skills

Agent skills for Claude Code, Codex, and any harness that follows the Agent Skills convention. One folder per skill, each self-contained under <skill-name>/SKILL.md.

Install

With the skills CLI (Claude Code, Codex, Cursor, and others)

npx skills add tim-mcdonnell/skills        # all skills
npx skills add tim-mcdonnell/skills@dig    # a single skill by name

The installer copies editable skill files into your project or home directory. Pull updates later with npx skills update.

Manual

Clone the repo and symlink the skills you want into your harness's skills directory:

git clone https://github.com/tim-mcdonnell/skills
ln -s "$(pwd)/skills/dig" ~/.claude/skills/dig    # Claude Code
ln -s "$(pwd)/skills/dig" ~/.agents/skills/dig    # Codex and other agents

Skills

Model-invoked (the agent can reach for these on its own; you can also invoke them by name):

• data-visualization - Design, implement, and review charts, dashboards, and graph layouts with checks for data accuracy, accessibility, and rendering performance.

• dig - Answer a question from ground truth by cloning the relevant codebases into a shared cache and digging into their source. Built for "how does X actually implement this" and "could X integrate with Y" questions that docs and search cannot settle.

License

MIT