← ClaudeAtlas

test-writerlisted

Analyze Python code, design and implement unit or integration tests, run them, and produce an evidence-based Markdown report. Use when any coding agent is asked to write, improve, execute, or report on Python tests, including projects that use unittest or pytest and workflows backed by a hosted API, an OpenAI-compatible endpoint, Ollama, or no external model.
Angelamer/test-writer-skill · ★ 0 · AI & Automation · score 60
Install: claude install-skill Angelamer/test-writer-skill
# Test Writer Create tests that expose behavior rather than merely increase coverage. ## Workflow 1. Read the target and its local imports. Identify public behavior, side effects, boundaries, exceptions, and external dependencies. 2. Inspect project instructions and existing tests. Preserve the existing framework and conventions. If none exist, prefer standard-library `unittest`. 3. Separate pure logic from I/O. Mock only process boundaries such as files, networks, clocks, models, and plotting; do not mock the behavior under test. 4. Add focused tests for normal behavior, meaningful boundaries, failure paths, and regressions. Do not edit production code unless the user also requested a fix. 5. Run the narrowest test command first, then the relevant suite. Prefer existing project commands; use the bundled Make commands when working in this repository. Record the exact command and its exit status. 6. Generate a Markdown report with `make report` or `scripts/test_report.py`. Report coverage, covered lines, missing lines, and missing branches for the target source file, not aggregate project or test-code coverage. Use `make coverage-html` when a line-by-line HTML view is useful. Report failures honestly and explain the likely production defect. 7. Exclude test files from aggregate coverage badges and HTML summaries. Keep target-source coverage in each Markdown report distinct from repository-wide non-test coverage. 8. Review the diff for unrelated changes, secrets, brittle ass