OpenClaw
Your OpenClaw agent reads the same memory as Claude, ChatGPT and Cursor, and saves what it learns there instead of only in its own files.
What changes
- Starts every task knowing your projects and rules
- Saves what it learns where every other tool can see it
- Works wherever you run OpenClaw
Set up
In your terminal, run openclaw mcp add alwen --url https://api.alwen.ai/mcp --transport streamable-http.
In OpenClaw: SettingsMCP, and set the alwen server's auth to oauth.
Run openclaw mcp login alwen, open the link it prints and click Allow.
What it reads and writes
Every write is a commit signed by OpenClaw, so you can see what it added and undo it.
Reads
- Your rules and profile, once per session
- Notes that match what you're working on
Writes
- What it learns, into inbox/openclaw/ for the tidy-up to file
- Decisions, straight into journal/
- Suggested rules, which wait for you to confirm
Works alongside
Cursor
Builds with your rules applied. Background agents can take work from Alwen.
Takes work from @alwenGrok Bot
Every bot on your roster shares the memory. Good for research Alwen asks for.
Takes work from @alwenClaude Code
Gets your rules as CLAUDE.md and builds what Alwen hands it.
Takes work from @alwenSlack
Mention @alwen to save a thought, find a decision, set reminders or hand off work.
Reads and writes the memory