write-agents-md
Write AGENTS.md
Produce a small, durable instruction file for the agents that automatically receive it. This skill authors guidance; it does not implement the application or perform its QA.
Admission rule
A statement belongs in AGENTS.md only when it passes all three tests:
- Audience: every agent entering this directory must know it, regardless of its task. Root means the whole repository; nested means that entire subtree.
- Value: it changes a decision the agent could otherwise get wrong. It is not standard tooling knowledge, an obvious layout, or something immediately discoverable in manifests/source.
- Stability: it is an enduring convention or invariant, not a session observation, changing version, task status, temporary workaround, or machine fact.
A statement failing audience goes in a task-specific skill. Detailed explanations/evidence go in documentation; execution state goes in the project tracker. A stable documentation entry point can stay in AGENTS.md; a task-by-task document/skill catalog cannot. Do not invent skill names or create additional skills without authorization.
TDD, the seven-step Manual QA procedure, fresh endpoint captures, verification, commit policy, and honest completion are standing obligations for every contributor. They distinguish runtime from documentation work without becoming scenario runbooks. Keep those quality gates in AGENTS.md; put commands, fixture recipes, particular flows and recovery procedures in skills.