Resolving Edge Cases

Use this to actively inspect relevant code, find plausible edge-case bugs, and fix or harden them. Do not wait for the user to enumerate edge cases.

Default Scope

When the user does not provide files, a commit, or a diff, use changed paths as the starting point: run git diff --name-only main...HEAD; if the repo uses master, run git diff --name-only master...HEAD instead. If there is no diff, inspect the files implied by the request; if no scope can be inferred, ask for one target.

Loop

1. Infer intended behavior from the user request, code, tests, docs, and sibling flows; state assumptions, but only ask when the rule is ambiguous.
2. Trace the real flow end to end, including callers, sibling routes, cleanup paths, and stored state.
3. Proactively inspect plausible edge cases for that flow; skip generic checklists that cannot occur.
4. Fix or harden each plausible bug at the shared root path, not only the reported symptom.
5. Add the smallest regression test or executable check that fails without the fix.
6. Run the narrow check, then scan sibling callers/routes for the same pattern.
7. Repeat while each fix exposes a concrete adjacent edge case.
8. Run the repo's normal verification gate before final response.