agentsclimarketplace

Rc tdd

Skill juandarn/rawcode/skills/rc-tdd

Enforce a red → green → refactor TDD loop for a change. Trigger: implementing a behavior change where a test can define done. Composes with Gentle AI's sdd-apply.From its SKILL.md

Install
npx -y skills add juandarn/rawcode --skill rc-tdd

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

One thing to look at

  • 1 stars1 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

What its file declares

Copied from the file, not written here

The file declares its own license as MIT. That is the author’s claim about this one file, and it is not the same thing as the license GitHub reports for the repository, which is listed with the other numbers below.

SKILL.md

2.6 KB, as published. Nobody here has run it

Purpose

A rawcode extension for Gentle AI. It turns "add X" into a verifiable loop so that done means a test passed, not that code exists. It does not replace sdd-apply — it supplies the discipline sdd-apply runs under.

Default: strict

TDD is on by default for any code change. The tdd-gate Stop hook enforces it: if you change code and end the turn without running a test, it blocks once and asks for the test. You only skip it when the user explicitly says so — "urgent", "hotfix", "skip tdd" (or a .rawcode-no-tdd marker file). Pure refactors under existing coverage, config, and docs are not code changes and don't trigger the gate.

When to use

  • Implementing or fixing a behavior that a test can pin down.
  • Any sdd-apply task whose success criterion is checkable by code.

The loop

  1. Define done. Restate the task as a failing test. "Fix the bug" → "write a test that reproduces it." "Add validation" → "write a test for invalid input." State the success criterion in one line before touching implementation.
  2. Red. Write the smallest test that fails for the right reason. Run it. Confirm it fails, and that the failure message is the one you expect — not an import or syntax error masquerading as a red.
  3. Green. Write the minimum code to pass. No speculative abstraction, no handling for cases no test demands.
  4. Refactor. Clean up with the test green. Re-run after every change.
  5. Verify honestly. Run the test and quote the real result. If you did not run it, say so — never report green you did not see.

Rules (rawcode discipline)

  • One behavior per loop. Don't batch unrelated changes into one red-green.
  • The test is the contract: if the test is wrong, fix the test first, visibly.
  • Match the project's existing test framework and layout — detect it, don't impose one.
  • No stub or TODO left passing as done.

Compose with Gentle AI

  • Under sdd-apply: each task in tasks.md enters this loop; check the task off only when its test is green.
  • Persist nothing new — the task list and its status live wherever the active SDD artifact store already keeps them (engram / openspec).
  • Honors strict-TDD mode: if the project's sdd-init marked strict_tdd: true, this loop is mandatory, not optional.

Keep looking

Skills are one crate of 608,475. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.