## Problem `tests/` contains only `manual_*.py` scripts — no automated suite. A test-generation tool with no tests is a credibility + regression risk. ## Tasks - [ ] Convert manual scripts into real `pytest` tests. - [ ] Unit-test: intent extraction, confidence scoring, verifier output parsing, diagnosis classification, adapters (template render). - [ ] Mock LLM calls (no live API in CI). - [ ] Add coverage reporting; set a baseline gate. ## Acceptance `pytest` runs green locally and in CI with meaningful coverage of core/ and agents/.
Problem
tests/contains onlymanual_*.pyscripts — no automated suite. A test-generation tool with no tests is a credibility + regression risk.Tasks
pytesttests.Acceptance
pytestruns green locally and in CI with meaningful coverage of core/ and agents/.