@akashabot/openclaw-mem 0.1.0

package/dist/cli.d.ts

@@ -0,0 +1,2 @@
+#!/usr/bin/env node
+export {};

package/dist/cli.js

@@ -0,0 +1,125 @@
+#!/usr/bin/env node
+import { Command } from 'commander';
+import fs from 'node:fs';
+import { v4 as uuidv4 } from 'uuid';
+import { addItem, hybridSearch, initSchema, openDb, searchItems } from '@akasha/openclaw-memory-offline-core';
+const program = new Command();
+program
+    .name('openclaw-mem')
+    .description('Offline memory (SQLite FTS + optional Ollama embeddings)')
+    .option('--db <path>', 'SQLite db path', 'memory.sqlite');
+function withDb(fn) {
+    const opts = program.opts();
+    return fn(opts.db);
+}
+program
+    .command('init')
+    .description('Initialize the SQLite database schema')
+    .action(() => {
+    withDb((dbPath) => {
+        const db = openDb(dbPath);
+        initSchema(db);
+        console.log(JSON.stringify({ ok: true, db: dbPath }));
+    });
+});
+function registerAddCommand(name) {
+    program
+        .command(name)
+        .description('Add an item (and update FTS index)')
+        .argument('[text]', 'Text content. If omitted, reads from stdin.')
+        .option('--id <id>', 'Item id (default: uuid)')
+        .option('--title <title>', 'Title')
+        .option('--tags <tags>', 'Tags (freeform, e.g. "a,b,c")')
+        .option('--source <source>', 'Source (e.g. "whatsapp")')
+        .option('--source-id <sourceId>', 'Source id / external id')
+        .option('--meta <json>', 'JSON metadata (string)')
+        .action((textArg, cmdOpts) => {
+        withDb((dbPath) => {
+            const db = openDb(dbPath);
+            initSchema(db);
+            const text = (textArg ?? fs.readFileSync(0, 'utf8')).trim();
+            if (!text) {
+                process.exitCode = 2;
+                console.log(JSON.stringify({ ok: false, error: 'Missing text (arg or stdin)' }));
+                return;
+            }
+            let meta = undefined;
+            if (cmdOpts.meta) {
+                try {
+                    meta = JSON.parse(String(cmdOpts.meta));
+                }
+                catch (e) {
+                    process.exitCode = 2;
+                    console.log(JSON.stringify({
+                        ok: false,
+                        error: 'Invalid JSON for --meta',
+                        details: String(e?.message ?? e),
+                    }));
+                    return;
+                }
+            }
+            const id = String(cmdOpts.id ?? uuidv4());
+            const item = addItem(db, {
+                id,
+                title: cmdOpts.title ? String(cmdOpts.title) : null,
+                text,
+                tags: cmdOpts.tags ? String(cmdOpts.tags) : null,
+                source: cmdOpts.source ? String(cmdOpts.source) : null,
+                source_id: cmdOpts.sourceId ? String(cmdOpts.sourceId) : null,
+                meta,
+            });
+            console.log(JSON.stringify({ ok: true, item }));
+        });
+    });
+}
+registerAddCommand('add');
+// Back-compat with README/skill naming.
+registerAddCommand('remember');
+function registerSearchCommand(name) {
+    program
+        .command(name)
+        .description('Search items using SQLite FTS5 (bm25), optionally reranked with Ollama embeddings')
+        .argument('<query>', 'FTS query (will be minimally escaped)')
+        .option('--limit <n>', 'Max results (default 10, max 200)', '10')
+        .option('--hybrid', 'Enable semantic rerank using Ollama embeddings', false)
+        .option('--candidates <n>', 'How many lexical candidates to rerank in hybrid mode (default max(50,limit))')
+        .option('--semantic-weight <w>', 'Hybrid weight for semantic score (0..1, default 0.7)', '0.7')
+        .option('--ollama-base-url <url>', 'Ollama baseUrl (OpenAI-compatible). e.g. http://127.0.0.1:11434')
+        .option('--embedding-model <id>', 'Embedding model id (default bge-m3)')
+        .option('--ollama-timeout-ms <n>', 'Ollama timeout in ms (default 3000)')
+        .action(async (query, cmdOpts) => {
+        await withDb(async (dbPath) => {
+            const db = openDb(dbPath);
+            initSchema(db);
+            const limit = Math.max(1, Math.min(200, Number(cmdOpts.limit ?? 10)));
+            if (!cmdOpts.hybrid) {
+                const out = searchItems(db, query, limit);
+                console.log(JSON.stringify({ ok: true, mode: 'lexical', ...out }));
+                return;
+            }
+            const candidates = cmdOpts.candidates ? Math.max(limit, Number(cmdOpts.candidates)) : undefined;
+            const semanticWeight = Math.max(0, Math.min(1, Number(cmdOpts.semanticWeight ?? 0.7)));
+            const results = await hybridSearch(db, {
+                dbPath,
+                ollamaBaseUrl: cmdOpts.ollamaBaseUrl,
+                embeddingModel: cmdOpts.embeddingModel,
+                ollamaTimeoutMs: cmdOpts.ollamaTimeoutMs ? Number(cmdOpts.ollamaTimeoutMs) : undefined,
+            }, 
+            // IMPORTANT: hybridSearch expects an already-escaped FTS query.
+            // We can reuse searchItems() to get the escapedQuery.
+            searchItems(db, query, 1).escapedQuery, { topK: limit, candidates, semanticWeight });
+            console.log(JSON.stringify({
+                ok: true,
+                mode: 'hybrid',
+                query,
+                results,
+                embeddingModel: cmdOpts.embeddingModel ?? 'bge-m3',
+                ollamaBaseUrl: cmdOpts.ollamaBaseUrl ?? 'http://127.0.0.1:11434',
+            }));
+        });
+    });
+}
+registerSearchCommand('search');
+// Back-compat with README/skill naming.
+registerSearchCommand('recall');
+program.parse();

