Skills & Technical Knowledge

A personal library of reusable AI skills, prompts, and instructions, plus a Go CLI that installs skills into your projects and keeps them up to date. Supports Claude Code, Codex, and Gemini CLI.

Browse my skill catalog. These skills are personal and shaped around my daily workflow and tools. Take ideas from them, and use any that work for you.

On macOS, install with Homebrew:

brew install --cask madalinpopa/tap/skills

Or install with Go:

go install github.com/madalinpopa/skills@latest

Or download a binary for macOS, Linux, or Windows from Releases. You also need Git 2.45 or newer on your PATH.

Run these commands from a project:

skills init                 # create the config and clone the skill store
skills ls                   # list published skills
skills install demo         # install a skill into this project
skills ls --local           # list this project's installed skills
skills sync                 # fetch the latest store content
skills update               # update installed skills from the store
skills remove demo          # uninstall a skill

The store is a local clone of your skills repository, shared by all your projects. sync pulls new changes into it. update copies those changes into your installed skills. You don't have to run init first: the CLI sets things up the first time a command needs the store.

Your local edits are safe. If you changed an installed skill, update skips it. Run skills diff <skill> to see what you changed. Add --force to install, update, or remove to replace your edits; the CLI backs them up first.

Add --dry-run to see what would change without writing anything, and -v to list each file. Run skills <command> --help for all options. Exit code 3 means a skill needs your attention. skills version shows the CLI version and the store commit.

The output below is an example. Use a skill name from skills ls instead of demo, and claude instead of codex if that is your agent. $ marks the command you type.

Install a skill for one agent:

$ skills install demo --agent codex
  + demo   added

  1 skill, 1 changed

Pull new changes, then update a skill you have not edited. This assumes the store has a newer version of demo:

$ skills sync
  pulled store  a1b2c3d -> e4f5a6b

$ skills update demo --agent codex
  ~ demo   updated

  1 skill, 1 changed

If you edited the skill, the CLI keeps your version and tells you what to do:

$ skills update demo --agent codex
  ! demo   skipped, you edited it

  1 skill, 0 changed, 1 needs attention
  Run 'skills diff demo --agent codex' to see your changes,
  or 'skills update demo --agent codex --force' to overwrite (backed up).

Run the diff command to see your edits. Use --force only when you want the store version instead.

See what removing a skill would do, without changing any files:

$ skills remove demo --agent codex --dry-run
  - demo   would remove
      would back up first

  1 skill, 1 would change

Drop --dry-run to remove it for real. Add -v to see each file under the skill.

You don't need to fork this repository to have your own skill library. Create a Git repository with a skills/ folder, one folder per skill:

skills/
  go-review/
    SKILL.md                the skill itself
    agents/openai.yaml      optional, copied only for Codex
    references/             optional, copied for every agent

The folder name is the skill name. The CLI reads only committed files under skills/, so the rest of the repository can hold anything you like. Look at the demo skill for a full example.

Then point the CLI at your repository in ~/.config/skills/config.toml:

[store]
repo = "https://github.com/your-name/your-skills"
branch = "main"

If you already ran the CLI with another repository, move the old store aside first, since it may hold local work you want to keep:

mv ~/.config/skills/store ~/.config/skills/store.old
skills init

Skills you installed from the old repository stay as they are. install and update skip them, even with --force. To switch one over, run skills remove and then skills install again.

To publish a skill, commit and push it to your branch, then run skills sync. A skill with status: draft in its frontmatter is not installed; a skill with no status is published. Changing a skill never needs a new CLI release.

The first run creates ~/.config/skills/config.toml and a store at ~/.config/skills/store/. If XDG_CONFIG_HOME is set, the CLI uses it. All projects on your machine share the config and the store.

By default, skills are installed for both agents:

Gemini CLI also reads .agents/skills/, so --agent codex covers it too. Use --agent to pick one agent, and --global to install for all your projects:

skills install demo --agent codex
skills install demo --global
skills ls --global

In a project, skills go into the repository root. Outside a repository, they go into the current folder. Edit the config to change agent paths or defaults. The default configuration shows the full format.

Fork this repository only if you want to work on the CLI itself. docs/SPEC.md describes how the CLI behaves, and AGENTS.md describes how changes are made.

Use Task for local checks:

task test:unit           # tests without network or Docker
task test:integration    # integration tests; requires Docker
task lint                # lint and dependency checks
task format              # format Go code

See Taskfile.yml for all tasks.