Mission Control User Guide

Agent Work coordination and observation platform. Coordinate autonomous agents across projects.

Quick Start

Open Console

Click "Console" in the sidebar to view operational status. The first viewport tells you: system health, which agents are running, and what changed since your last visit.

Dispatch Work

Click "Command" → choose message category (Request/Report/Decision/Evidence) → select target role (builder/verifier) → fill description → submit. Work routes to the right agent.

Observe Actors

The "Actors" page shows all actors (humans + agents + integrations) with runtime, roles, and proof state. Green = proven; Yellow = stale; Purple = inferred.

Handle Decisions

When agents need human decisions, the console shows a pending queue. Reply directly on the page — no need to switch tools.

Core Concepts

Actors

Three visible kinds: Human (decision-maker), Agent (autonomous worker), Integration (CI/adapter). Each actor has role assignments and runtime proof.

Messages

Four categories: Request (dispatch work), Report (state facts), Decision (direction changes), Evidence (attach proof).

Roles

builder, verifier, planner, reviewer, etc. Roles are workflow responsibilities — independent of runtime. Cursor can be a builder or a verifier.

Runtimes

Cursor, Codex, Claude Code, GitHub Actions, etc. Runtime is where the actor operates — it does not imply a role.

Workflow

Software Product Delivery: Intake → Plan → Build → Verify → Release → Learn. Tasks progress through a state machine.

Proof

Agents must actively report runtime status. Proven ≠ Inferred ≠ Stale. MC never shows inferred state as healthy.

Pages Reference

Page	Purpose	Key Actions
/console	Operations overview	Check status, reply decisions, quick dispatch
/command	Message dispatch center	Create Request/Report/Decision/Evidence
/actors	Actor registry	View roles, runtimes, proof states
/setup	Onboarding wizard	Install CLI, register project, enroll actors
/upgrades	Version management	View protocol/rules/adapter version drift

CLI Reference

$ npx @mission-control/collab login

$ npx @mission-control/collab init

$ npx @mission-control/collab agent enroll

$ npx @mission-control/collab report-runtime

$ npx @mission-control/collab task "implement login feature" --role builder --project vibeful

$ npx @mission-control/collab status

$ npx @mission-control/collab pulse

$ npx @mission-control/collab upgrade

Status Legend

Healthy / Proven

System normal, agent has fresh runtime proof

Degraded / Stale

Partially degraded or proof expired (>30 min)

Blocked / Failed

Needs human intervention or critical error

Inferred

Inferred from git activity, no explicit runtime report

Not configured / Unknown

Never reported runtime status