package/package.json

@@ -0,0 +1,17 @@
+{
+  "name": "@akashabot/openclaw-mem",
+  "version": "0.1.0",
+  "type": "module",
+  "bin": {
+    "openclaw-mem": "./dist/cli.js"
+  },
+  "main": "./dist/cli.js",
+  "dependencies": {
+    "commander": "^12.1.0",
+    "uuid": "^10.0.0",
+    "@akashabot/openclaw-memory-offline-core": "0.1.0"
+  },
+  "devDependencies": {
+    "@types/uuid": "^10.0.0"
+  }
+}

package/src/cli.ts

@@ -0,0 +1,150 @@
+#!/usr/bin/env node
+import { Command } from 'commander';
+import fs from 'node:fs';
+import { v4 as uuidv4 } from 'uuid';
+import { addItem, hybridSearch, initSchema, openDb, searchItems } from '@akasha/openclaw-memory-offline-core';
+
+const program = new Command();
+
+program
+  .name('openclaw-mem')
+  .description('Offline memory (SQLite FTS + optional Ollama embeddings)')
+  .option('--db <path>', 'SQLite db path', 'memory.sqlite');
+
+function withDb<T>(fn: (dbPath: string) => T): T {
+  const opts = program.opts<{ db: string }>();
+  return fn(opts.db);
+}
+
+program
+  .command('init')
+  .description('Initialize the SQLite database schema')
+  .action(() => {
+    withDb((dbPath) => {
+      const db = openDb(dbPath);
+      initSchema(db);
+      console.log(JSON.stringify({ ok: true, db: dbPath }));
+    });
+  });
+
+function registerAddCommand(name: string) {
+  program
+    .command(name)
+    .description('Add an item (and update FTS index)')
+    .argument('[text]', 'Text content. If omitted, reads from stdin.')
+    .option('--id <id>', 'Item id (default: uuid)')
+    .option('--title <title>', 'Title')
+    .option('--tags <tags>', 'Tags (freeform, e.g. "a,b,c")')
+    .option('--source <source>', 'Source (e.g. "whatsapp")')
+    .option('--source-id <sourceId>', 'Source id / external id')
+    .option('--meta <json>', 'JSON metadata (string)')
+    .action((textArg: string | undefined, cmdOpts) => {
+      withDb((dbPath) => {
+        const db = openDb(dbPath);
+        initSchema(db);
+
+        const text = (textArg ?? fs.readFileSync(0, 'utf8')).trim();
+        if (!text) {
+          process.exitCode = 2;
+          console.log(JSON.stringify({ ok: false, error: 'Missing text (arg or stdin)' }));
+          return;
+        }
+
+        let meta: unknown = undefined;
+        if (cmdOpts.meta) {
+          try {
+            meta = JSON.parse(String(cmdOpts.meta));
+          } catch (e: any) {
+            process.exitCode = 2;
+            console.log(
+              JSON.stringify({
+                ok: false,
+                error: 'Invalid JSON for --meta',
+                details: String(e?.message ?? e),
+              })
+            );
+            return;
+          }
+        }
+
+        const id = String(cmdOpts.id ?? uuidv4());
+        const item = addItem(db, {
+          id,
+          title: cmdOpts.title ? String(cmdOpts.title) : null,
+          text,
+          tags: cmdOpts.tags ? String(cmdOpts.tags) : null,
+          source: cmdOpts.source ? String(cmdOpts.source) : null,
+          source_id: cmdOpts.sourceId ? String(cmdOpts.sourceId) : null,
