audit-mirage-analyze

Phases 2-3 and 7: Systematic Audit, Green Mirage Patterns, Fix Verification

Invariant Principles

1. Every test function gets audited - No skipping tests that "look fine"; line-by-line analysis catches what scanning misses
2. Assertions determine test value - A test without meaningful assertions is worse than no test: it creates false confidence and hides production defects
3. Score by pattern, not by intuition - Apply every Green Mirage Pattern as the scoring rubric
4. A fix is a new suspect - A remediated assertion gets the same adversarial treatment as the assertion it replaced

Phase 2: Systematic Line-by-Line Audit

For EACH test file, work through EVERY test function. For tests with multiple actions, apply one "Action Analysis" block per action.

### Test: `test_function_name` (file.py:line)

**Purpose (from name/docstring):** What this test claims to verify

**Setup Analysis:**
- Line X: [what's being set up]
- Line Y: [dependencies/mocks introduced]
- Concern: [any setup that hides real behavior?]

**Action Analysis:**
- Line Z: [the actual operation being tested]
- Code path: function() -> calls X -> calls Y -> returns
- Side effects: [files created, state modified, etc.]

**Assertion Analysis:**
- Line A: `assert condition` - Would catch: [what failures] / Would miss: [what failures]

**Verdict:** SOLID | GREEN MIRAGE | PARTIAL
**Gap (if any):** [Specific scenario that passes test but breaks production]
**Fix (if any):** [Concrete code to add]

Code Path Tracing

Trace the COMPLETE path for each test action:

test_function()
  |-> production_function(args)
        |-> helper_function()
        |     |-> external_call() [mocked? real?]
        |     |-> returns value
        |-> processes result
        |-> returns final
  |-> assertion checks final