write-a-skill

Create and validate agent skills with SKILL.md frontmatter and bundled resources.

Updated Apr 6, 2026
One-click install
npx skills add https://github.com/toderian/project_template --skill write-a-skill-toderian
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: write-a-skill
Source: https://github.com/toderian/project_template/tree/main/plugins/agents-extras/skills/write-a-skill
Command: npx skills add https://github.com/toderian/project_template --skill write-a-skill-toderian

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Writing a reusable agent skill requires getting many details right at once: choosing the correct plugin, authoring valid SKILL.md frontmatter, writing a description that actually triggers, and passing validation. This Skill guides that entire authoring workflow so new skills are discoverable and correctly structured in both Claude Code and Codex. ## Core Features & Use Cases - Structured authoring process: Capture intent, interview for edge cases, draft the skill in place, and test it against realistic prompts before review. - Frontmatter and layout rules: Enforces name/description requirements, plugin selection by audience, and correct use of scripts/, references/, and assets/ directories. - Validation and quality bar: Checks description length, body size, build script output, and plugin validation so the skill passes marketplace standards. - Use Case: You want to add a new code-review skill to the agents-extras plugin. Use this Skill to pick the right plugin, write the SKILL.md with a trigger-carrying description, bundle a helper script, and validate the result. ## Quick Start Help me create a new skill for the agents-extras plugin that reviews pull request descriptions.

Frequently Asked Questions about write-a-skill

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I create a new agent skill for Claude Code and Codex?

Create a directory under plugins/<plugin>/skills/<name>/ containing a SKILL.md file with name and description frontmatter. Both harnesses discover skills the same way: every subdirectory of a plugin's skills/ folder with a SKILL.md is a skill.

What should a skill description include to trigger correctly?

The description must be under 1,024 characters, harness-neutral, and carry the specific trigger branches since it is the only text the agent sees when picking skills. State what the skill does and when to use it.

When should I bundle scripts with a skill?

Bundle a script when the operation is deterministic, when the same code would be regenerated on every invocation, or when errors need explicit handling. Scripts are limited to bash, python3 stdlib, and jq with no new runtime dependencies.

Which plugin should a new skill go in?

Choose by audience: agents-core for core coding workflow, agents-tasks for task ledgers and knowledge bases, agents-extras for situational review and authoring tools, and agents-personal for personal writing or niche migrations.

How do I validate a skill before publishing it?

Run claude plugin validate on the plugin and confirm no warnings appear. Also run python3 scripts/build.py --check for Codex policy currency and bash scripts/tests/run-all.sh for the full test suite.