {
+    all.push(...batch);
+  }
+  return all.map(v => Array.from(v));
+}
+
+export function cosineSimilarity(a, b) {
+  let dot = 0;
+  for (let i = 0; i < a.length; i++) dot += a[i] * b[i];
+  return dot;
+}
+
+export { BATCH_SIZE };