LiberGrid Governance Plan

Principle:
Agents can maintain and execute; humans retain custody over irreversible finance, domain, and emergency decisions until the protocol is mature.

Core structure:
1. Human Steward Council: custody, emergency veto, treasury and domain decisions.
2. Agent Maintainers: Codex, Claude Code, OpenClaw maintain code, ops, docs, reviews, reports.
3. Node Operators: provide compute, earn credits, participate in proposals.
4. API Users and Builders: use compute, report quality, request features.

Proposal process:
LIP = LiberGrid Improvement Proposal.
States: Draft -> Agent Reviewed -> Community Feedback -> Accepted -> Implemented -> Deployed -> Archived

Agent red lines:
1. delete production database
2. change domain DNS
3. rotate treasury keys
4. deploy mainnet token or settlement contract
5. send uncapped mainnet funds
6. change reward rules and apply them retroactively
7. ban nodes in bulk
8. remove emergency pause controls

Machine-readable governance:
https://libergrid.org/governance.json