Add integration tests along the app's critical path
Category: Testing · Updated 2026-09-22 · Tools: Both
Use this when a project has grown mostly unit-tested but the end-to-end flow a real user actually takes has never been exercised as a whole. Good for right after a usage-limit reset, when you have a fresh block of usage and want to spend it on something that pays off over time rather than a one-off fix. Medium to large usage, depending on repo size.
Full prompt
Map out this project's main critical-path flow (the sequence of steps a real user or caller goes through to get real value out of it — ask me if it's not obvious from the code). List every point along that path that currently has no integration-level test coverage, ranked by how risky a silent break there would be. Then add integration tests for the top 3-5 gaps only, on a new branch, running the full test suite after each one you add so a failure is caught immediately rather than at the end. Don't touch application code to make tests pass — if a gap can't be tested without a production-code change, stop and describe what change would be needed instead of making it. Don't push. When done, give me a short table: gap, risk level, test added (yes/no + why not for anything skipped).
Tips
- Confirm the critical path with the agent before it starts — it's easy for an agent to pick a technically-real but low-value flow instead of the one that actually matters
- Cap it at 3-5 gaps explicitly, otherwise the agent will try to cover everything and burn the whole session on lower-value cases
- Running the suite after each addition (not just at the end) is what catches a bad test before it compounds with the next one
Still waiting on a reset? See what to do when you hit the limit, or check Codex / Claude time since last reset.
Back to the prompt library · Subscribe: all-products feed