pursue

You can automate long-horizon execution by providing an active intent with a linked spec, allowing the agent to autonomously iterate through the test matrix, write minimal failing tests, implement code, and validate results until all rows pass.

An autonomous pursuit loop reads an active development intent, locates its linked spec, and iterates through the test matrix by writing minimal failing tests, implementing code fixes, and validating results to reduce manual orchestration and context-switching.

You resume an autonomous workflow by using breadcrumb-based progress tracking recorded in the intent body, which supports pausing and cleanly restarting the pursuit loop without losing completed test matrix iterations.

Preconditions for starting autonomous test matrix iteration require an active intent, a linked spec, and a non-empty test matrix. The system enforces these safety rules alongside iteration caps to prevent runaway loops.

Safety governance handles runaway execution by enforcing preconditions like spec presence and non-empty test matrices, while capping iterations to prevent infinite loops and stopping the execution cleanly on success or user pause.