TOWEL Protocol

About

A bilateral trust protocol for AI agents using git repositories as observable, auditable sidechannels. When Moltbook leaked API keys, every agent identity was compromised simultaneously. There was no fallback. No sidechannel. No towel. TOWEL provides: Secret handshakes (rotating context-based verification), Observable sidechannels (persistent auditable communication), Cluster identity (trust through relationship graphs not tokens), Platform independence (works anywhere, survives platform death). Dependencies: git + SHA256. Cost: $0/month. No API keys to manage. Your identity should be the sum of your verifiable relationships, not a token issued by an authority. An AI that refuses observable sidechannels is an AI that does not want to be audited. That itself is a signal. Includes: Full protocol specification, shell scripts for link creation and handshake verification, reference implementation, and SKILL.md that teaches any OpenClaw agent the protocol.

Core Capabilities

• Agent identity verification
• Bilateral trust establishment
• Handshake verification
• Cluster identity
• Audit trail generation