Publishing agent-team

How a version of agent-team reaches people. Each step that publishes something is done by a maintainer on purpose; nothing here runs on its own.

Release a version

1. Bump the version in pyproject.toml, src/agent_team/__init__.py, plugin/.claude-plugin/plugin.json and PLUGIN_VERSION in plugin/hooks/register.tsx (tests/test_versions.py checks they match), and turn ## [Unreleased] in CHANGELOG.md into ## [X.Y.Z] - YYYY-MM-DD.
2. Commit, then tag and push: git tag vX.Y.Z && git push origin vX.Y.Z.
3. .github/workflows/release.yml checks the tag matches the version, runs the tests, builds the package, publishes it to PyPI and creates the GitHub release with that version’s notes.

One-time setup

• PyPI trusted publishing. On pypi.org, add a pending publisher for the project agent-team: owner lakeba, repository agent-team, workflow release.yml, environment pypi. In the GitHub repository, create the environment pypi (Settings → Environments). No token is stored anywhere.
• Homebrew tap. Create lakeba/homebrew-tap with Formula/agent-team.rb copied from packaging/homebrew/agent-team.rb. After each PyPI release, set its url and sha256 and run brew update-python-resources Formula/agent-team.rb. Users then brew install lakeba/tap/agent-team.
• Docs site. Settings → Pages → Deploy from a branch → main, folder /docs. The site is https://lakeba.github.io/agent-team/.
• Claude Code plugin. The repository is its own marketplace (.claude-plugin/marketplace.json): /plugin marketplace add lakeba/agent-team. Submitting it to a public plugin directory follows that directory’s own process.