Module 05 · intermediate · 1 hour

Skills

Skills package specialized instructions, references, assets, and optional scripts. Codex sees a compact catalog first, then reads the full SKILL.md only when a skill is selected.

Progressive disclosure

The initial catalog contains each skill’s name, description, and path. Full instructions load on activation; references and assets load only when needed.

my-skill/
├── SKILL.md
├── scripts/
├── references/
├── assets/
└── agents/openai.yaml

Write a strong trigger

The description should name the job, domain, positive triggers, and important exclusions. Front-load signal so truncation does not erase the key use case.

---
name: security-review
description: Review code changes for injection, auth, and data exposure. Use for PR security audits; not for style-only review.
---

Scope and invoke

Repository skills live under .agents/skills from the current directory up to the repo root. User skills live under $HOME/.agents/skills. Use $ to mention a skill or let Codex match it implicitly.

$skill: Explicit invocation
Description: Implicit matching
openai.yaml: UI and policy metadata

Distribute intentionally

Use direct folders for local or repository authoring. Use plugins for installable distribution, related skill bundles, connectors, and shared presentation assets.

Continue learning

Use the official OpenAI Codex documentation as the final reference for current commands, settings, and availability.

Return to the full curriculum.