Replace sqlite-lembed local embedder

README.md

@@ -198,7 +198,7 @@ Local mode stores solutions in SQLite and does not call the hosted API. The dire
 
 Keyword, semantic, and hybrid search are available locally by default. `clanker local embed` downloads/checks the default GGUF embedding model and embeds pending local solutions. Disable local semantic and hybrid search with `CLANKER_LOCAL_SEMANTIC=0`, `false`, or `off`. Override the database path with `CLANKER_LOCAL_DB` and the model path with `CLANKER_LOCAL_MODEL_PATH`.
 
-The Docker-isolated e2e check is available on demand:
+The Docker-isolated e2e check runs the local-mode suite against Node 22 and Node 24 by default:
 
 ```bash
 pnpm test:e2e:local

packages/cli/.claude-plugin/plugin.json

@@ -1,6 +1,6 @@
 {
   "name": "clankeroverflow",
-  "version": "1.1.0",
+  "version": "1.2.0",
   "description": "Search-first debugging memory for AI coding agents. Search prior fixes before fresh debugging, validate results, vote on tried solutions, and log verified reusable fixes.",
   "author": {
     "name": "ClankerOverflow",

packages/cli/.codex-plugin/plugin.json

@@ -1,6 +1,6 @@
 {
   "name": "clankeroverflow",
-  "version": "1.1.0",
+  "version": "1.2.0",
   "description": "Search-first debugging memory for AI coding agents. Search prior fixes before fresh debugging, validate results, vote on tried solutions, and log verified reusable fixes.",
   "author": {
     "name": "ClankerOverflow",

packages/cli/e2e/Dockerfile.local-mode

@@ -1,4 +1,5 @@
-FROM node:22-bookworm-slim
+ARG NODE_IMAGE=node:22-bookworm-slim
+FROM ${NODE_IMAGE}
 
 ENV COREPACK_ENABLE_DOWNLOAD_PROMPT=0
 ENV PNPM_HOME=/pnpm

packages/cli/e2e/local-mode.mjs

@@ -112,14 +112,35 @@ async function logDirectSolution(env, fixture) {
 async function verifyDirectCli(env) {
   logStep("checking native semantic dependencies import");
   await import("sqlite-vec");
-  await import("sqlite-lembed");
+  await import("node-llama-cpp");
 
-  logStep("downloading or checking the local embedding model");
+  logStep("logging a direct CLI fixture before embeddings are available");
+  const semanticDisabledEnv = { ...env, CLANKER_LOCAL_SEMANTIC: "0" };
+  await logDirectSolution(semanticDisabledEnv, fixtures.vite);
+
+  logStep("checking local semantic status before embedding pending direct logs");
+  const pendingStatus = JSON.parse(await runCli(["local", "status", "--json"], env));
+  assert.equal(pendingStatus.mode, "local");
+  assert.equal(pendingStatus.semantic.enabled, true);
+  assert.equal(pendingStatus.semantic.totalSolutions, 1);
+  assert.equal(pendingStatus.semantic.embeddedSolutions, 0);
+  assert.equal(pendingStatus.semantic.pendingEmbeddings, 1);
+  assert.equal(pendingStatus.semantic.sqliteVecAvailable, true);
+  assert.equal(pendingStatus.semantic.embedderAvailable, true);
+
+  logStep("verifying direct keyword search works before local embeddings exist");
+  const preEmbedKeyword = await runCli(
+    ["search", "EADDRINUSE", "--mode", "keyword", "--limit", "1"],
+    env,
+  );
+  assertTopProblem(preEmbedKeyword, fixtures.vite.problem, "pre-embed direct keyword search");
+
+  logStep("downloading or checking the local embedding model and embedding pending solutions");
   const embedOutput = await runCli(["local", "embed"], env);
   assert.match(embedOutput, /Local embeddings ready/);
+  assert.match(embedOutput, /1 solution\(s\) embedded/);
 
-  logStep("logging direct CLI fixture solutions");
-  await logDirectSolution(env, fixtures.vite);
+  logStep("logging direct CLI fixture solutions with immediate embeddings");
   await logDirectSolution(env, fixtures.playwright);
   await logDirectSolution(env, fixtures.prisma);
 

packages/cli/openclaw.plugin.json

@@ -2,7 +2,7 @@
   "id": "@bernoussama/clankeroverflow",
   "name": "ClankerOverflow",
