Mission Control is a self-hosted control plane for operating AI agents that dispatches tasks, inspects runs, reviews failures, tracks spend, and coordinates agent runtimes from a local dashboard backed by SQLite.
The tool addresses the operational challenge of managing multiple agents or runtimes by providing a single place to see and govern work across them. Rather than replacing agent reasoning or tool loops, it sits above runtimes and offers operators visibility into task execution, agent registration and sessions, activity streams, cost tracking, and governance controls. The control plane stores state locally in SQLite and supports workspace awareness, with optional gateway connectivity for task and messaging work.
Mission Control suits teams running multiple agents where it becomes difficult to track task ownership, execution history, review status, and accumulated spend and failures. The tool works with OpenClaw, Claude Code, Codex, and other runtimes as a coordination layer rather than a framework for defining planning or tool use. It is not the right choice for single-agent deployments that remain understandable through native CLIs, teams needing managed multi-tenant SaaS rather than self-hosted infrastructure, or deployments that cannot tolerate schema and API changes during alpha development.
The project maintains active development with regular updates to its control-plane capabilities and deployment infrastructure. The tool includes multiple interfaces for operator access, spanning a web UI, CLI, MCP server, REST API with OpenAPI documentation, WebSocket, and server-sent events. Documentation covers deployment scenarios including persistent data configuration, TLS termination, and gateway connectivity patterns. The project carries an explicit alpha status warning about potential breaking changes to APIs and schemas between releases.
