Executes test-driven development plans using Red-Green-Refactor cycles and Tidy First principles.
npx skills add https://github.com/kohdice/dotfiles --skill tdd-kohdice
Please help me install this Agent Skill. Skill: tdd Source: https://github.com/kohdice/dotfiles/tree/main/config/agents/skills/tdd Command: npx skills add https://github.com/kohdice/dotfiles --skill tdd-kohdice
.plans/ directory, finds the next unchecked item, and works through Test:, Test (integration):, and Refactor: items in order. - Red-Green-Refactor Cycle: Writes a failing test, implements the minimum code to pass it, then refactors only when all tests pass, with fast-tier and slow-tier test runs at the right checkpoints. - Living Plan Management: Appends newly discovered test cases to the plan, reports phase completions, and keeps structural and behavioral commits separate. - Use Case: After creating a TDD plan for a new feature, invoke the skill with the plan filename and it implements each test increment, marking checkboxes and suggesting commits per phase. ## Quick Start Ask the AI to run the tdd skill on your plan file, for example by saying: execute the TDD plan in .plans/watch-refresh.md.High-intent search queries and answers about installing and using this skill.
Invoke the skill with a plan filename, such as /tdd watch-refresh.md, and it resolves the file relative to the .plans directory. Without a filename it uses the most recent plan from the current session or asks which existing plan to use.
Red writes one failing test for a small behavior increment, Green implements the minimum code to make it pass, and Refactor cleans up structure only while all tests pass. Each phase runs the fast test tier to confirm the expected state.
Integration tests in the slow tier run only when a Test (integration) plan item is current, when a phase or plan completes, or before a commit. Routine checkpoints run only the fast unit-test tier to keep feedback quick.
An immediately passing test is treated as a signal, not a success. The skill inspects whether the behavior already exists, in which case the item is marked complete, or whether the test fails to exercise the intended path, in which case the test is fixed.
Yes, standalone refactoring runs Phase 3 on its own with the invariant that all tests pass before and after. Structural changes are validated by comparing identical passing test name sets across both runs, and no tests are added or modified.
The skill does not invent work when a plan is complete and never commits uninvited. Discovered tests must describe observable behavior of the project's own code, and tests for third-party library responsibilities are explicitly excluded.
stagewise-ioCreates structured implementation plans as markdown files before coding begins.
fallow-rsCommits, pushes, and opens a ready pull request for local changes on GitHub.
ssrxsVerifies Laravel projects' environment, code quality, security, and deployment readiness.
chrisliu298Enforce a test-first workflow requiring a failing test before production code.