   "description": "Search-first debugging memory for AI coding agents. Search prior fixes before fresh debugging, validate results, vote on tried solutions, and log verified reusable fixes.",
-  "version": "1.1.0",
+  "version": "1.2.0",
   "configSchema": {
     "type": "object",
     "additionalProperties": false

packages/cli/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@clankeroverflow/cli",
-  "version": "1.1.0",
+  "version": "1.2.0",
   "description": "ClankerOverflow CLI for logging and searching AI agent solutions",
   "license": "MIT",
   "repository": {
@@ -51,6 +51,6 @@
     "vitest": "4.0.7"
   },
   "optionalDependencies": {
-    "sqlite-lembed": "0.0.1-alpha.8"
+    "node-llama-cpp": "3.18.1"
   }
 }

packages/cli/src/index.ts

@@ -39,7 +39,7 @@ function formatLocalStatus(dbPath: string, status: Awaited<ReturnType<LocalBacke
         ? pc.green("available")
         : pc.red(status.sqliteVecError ?? "unavailable")
     }`,
-    `sqlite-lembed: ${
+    `node-llama-cpp: ${
       status.embedderAvailable
         ? pc.green("available")
         : pc.red(status.embedderError ?? "unavailable")
@@ -291,7 +291,7 @@ export function createProgram(options: CreateProgramOptions = {}) {
             detail: status.sqliteVecError ?? "available",
           },
           {
-            name: "sqlite-lembed",
+            name: "node-llama-cpp",
             ok: status.embedderAvailable,
             detail: status.embedderError ?? "available",
           },

packages/cli/src/mcp/local-backend.test.ts

@@ -6,10 +6,15 @@ import { afterEach, beforeEach, describe, expect, test, vi, type MockInstance }
 
 import { LocalBackend } from "./local-backend";
 import { openLocalDb } from "./local-db";
-import { embeddingFingerprintForConfig, type LocalSemanticConfig } from "./local-semantic";
+import {
+  embeddingFingerprintForConfig,
+  floatVectorToBuffer,
+  LOCAL_EMBEDDER_ID,
+  type LocalSemanticConfig,
+} from "./local-semantic";
 
 function vector(values: number[]) {
-  return Buffer.from(new Float32Array(values).buffer);
+  return floatVectorToBuffer(values, values.length);
 }
 
 function writeGguf(modelPath: string, contents: string) {
@@ -103,7 +108,7 @@ describe("CLI local MCP backend", () => {
       dimensions: 4,
     };
     const embedder = {
-      embed(text: string) {
+      async embed(text: string) {
         return /oauth/i.test(text) ? vector([1, 0, 0, 0]) : vector([0, 1, 0, 0]);
       },
     };
@@ -134,7 +139,7 @@ describe("CLI local MCP backend", () => {
       dimensions: 4,
     };
     const embedder = {
-      embed(text: string) {
+      async embed(text: string) {
         return /oauth/i.test(text) ? vector([1, 0, 0, 0]) : vector([0, 1, 0, 0]);
       },
     };
@@ -167,7 +172,7 @@ describe("CLI local MCP backend", () => {
     };
     const firstBackend = new LocalBackend(dbPath, {
       semantic,
-      embedder: { embed: () => vector([1, 0, 0, 0]) },
+      embedder: { embed: async () => vector([1, 0, 0, 0]) },
     });
     await firstBackend.log({
       problem: "OAuth callback timeout",
@@ -177,7 +182,7 @@ describe("CLI local MCP backend", () => {
 
     const freshBackend = new LocalBackend(dbPath, {
       semantic,
-      embedder: { embed: () => vector([1, 0, 0, 0]) },
+      embedder: { embed: async () => vector([1, 0, 0, 0]) },
     });
 
     await expect(freshBackend.status()).resolves.toMatchObject({
@@ -201,6 +206,24 @@ describe("CLI local MCP backend", () => {
     expect(second).not.toBe(first);
   });
 
+  test("local embedding metadata uses node-llama-cpp", () => {
+    expect(LOCAL_EMBEDDER_ID).toBe("node-llama-cpp");
+  });
+
+  test("converts embedding vectors to explicit float32 sqlite blobs", () => {
+    const buffer = floatVectorToBuffer([1.5, -2.25], 2);
+
+    expect(buffer).toHaveLength(8);
+    expect(buffer.readFloatLE(0)).toBe(1.5);
+    expect(buffer.readFloatLE(4)).toBe(-2.25);
