Skip to content

CI/CD Workflows

GitHub Actions workflows that automate testing, releasing, and documentation deployment for Zolletta-metaskill.

Workflow overview

Workflow File Trigger Scope
Tests .github/workflows/tests.yml Push (all branches), pull request pytest + coverage, ruff, ty, mypy, vulture
Semantic Release .github/workflows/semantic-release.yml Push to main Calls Tests, then creates a version tag + GitHub Release via python-semantic-release
Docs (Pages) .github/workflows/docs-pages.yml Push to main (docs changes only) Builds MkDocs Material site and deploys to GitHub Pages

Tests

Runs on every push and pull request. Executes the full test suite with coverage reporting, followed by static analysis (ruff, ty, mypy, vulture). Also callable as a reusable workflow by Semantic Release.

  • Triggers: push (all branches), pull_request, workflow_call
  • Jobs: pytest → quality (quality depends on pytest passing)
  • Coverage: Uploaded to Codecov

Semantic Release

Runs only on pushes to main. It calls the Tests workflow as a dependency, then — if all checks pass — determines the next version from commit history, stamps it into pyproject.toml, creates a git tag, and publishes a GitHub Release.

  • Trigger: push to main
  • Depends on: Tests workflow (via workflow_call)
  • Versioning: python-semantic-release with conventional commits
  • Release types: feat → minor, fix/perf/refactor → patch, breaking changes → major
  • Ignored types: chore, docs, deps, ci (no version bump)
  • Authentication: Uses the built-in GITHUB_TOKEN

Docs (Pages)

Runs only when documentation files change on main. Builds the MkDocs Material site and deploys it to GitHub Pages at metaskill.zolletta.org.

  • Trigger: push to main with changes in docs/**, mkdocs.yml, or the workflow file itself
  • Concurrency: Only one deployment at a time; in-progress runs are cancelled
  • Path filters: Does not run on code-only changes

Shared setup

All workflows use a composite action (.github/actions/setup-env/action.yml) that installs uv, sets up Python, caches the .venv directory, and runs uv sync. This avoids duplicating setup steps across jobs and speeds up runs via cache reuse.