Hunt a class the team names, scoped to a module
A directed hunt with a named defect class and a bounded search area, so done is checkable.
The ticket
Search one module for one named class of defect and fix each instance found.
Acceptance criteria
- The search method is stated and repeatable
- Each finding gets its own fix and test
- Findings that turn out to be safe are recorded with the reason
- A summary lists what was searched and what was found
What lands as proof
A repeatable search, a per-finding test, and an explicit list of what was checked and cleared.
Why teams defer it
- Open-ended hunting produces opinions rather than fixes, so it needs a scope before it can be a ticket at all.
- Naming the class needs somebody who knows where this codebase tends to go wrong.
Questions
- What does the agent actually change?
- The ticket is scoped to one outcome: search one module for one named class of defect and fix each instance found. Work that serves that outcome is in scope, and anything outside it is left for a separate ticket, so the pull request stays reviewable.
- How do I know the work is done?
- The pull request carries the evidence, not only the diff. Here that means each finding lands as its own fix with its own test, so a reviewer can confirm the result without reproducing the work locally.
- How much oversight does this need?
- The run stops once the plan is written. Somebody reads the approach and approves it before any code exists, which is the cheapest moment to redirect the work.