agent-skills

My personal collection of skills for LLM assistants.

Each top-level directory is one skill. Inside each, a SKILL.md file holds the YAML frontmatter and the skill body. Supporting files (examples, references, small scripts) live in the same directory when a skill needs them.

Skills

• blog-post-writer: Drafts blog posts about software topics in the author's voice.
• linkedin-companion-writer: Writes short LinkedIn posts that point at an existing blog post.
• ai-first-docs: Migrates a project's documentation to AI-first docs approach, where the same content serves both humans and AI agents.

Adding a new skill

1. Create a directory with a kebab-case name (for example code-reviewer).

2. Add a SKILL.md inside it with this frontmatter:

   ---
   name: code-reviewer
   description: "When the assistant should use this skill, and when not to."
   ---

3. Keep name identical to the directory name.

4. Write the skill body below the frontmatter. The description is what the assistant uses to decide when to trigger the skill, so be specific about both trigger cases and non-trigger cases.

Format note

The SKILL.md convention is currently Claude's, but the content inside (the instructions, the rules, the voice guidance) is portable. If another assistant adopts a compatible format in the future, the skills here should port with minor wrapper changes at most.