Specship picks up work from human review comment, requested change, or review bot comment.
Use case
AI agent that responds to PR review comments with code.
Specship turns work from Git pull requests into acceptance criteria, failing tests, implementation commits, and a review-ready pull request. It is designed for teams that want async implementation without losing specs, tests, review, or budget control.
Workflow
How the ticket becomes a pull request.
The agent drafts observable acceptance criteria before code starts.
Failing tests encode the behavior so reviewers can see the intended change.
The output is a normal branch and pull request with commands, checks, and review notes.
Proof points
What this workflow ships with.
Specship is built for normal engineering artifacts: comments, status transitions, branches, tests, pull requests, and review evidence your team can audit later.
Direct answer: AI agent responds to PR comments is useful when a team wants implementation work to begin in its planning tool and end in a reviewable pull request, not in a one-off chat transcript.
Use it when
- The ticket has a clear user or system outcome.
- The desired behavior can be tested before implementation.
- The PR should stay inside normal Git review and CI.
- Budget, protected paths, or merge policy matter.
Do not use it when
- The product decision is still unresolved.
- The task requires production credentials in a public form.
- The change affects sensitive systems without human review.
- The ticket cannot define what done means.
Security notes
Keep the agent inside policy.
The agent can be autonomous inside the rules you set, but repo access, credentials, test execution, and merge behavior stay bounded by workspace policy.