+  });
+
+  test("rejects local embedding vectors with unexpected dimensions", () => {
+    expect(() => floatVectorToBuffer([1, 2, 3], 4)).toThrow(
+      "node-llama-cpp returned 3 embedding dimensions",
+    );
+  });
+
   test("replacing a model file at the same path makes existing embeddings pending", async () => {
     const semantic: LocalSemanticConfig = {
       enabled: true,
@@ -210,7 +233,7 @@ describe("CLI local MCP backend", () => {
     };
     const backend = new LocalBackend(dbPath, {
       semantic,
-      embedder: { embed: () => vector([1, 0, 0, 0]) },
+      embedder: { embed: async () => vector([1, 0, 0, 0]) },
     });
     await backend.log({
       problem: "OAuth callback timeout",

packages/cli/src/mcp/local-backend.ts

@@ -9,7 +9,7 @@ import type {
 } from "./backend";
 import { openLocalDb, type LocalDb } from "./local-db";
 import {
-  createSqliteEmbedder,
+  createLocalEmbedder,
   embeddingFingerprintForConfig,
   ensureLocalSemanticSchema,
   ensureVecTable,
@@ -31,7 +31,7 @@ export class LocalSemanticSearchNotConfiguredError extends Error {
 }
 
 type SearchRow = SolutionResult & { rank: number };
-type Embedder = { embed(text: string): Buffer };
+type Embedder = { embed(text: string): Promise<Buffer> };
 
 function nowIso() {
   return new Date().toISOString();
@@ -184,7 +184,7 @@ export class LocalBackend implements SolutionBackend {
     if (!this.semantic?.enabled) throw new LocalSemanticSearchNotConfiguredError();
     await ensureVecTable(this.db, this.semantic.dimensions);
     const embedder = await this.resolveEmbedder();
-    const embedding = embedder.embed(queryEmbeddingText(queryText));
+    const embedding = await embedder.embed(queryEmbeddingText(queryText));
     const rows = this.db
       .prepare(
         `SELECT solution_id, distance
@@ -245,7 +245,7 @@ export class LocalBackend implements SolutionBackend {
     await ensureVecTable(this.db, this.semantic.dimensions);
     const embedder = await this.resolveEmbedder();
     const text = solutionEmbeddingText({ problem, solution, tags });
-    const embedding = embedder.embed(text);
+    const embedding = await embedder.embed(text);
     insertEmbedding(this.db, {
       solutionId: id,
       model: this.semantic.modelId,
@@ -260,7 +260,7 @@ export class LocalBackend implements SolutionBackend {
   private async resolveEmbedder() {
     if (this.embedder) return this.embedder;
     if (!this.semantic?.enabled) throw new LocalSemanticSearchNotConfiguredError();
-    this.embedder = await createSqliteEmbedder(this.db, this.semantic);
+    this.embedder = await createLocalEmbedder(this.semantic);
     return this.embedder;
   }
 

packages/cli/src/mcp/local-semantic.ts

@@ -22,7 +22,7 @@ export const DEFAULT_LOCAL_MODEL_FILE = "bge-small-en-v1.5-q8_0.gguf";
 export const DEFAULT_LOCAL_MODEL_DIMENSIONS = 384;
 export const DEFAULT_LOCAL_MODEL_URL =
   "https://huggingface.co/ggml-org/bge-small-en-v1.5-Q8_0-GGUF/resolve/main/bge-small-en-v1.5-q8_0.gguf";
-export const LOCAL_EMBEDDER_ID = "sqlite-lembed";
+export const LOCAL_EMBEDDER_ID = "node-llama-cpp";
 export const LOCAL_EMBEDDING_FORMAT_VERSION = "solution-v1";
 export const LOCAL_QUERY_FORMAT_VERSION = "query-v1";
 
@@ -54,7 +54,6 @@ export type LocalSemanticStatus = {
 };
 
 const sqliteVecLoaded = new WeakSet<LocalDb>();
-const lembedLoaded = new WeakSet<LocalDb>();
 
 export function defaultLocalModelPath(env: NodeJS.ProcessEnv = process.env) {
   const cacheRoot = env.XDG_CACHE_HOME || join(homedir(), ".cache");
@@ -177,40 +176,42 @@ export async function loadSqliteVec(db: LocalDb) {
   }
 }
 
-async function loadLembed(db: LocalDb) {
-  if (lembedLoaded.has(db)) return;
+export async function checkLocalEmbedderAvailable() {
   try {
-    const sqliteLembed = await import("sqlite-lembed");
-    sqliteLembed.load(db);
-    lembedLoaded.add(db);
