version: "1.0.0" name: python-mutation-testing description: Assess test suite strength by introducing deliberate bugs (mutations) one at a time and checking whether any test catches each one. Reports on test suite gaps and optionally implements missing tests. user-invocable: true license: MIT compatibility: Designed for Claude Code or similar AI coding agents, and for projects using Python. metadata: author: honnibal version: "1.0.0" openclaw: emoji: "\U0001F9EA" homepage: https://github.com/honnibal/claude-skills

Mutation Testing

Assess the strength of the project's test suite by introducing deliberate bugs (mutations) into the source code, one at a time, and checking whether any test fails. A mutation that the tests don't catch reveals a gap in coverage.

Scope

• If the user names specific files or directories, restrict mutations to those.
• Only mutate production code, never test files.
• Prioritise code with meaningful logic (branching, arithmetic, state changes).

Pre-flight

1. Clean working tree. Run git status. If there are uncommitted changes, ask the user to commit or stash first.
2. Find the test runner. Look for pytest.ini, pyproject.toml [tool.pytest], or a tests/ directory. Confirm the suite passes before starting.
3. Map the code. Read the files in scope. Build a mental model of data flow.

Workflow

1. Choose mutations. Read the file and identify 3–8 candidate mutations from the catalogue below.
2. Apply, test, revert. For each mutation:

a. Apply the mutation using a single-line change. b. Run the test suite. c. Record the result: Killed (a test failed) or Survived (no test failed). d. Revert the mutation with git checkout -- <file>.

Mutation Catalogue

1. Delete or skip a side effect — remove an assignment, method call, append.
2. Negate or invert a condition — if x → if not x, and → or.
3. Change a boundary — < → <=, >= → >, == 0 → == 1.
4. Swap or hardcode a return value — return score → return 0.
5. Delete an early return or guard clause — remove if not items: return [].
6. Change an operator — + → -, * → /.
7. Modify a default argument or constant — change a default parameter value.
8. Swap argument order — reverse arguments in a non-commutative call.

Assessing test failures

Rate diagnostic quality:

• Clear — test name and failure message immediately point to the bug.
• Indirect — failure describes a symptom rather than root cause.
• Cascading — many tests failed, making it hard to locate the root cause.

Reporting

Produce a summary table:

| # | File | Mutation | Result | Diagnostic | Notes |
|---|------|----------|--------|------------|-------|

Then provide:

1. Mutation score: killed / total (e.g. 6/8 = 75%)
2. Uncaught mutations: list each survived mutation with explanation
3. Diagnostic quality: note indirect/cascading failures
4. Recommended tests: for each survived mutation, describe a test that would catch it

Critical Rules

• Always revert after every test run. Never stack mutations.
• Don't mutate test files, imports, type annotations, or docstrings.
• 3–8 mutations per file is enough. Quality over quantity.

Cross-References

• python-testing — pytest patterns, coverage
• python-hypothesis — property-based testing