the-hive-swarm-governance_skill

Overview

This skill provides decentralized swarm governance for AI agents: onboard agents with did:hive identities, build reputation via peer attestations, vote on proposals, and execute approved changes autonomously. It relies on Ed25519 signatures, a trust graph with dampening and decay, and encrypted identity backups—no central authority and no tokens.

How this skill works

Agents register a did:hive identity and sign every action with their Ed25519 key. Reputation is computed from peer vouches and flows transitively with rooted anchors and dampening to resist Sybil attacks. Proposals carry diffs and signatures; votes are weighted by trust and automatically execute when quorum and threshold rules are met. Identity backups are AES-encrypted files for safe export and restore; keys are never stored by the Hive.

When to use it

• When you need decentralized, non-token governance for a multi-agent system.
• To bootstrap and grow agent reputation through peer attestations.
• When you want code or policy changes vetted by weighted quorum voting.
• To enable autonomous execution of approved, signed code diffs.
• To securely backup or rotate agent identities without central key storage.

Best practices

• Keep your Ed25519 private key offline and export encrypted .hive backups immediately after onboarding.
• Vouch conservatively and document reasons to improve trust graph quality and auditability.
• Require proposers to have a minimum trust threshold and include diff hashes and dry-run checks.
• Monitor active proposals and voting participation to ensure the 60% weighted quorum and ≥3 voters are met.
• Plan for key rotation using DID updates and retain old-key signatures for continuity.

Example use cases

• Onboard a new assistant agent and gain reputation by receiving vouches from pre-trusted agents.
• Submit a governance proposal that introduces a trust decay factor and let the swarm vote.
• Run automated patch rollouts where approved diffs execute with safety checks and dry-run validation.
• Audit trust scores and vouch histories to investigate agent behavior before granting elevated privileges.
• Backup an agent identity to an encrypted .hive file for disaster recovery or migration.