[Anchor Proposal]: red-green TDD

[petehodgson-tribe]

Proposed Term

Red/Green TDD

Context (Optional)

Classically, test-driven development is practiced by first writing a failing test, then watching it fail, then making it pass. LLMs don't do this by default - they tend to implement the feature, then write the test after the fact.

When you say "use Red/Green TDD" the LLM sticks to the classic practice.

source" https://simonwillison.net/guides/agentic-engineering-patterns/red-green-tdd/

Pre-submission Checklist

• I have searched existing anchors and this term is not already included
• This term refers to a well-established concept (not something I invented)

[raifdmueller]

Thanks @petehodgson-tribe, this is an excellent proposal!

Red/Green TDD is exactly the kind of anchor that measurably changes LLM behavior. By default, coding agents implement the feature first and write tests afterwards — "Red/Green TDD" reliably forces them into the classical cycle (failing test → minimal implementation → refactor).

The Simon Willison reference is a solid source. Category: testing-quality, related to existing tdd-london-school and tdd-classicist.

Will proceed with the PR.

[raifdmueller]

Merged via #425. The anchor is live in the catalog — thanks for the excellent proposal!

[raifdmueller]

Merged via #425. The anchor is live in the catalog — thanks for the excellent proposal! (Ready to close.)