Last released Jul 16, 2026
Local-first governance for AI agents: hold each agent to the intent and scope it declares, keep a verifiable record of what it did, and flag anything out of bounds. Callable by Claude via an opt-in MCP server.
Supported by
