You have run a task end to end. What you read next depends on what you are trying to do.
I want to understand how this actually works
Start with How a task becomes a pull request, then Agent roles. Together they explain every stage and who performs it.
After that, Specifications is the highest-value page in the documentation — the quality of what Coroid builds is bounded by the quality of what it is asked to build.
I want to run more work at once
Read Capacity and agent slots. Slots are the unit of parallelism, and the most common source of confusion is expecting them to make a single task faster. They do not.
Then Tasks, plans, sprints and lanes for how to structure work larger than one pull request.
I want better output
Three levers, in order of impact:
- Write better specifications. See Specifications.
- Give Coroid more context. See Project context.
- Change the models. See AI profiles. A profile carries a model for every agent, so moving a piece of work up a profile lifts planning, building and review together.
I want tighter control over what ships
Quality gates explains how policy decides what can pass, and where a gate can stop a task.
I want to connect my other tools
The Integrations overview maps what connects to what and in which direction — issue trackers in, chat both ways, review evidence onto pull requests.
I want to know what something means
Glossary defines every term the interface uses, and Task statuses explains every status a task can be in and what to do about it.