Hey Hunters, Timur here. I built cookiedclaw because at 3am I want to text my agent about an idea I just had and kick off the implementation. Without opening a laptop. And running on my own server, not Anthropic's. The tagline is half a joke. OpenClaw's revolution was bringing autonomous agents into the messaging apps you already live in. cookiedclaw inherits that idea and plugs a real coding agent (Claude Code) behind it. Channels hides tool calls and dies on terminal close. Remote Control is mobile-only, can't start new sessions, and bridges through Anthropic. OpenClaw shells out to the Claude CLI, which breaks the moment Anthropic flips `--bare` to default. cookiedclaw runs as a Claude Code plugin with a bundled MCP server inside your real session, on your VPS. Your Max plan keeps working. You are the session, cookiedclaw just adds Telegram on top. Live tool calls in your DM, permission relay from your phone, image dispatch, daemon mode that survives terminal close. Roadmap: configurable heartbeat, crons, multi-agent orchestration. Plus more transports (Discord / Slack / iMessage) and more runtimes (Codex / Cursor). Transport-agnostic gateway. Any messenger, any coding agent. Would love your feedback. Especially if you're running OpenClaw on Telegram with Max: what would make this an obvious migration?

cookiedclaw

What if OpenClaw and Claude Code had a baby

What if OpenClaw and Claude Code had a baby