When the agent reaches for it
The agent keeps every installed skill's description in view and loads the full instructions when a request matches. This is the description it reads:
Test-driven development. Use when the user wants to build features or fix bugs test-first, mentions "red-green-refactor", or wants integration tests.
What it asks of the agent
Agents write tests easily and write bad ones just as easily. This skill is the reference that keeps the red-green loop honest. Tests go through public interfaces and read like a specification. Before writing any, the agent names the seams it plans to test and gets your agreement, so effort lands on the paths that matter.
- Implementation-coupled tests, which break on a refactor when behavior has not changed, are out.
- Tautological tests, which recompute the expected value the way the code does, are out.
- Writing every test first and every implementation after is out. One seam, one test, one change per cycle.
What the instructions cover
The sections of its SKILL.md, in order.
- What a good test is
- Seams: where tests go
- Anti-patterns
- Rules of the loop
What ships with it
Besides SKILL.md, the folder carries these. The agent opens them only when a step in the instructions calls for one.
-
agents/ -
mocking.md -
tests.md
Install the TDD skill
In Claude Code, run this in a session. It comes from the publisher's own instructions.
/plugin install mattpocock-skills License: MIT.
In Longwave, add Engineering Skills from the marketplace. It installs all 37 of its skills, and your Claude Code, Codex and OpenCode threads can use them.