Problem
No agent has guidance on how many tests are sufficient. The planner writes "Tests to Write: 1, 2, 3..." but there is no criterion for adequacy. Without a heuristic, test quantity is inconsistent across tasks.
Proposed Fix
In prd-task-planner.md, add to the TDD Mode template (after line 175):
### Coverage Heuristic
Write at least one test per acceptance criterion. For each, cover:
- Happy path
- One error/edge case (empty input, null, boundary value)
Simple, concrete, no tooling required. Mutation testing would be overkill at this stage.
Rationale
From the article: "concrete grading criteria" — a gradable heuristic that the implementer can follow and the reviewer can verify.
Effort
~5 minutes — prompt addition in 1 file.
Problem
No agent has guidance on how many tests are sufficient. The planner writes "Tests to Write: 1, 2, 3..." but there is no criterion for adequacy. Without a heuristic, test quantity is inconsistent across tasks.
Proposed Fix
In
prd-task-planner.md, add to the TDD Mode template (after line 175):Simple, concrete, no tooling required. Mutation testing would be overkill at this stage.
Rationale
From the article: "concrete grading criteria" — a gradable heuristic that the implementer can follow and the reviewer can verify.
Effort
~5 minutes — prompt addition in 1 file.