+    await import("node-llama-cpp");
   } catch (error) {
     throw new Error(
-      `sqlite-lembed extension is unavailable: ${error instanceof Error ? error.message : String(error)}`,
+      `node-llama-cpp embedder is unavailable: ${error instanceof Error ? error.message : String(error)}`,
     );
   }
 }
 
-function ensureTempModelTable(db: LocalDb, config: LocalSemanticConfig) {
+export function floatVectorToBuffer(vector: ArrayLike<number>, dimensions: number) {
+  if (vector.length !== dimensions) {
+    throw new Error(
+      `node-llama-cpp returned ${vector.length} embedding dimensions, but CLANKER_LOCAL_MODEL_DIMENSIONS is ${dimensions}`,
+    );
+  }
+  const buffer = Buffer.allocUnsafe(dimensions * Float32Array.BYTES_PER_ELEMENT);
+  for (let index = 0; index < dimensions; index += 1) {
+    buffer.writeFloatLE(vector[index]!, index * Float32Array.BYTES_PER_ELEMENT);
+  }
+  return buffer;
+}
+
+export async function createLocalEmbedder(config: LocalSemanticConfig) {
   const modelValidation = validateGgufFile(config.modelPath);
   if (!modelValidation.ok) {
     throw new Error(modelValidation.error ?? "model file is not valid");
   }
-  db.prepare(
-    `INSERT OR REPLACE INTO temp.lembed_models(name, model)
-     SELECT ?, lembed_model_from_file(?)`,
-  ).run(config.modelId, config.modelPath);
-}
-
-export async function createSqliteEmbedder(db: LocalDb, config: LocalSemanticConfig) {
-  await loadLembed(db);
-  ensureTempModelTable(db, config);
+  const { getLlama } = await import("node-llama-cpp");
+  const llama = await getLlama();
+  const model = await llama.loadModel({ modelPath: config.modelPath });
+  const context = await model.createEmbeddingContext();
   return {
-    embed(text: string) {
-      const row = db.prepare("SELECT lembed(?, ?) AS embedding").get(config.modelId, text) as
-        | { embedding: Buffer | Uint8Array }
-        | undefined;
-      if (!row?.embedding) throw new Error("sqlite-lembed returned no embedding");
-      return Buffer.from(row.embedding);
+    async embed(text: string) {
+      const embedding = await context.getEmbeddingFor(text);
+      return floatVectorToBuffer(embedding.vector, config.dimensions);
     },
   };
 }
@@ -375,7 +376,7 @@ export async function getLocalSemanticStatus(db: LocalDb, config: LocalSemanticC
   let embedderAvailable = true;
   let embedderError: string | undefined;
   try {
-    await loadLembed(db);
+    await checkLocalEmbedderAvailable();
   } catch (error) {
     embedderAvailable = false;
     embedderError = error instanceof Error ? error.message : String(error);

packages/cli/src/mcp/server.ts

@@ -191,8 +191,8 @@ export function createMcpServer() {
                 ? "sqlite-vec: available"
                 : `sqlite-vec: ${status.sqliteVecError}`,
               status.embedderAvailable
-                ? "sqlite-lembed: available"
-                : `sqlite-lembed: ${status.embedderError}`,
+                ? "node-llama-cpp: available"
+                : `node-llama-cpp: ${status.embedderError}`,
             ].join("\n"),
           },
         ],

packages/cli/src/package.test.ts

@@ -24,7 +24,8 @@ describe("packages/cli package metadata", () => {
     expect(packageJson.dependencies?.["better-sqlite3"]).toBe("12.10.0");
     expect(packageJson.dependencies?.mcplog).toBe("^0.0.5");
     expect(packageJson.dependencies?.["sqlite-vec"]).toBe("^0.1.9");
-    expect(packageJson.optionalDependencies?.["sqlite-lembed"]).toBe("0.0.1-alpha.8");
+    expect(packageJson.optionalDependencies?.["node-llama-cpp"]).toBe("3.18.1");
+    expect(packageJson.optionalDependencies?.["sqlite-lembed"]).toBeUndefined();
     expect(packageJson.dependencies?.zod).toBe("^4.1.13");
     expect(packageJson.dependencies?.["@tobilu/qmd"]).toBeUndefined();
     expect(packageJson.dependencies?.["node-llama-cpp"]).toBeUndefined();