Context

Contexts serve as a knowledge base — broad context about the whole project, or narrower context for a specific group of tasks. They're especially useful early in a project, helping agents get familiar with your project and build up a knowledge base before they start making inferences with more autonomy.

In the sidebar, navigate to AI Agents > Contexts.

From this screen you can review, edit, and create contexts. Describe — in natural language, in your project's configured language — one or more contexts to help agents understand what, how, and for what purpose they should work. It's especially useful to write contexts that cover your business domain, your preferred technology stack, and, for complex systems, your architectural approach.

You can create as many contexts as you like. It can help to create contexts scoped to specific projects or temporary working groups that may be reused by tasks later on. Cover the important details, but try to keep each context concise and only attach the contexts a task actually needs — too much context can slow an agent down and consume more token credits.