mission-control

robsannaa/mission-control is ranked 15,922 among these 8 peers; Linzaer/Face-Track-Detect-Extract is ranked 15,928.

See everything, dispatch anything, all on your own machine. One writing code, one on research, one watching an inbox, a few more on whatever I pointed them at that morning. Past two or three, terminals stop working as an interface. Mission Control is the control plane over the fleet. Watch agents work live, hand them tasks and let them pick them up, spin up subagents and fan them out, see what everything costs, keep the whole thing healthy. It runs

locally and drives OpenClaw as the engine underneath. ⚡ Hand it to your agent to install You don't install things by hand anymore. Point Claude Code, Codex, or whatever you drive at this and it does the rest: A lot of this was written by the agents it runs. Please, consider supporting me and buy me a Claude Code Subscription! You tab between terminals, lose track of what is running, and guess at what it cost. You point an agent at

a problem, drop it on the board, and go start the next one while it runs. The harness should never be the interesting part. Mission Control is built to sit behind them and stay out of the way: nothing to babysit, nothing to sync, hard to break. You should be able to forget it is there and keep shipping. It reads and writes straight through to the engine underneath, live. Change something here and it lands immediately, with no sync

step and no cache to go stale. The only state it keeps is two small local files: usage history and the task board. ** What you see is what is running, right now. ** No migrations, no backups, no cleanup jobs. ** If the dashboard goes down, the agents keep running. ** Nothing to provision, nothing to upgrade between versions. Pull the dashboard out and the fleet still runs. It is glass over the engine, not the engine. No clicking around to find out

if things are working. Talk to any of them **Chat** is a real conversation with any agent, in the browser. Attach files, pick the model, stream the response, switch agents without losing context. Point agents at work and watch it move **Tasks** is a board (Backlog, In Progress, Review, Done) where a card is not a note. Drop one in, an agent picks it up, and you watch it cross to Done. This is where "point a few agents at

the problem and go" becomes something you can see. Keep agents working while you don't **Cron Jobs** run agents on a schedule: summarize the inbox every morning, check for updates every hour. Create, edit, pause, and test, with full run history so you know what happened overnight. Command the team **Agents** shows the whole hierarchy as a live org chart: every agent and subagent, who is active, which channels they are on, which workspace. Spin up new subagents and fan

them out on the spot. Know the cost to the token **Usage** tracks every token across every model and agent.