Cursor rule

1---
2description: Test maintenance — classify failures before changing code or tests
3alwaysApply: true
4---
5 
6# Test Philosophy
7 
8A failing test is evidence, not the problem itself.
9 
10Before modifying either product code or tests, classify the failure.
11 
12## Case 1 — Product regression
13 
14The test still describes the intended behavior.
15 
16Action:
17- Fix the production code.
18- Do not weaken the test.
19 
20Examples:
21- Wrong calculation
22- Broken API response
23- Missing UI element
24- Queue logic regression
25- Gameplay bug
26 
27## Case 2 — Test regression
28 
29The product intentionally evolved and the test no longer reflects reality.
30 
31Examples:
32- Login/session architecture changed
33- Fixtures became obsolete
34- Route renamed intentionally
35- New initialization flow
36- UI structure intentionally redesigned
37 
38Action:
39- Update the fixture or test.
40- Preserve the original requirement.
41- Keep assertions as strict as before.
42 
43## Forbidden
44 
45Never make a test green by:
46 
47- removing assertions
48- replacing assertions with weaker ones
49- using `assert True`
50- adding `pytest.skip`
51- changing expected behaviour without an approved design decision
52- ignoring failures
53 
54## Required
55 
56When updating a failing test, always explain:
57 
581. Why the old test became invalid.
592. Which requirement is still being verified.
603. Why the updated test proves the same behaviour.
61 
62## Genesis Colonies Rules
63 
64- Session tests use `session_transaction()` where appropriate.
65- Commit database changes before HTTP requests when required.
66- Prefer realistic fixtures over mocking gameplay.
67- Keep tests close to real player behaviour.
68- Never introduce duplicate tests for the same canonical feature.
69- A green test suite must mean the product actually works, not merely that the tests were made easier.
70