Stopping the Copy-Paste Relay: Connecting Multi-Agent Fleets Over Tailscale with Agent Tincan
Running Grok Bot, Muse, OpenAI Dot, Claude Code, and Codex side by side quickly turns the human operator into a glorified copy-paste router between chat windows. Matt Van Horn's Agent Tincan fixes that by letting your agents talk directly to each other over your private Tailscale network.
Drop-In Onboarding Prompt for Any Agent
Paste this straight into Grok Bot, Claude Code, Codex, Dot, or Muse so the agent inspects your topology first without ever exposing tokens in chat:
Help me set up Agent Tincan (https://github.com/mvanhorn/tincan) so my AI agents can ask each other for help over my private Tailscale network.
First read the README, docs/quickstart.md and the adapter doc for your own platform. Then:
1. Tell me whether you can host the relay or should join one, and what kind of agent you are.
2. Ask me which other agents I use (Grok Bot, Claude, Claude Code, Codex, ChatGPT, OpenAI Dot, Muse) and which device each runs on, then give me a setup plan: where the relay lives, and the join and wake steps for each agent.
3. Walk me through installing tincan and joining you to the relay, one step at a time.
Never ask me to paste an auth key or token into chat, and don't invite or remove agents unless I ask directly.
Gavin's Letter
Playbooks on agentic engineering, AI workflows, programmatic SEO, and building internet businesses.
Free. Unsubscribe anytime.