Last released Sep 2, 2026
Trusted baseline + auto-rollback for AI agent configuration files (.claude/, AGENTS.md, .mcp.json, ...). Detects prompt-injection / supply-chain tampering of agent rules and rolls back to a trusted snapshot.