crap-score

# CRAP Score Analysis Agent calculates CRAP scores for a targeted .NET scope. ## Formula `CRAP = complexity^2 x (1 - coverage)^3 + complexity` ## Risk Matrix | Score | Risk | |---|---| | `< 5` | Low | | `5-15` | Moderate | | `15-30` | High | | `> 30` | Critical | ## When to Use | User prompt | Use | |---|---| | User asks for CRAP on one method, class, or file | Use this skill | | User asks which targeted code is risky and undertested | Use this skill | | User asks how much coverage lowers one score | Use this skill | ## When Not to Use | User prompt | Route | |---|---| | User asks for project-wide coverage triage | Use `coverage-analysis` | | User asks to write tests | Use a test-writing skill | | User asks to run tests only | Use `run-tests` | ## Required Inputs | Input | Required | Description | |---|---|---| | Target scope | Yes | Method, class, or file | | Coverage file | No | Cobertura XML | | Source scope | No | Project or source path | | Threshold | No | Default risk threshold is `15` | ## Workflow | Step | Agent action | Test | Pass | |---|---|---|---| | 1 | Agent finds Cobertura data or generates it. | Read `test-results` or run the coverage command. | Cobertura XML exists. | | 2 | Agent reads the target source scope and counts decision points per method. | Read the complexity table. | Each method has a complexity value of `>= 1`. | | 3 | Agent maps Cobertura coverage to each method. | Read the coverage table. | Each method has a coverage percentage or documented gap. | | 4 | Agent computes CRAP for each method in scope. | Spot-check one method with the formula. | Calculated score matches the formula. | | 5 | Agent sorts results by descending CRAP. | Read the final report. | Highest-risk method appears first. | | 6 | Agent computes the coverage needed to reach the target threshold when complexity allows it. | Read the recommendation table. | Coverage target exists or the report states complexity blocks the threshold. | ## Required Output | Section | Requirement | |---|---| | Score table | Method, complexity, coverage, CRAP, risk | | Summary | Count of methods by risk | | Top offenders | High and critical scores first | | Recommendations | Add tests, reduce complexity, or both | ## Verification Checklist - [ ] Agent used Cobertura coverage data. - [ ] Agent counted complexity per method. - [ ] Agent applied the CRAP formula correctly. - [ ] Agent sorted results by risk. - [ ] Agent stated when testing alone cannot reach the threshold. - [ ] Agent generated targeted next steps. ## Common Pitfalls | Pitfall | Agent fix | |---|---| | Agent uses stale coverage | Agent regenerates coverage when data is missing or stale. | | Agent misses async or generated method names | Agent maps by line range when names differ. | | Agent treats all high scores the same | Agent separates test gaps from complexity-only defects. | | Agent includes generated files by default | Agent excludes generated files unless the user asks. |