gl-cursor-skill

gl-cursor-skill installs AI-agent skills between common local skill folders.

Primary usage:

uvx gl-cursor-skill add cursor-agent

The cursor-agent skill is bundled with this package, so the command works even when the source skill does not already exist on the machine.

When options are omitted in an interactive terminal, the CLI asks where to install:

? Install for which agent?
  Codex
  Claude Code
  OpenCode

? Install scope?
  Global
  Project

For scripts, pass the choices explicitly:

uvx gl-cursor-skill add cursor-agent --agent codex --scope global --from bundled --yes

Supported Targets

claude and claudecode are accepted aliases for claude-code. open-code is an alias for opencode.

Commands

gl-cursor-skill add <skill>
gl-cursor-skill agents
gl-cursor-skill list

Examples:

# Interactive install
uvx gl-cursor-skill add cursor-agent

# Non-interactive install from Claude global skills into Codex global skills
uvx gl-cursor-skill add cursor-agent --agent codex --scope global --from claude --yes

# Non-interactive install from the bundled cursor-agent skill
uvx gl-cursor-skill add cursor-agent --agent codex --scope global --from bundled --yes

# Install from an explicit local path
uvx gl-cursor-skill add /path/to/cursor-agent --agent codex --scope global --yes

# Preview without copying files
uvx gl-cursor-skill add cursor-agent --agent codex --scope global --from claude --dry-run --yes

# Overwrite an existing target skill
uvx gl-cursor-skill add cursor-agent --agent codex --scope global --from claude --force --yes

Development

uv sync
uv run pytest
uv run gl-cursor-skill --help
uv build

Release From GitHub

The repository includes GitHub Actions workflows:

• .github/workflows/ci.yml runs tests and builds the package on push and pull requests.
• .github/workflows/publish.yml publishes to PyPI when a GitHub Release is published.

Recommended setup:

1. Create a GitHub repository and push this project.
2. On PyPI, configure a Trusted Publisher for project gl-cursor-skill.
3. Use these PyPI Trusted Publisher values:

Owner: <your GitHub username or org>
Repository name: <your repo name>
Workflow name: publish.yml
Environment name: pypi

4. For each release, bump the version in pyproject.toml, tag the commit as vX.Y.Z, and publish a GitHub Release from that tag.

The workflow checks that the release tag matches the package version. For example, version 0.1.0 must be released with tag v0.1.0.

Manual local publish also works:

UV_PUBLISH_TOKEN="pypi-..." uv publish

If you configure PyPI Trusted Publishing in CI, uv publish can run without a token.