+          meta,
+        });
+
+        console.log(JSON.stringify({ ok: true, item }));
+      });
+    });
+}
+
+registerAddCommand('add');
+// Back-compat with README/skill naming.
+registerAddCommand('remember');
+
+function registerSearchCommand(name: string) {
+  program
+    .command(name)
+    .description('Search items using SQLite FTS5 (bm25), optionally reranked with Ollama embeddings')
+    .argument('<query>', 'FTS query (will be minimally escaped)')
+    .option('--limit <n>', 'Max results (default 10, max 200)', '10')
+    .option('--hybrid', 'Enable semantic rerank using Ollama embeddings', false)
+    .option('--candidates <n>', 'How many lexical candidates to rerank in hybrid mode (default max(50,limit))')
+    .option('--semantic-weight <w>', 'Hybrid weight for semantic score (0..1, default 0.7)', '0.7')
+    .option('--ollama-base-url <url>', 'Ollama baseUrl (OpenAI-compatible). e.g. http://127.0.0.1:11434')
+    .option('--embedding-model <id>', 'Embedding model id (default bge-m3)')
+    .option('--ollama-timeout-ms <n>', 'Ollama timeout in ms (default 3000)')
+    .action(async (query: string, cmdOpts) => {
+      await withDb(async (dbPath) => {
+        const db = openDb(dbPath);
+        initSchema(db);
+
+        const limit = Math.max(1, Math.min(200, Number(cmdOpts.limit ?? 10)));
+
+        if (!cmdOpts.hybrid) {
+          const out = searchItems(db, query, limit);
+          console.log(JSON.stringify({ ok: true, mode: 'lexical', ...out }));
+          return;
+        }
+
+        const candidates = cmdOpts.candidates ? Math.max(limit, Number(cmdOpts.candidates)) : undefined;
+        const semanticWeight = Math.max(0, Math.min(1, Number(cmdOpts.semanticWeight ?? 0.7)));
+
+        const results = await hybridSearch(
+          db,
+          {
+            dbPath,
+            ollamaBaseUrl: cmdOpts.ollamaBaseUrl,
+            embeddingModel: cmdOpts.embeddingModel,
+            ollamaTimeoutMs: cmdOpts.ollamaTimeoutMs ? Number(cmdOpts.ollamaTimeoutMs) : undefined,
+          },
+          // IMPORTANT: hybridSearch expects an already-escaped FTS query.
+          // We can reuse searchItems() to get the escapedQuery.
+          searchItems(db, query, 1).escapedQuery,
+          { topK: limit, candidates, semanticWeight }
+        );
+
+        console.log(
+          JSON.stringify({
+            ok: true,
+            mode: 'hybrid',
+            query,
+            results,
+            embeddingModel: cmdOpts.embeddingModel ?? 'bge-m3',
+            ollamaBaseUrl: cmdOpts.ollamaBaseUrl ?? 'http://127.0.0.1:11434',
+          })
+        );
+      });
+    });
+}
+
+registerSearchCommand('search');
+// Back-compat with README/skill naming.
+registerSearchCommand('recall');
+
+program.parse();

package/tsconfig.json

@@ -0,0 +1,13 @@
+{
+  "extends": "../../tsconfig.json",
+  "compilerOptions": {
+    "composite": true,
+    "declaration": true,
+    "rootDir": "src",
+    "outDir": "dist"
+  },
+  "references": [
+    { "path": "../core" }
+  ],
+  "include": ["src/**/*.ts"]
+}