Quest Plan

Test Coverage Climb · 4 levels · populated

Level 1 · Find the gaps

Rank the gaps before you fill any of them. Identify undertested code and rank it by actual risk: highest usage, lowest coverage, worst bug history, or the code a recent incident touched. A rarely-run script with zero coverage matters less than a hot path with none. Produce a short ranked list of candidates rather than a raw dump of every uncovered line — the ranking is the point. When the list exists and the top of it is genuinely the riskiest gap, run quest_level complete.

Level 2 · Write the tests

Write for the highest-ranked gaps first, and write for real behavior. Cover the top of the level-1 list before the bottom. Test the actual behavior — edge cases, error paths, and failure modes — not just a happy-path call that re-confirms what already worked. A test that only exercises the easy case adds coverage numbers without adding protection. When the highest-risk gaps have real tests behind them, run quest_level complete.

Level 3 · Prove it catches bugs

Don’t trust a test suite you haven’t watched fail. Deliberately reintroduce one bug the new tests are meant to catch — revert a past fix, introduce an off-by-one, skip a validation check — and confirm the suite fails loudly and points at the right place. A test that would have stayed green through the bug it was written for isn’t protection, it’s decoration. Revert your deliberate break afterward and confirm green again. When you’ve watched the suite catch a real regression, run quest_level complete.

Level 4 · Report the delta

Show the before and after, and hand off what’s still open. Measure and report the coverage change for the targeted area specifically — not a vague project-wide percentage that hides where the work actually went. Include the ranked list of what’s still undertested so the next pass doesn’t have to start from zero. When the delta is measured and the next-priority list is written, run quest_level complete.