A lightweight, fully local memory layer for AI assistants — Claude Desktop, Claude Code, Cursor, Codex, and ChatGPT. Stores memories in a SQLite database with FTS5 full-text search and BM25 ranking. Version 0.4.10, active development, released biweekly. No cloud, no accounts, no API keys. Key differentiators: zero external dependencies beyond better-sqlite3 and MCP SDK; six MCP tools for querying memories via natural language; HTTP REST API for tools that don't support MCP; CLI for management, backup, and restore.
npm install knol-localNo compatibility data collected yet for this library.
Verified import paths — ran on the pinned version, not inferred.
Install knol-local globally, add a memory with tags and importance, search, list, start HTTP server, and launch MCP server for Claude/Cursor integration.
Upgrade Node.js to version 18 or later
Use `npx knol-local` or install globally with `npm install -g knol-local`
Ensure you pass `--key <token>` if you want auth, or omit for open access
Use `knol-local export --out file.json` for JSON export, `knol-local backup` for DB copy
Use exact phrases in quotes to bypass stemming if needed
Run `npm rebuild better-sqlite3` in the knol-local installation directory, or reinstall with `npm install -g knol-local`
Upgrade to latest knol-local: `npm install -g knol-local@latest`, then check if knex is listed in dependencies
Install globally: `npm install -g knol-local`. If still not found, check npm global bin directory is in PATH
Use a different port: `knol-local serve --port 3002`, or kill the process using port 3001