Automates the verification of implementation tasks within a Spec-Driven Development loop to ensure all acceptance criteria are met.
This skill serves as the critical validation stage for the ShipSpec development framework within Claude Code. It orchestrates the verification of coding tasks by reading loop state files, fetching specific acceptance criteria from TASKS.json, and delegating to a verification agent. By providing standardized completion markers—VERIFIED, INCOMPLETE, or BLOCKED—it enables Claude to cleanly exit task loops, manage local state cleanup, and maintain an accurate audit log of the development process, ensuring that no feature is marked complete until it truly meets its specifications.
Key Features
01Direct integration with ShipSpec planning files
02Detailed debug logging and status reporting
038 GitHub stars
04Session control via standardized completion markers
05Task loop state management and cleanup
06Automated acceptance criteria validation
Use Cases
01Automating session exits to prevent redundant AI iterations
02Finalizing a coding feature in a Spec-Driven Development workflow
03Ensuring code changes meet pre-defined requirements before status updates