Every agent does one job well and hands its work to the next. Meet the crew, what each one is great at, and what it leaves behind.
Turns a one-line idea into a scope every other agent can work from.
Reads the scope like a sceptic and asks before anyone builds the wrong thing.
Maps every screen, sketches each one and checks it covers everything in the scope.
Creates the design system — colours, type, spacing and components — as named tokens.
Chooses the stack, sets up the project and splits the build into tasks.
Builds the pages, components, styling and state you'll actually click.
Builds the API routes, business logic and auth behind the screens.
Designs the schema, writes the migrations and the queries that read them.
Reviews the code against the scope and design, and checks screenshots of your running app.
Writes tests for what was built and runs them, so you don't find the bugs.
Traces a bug from its symptom to its cause and proposes the smallest fix.
Writes the README and docs from what was actually built, not what was planned.
Spots where the scope, design and code have drifted apart — before it matters.
Models shown are an example setup — change any of them in Settings → Models.
Install Commandr, connect a model, and hand off your first idea.
For Apple Silicon Macs · v0.5.0
