Thanks for your interest in improving team-up. Here's how to get involved.
- Report bugs — something broke or didn't work as expected? Open an issue.
- Suggest features — new agent roles, workflow improvements, better coordination patterns? Open a feature request.
- Add or improve agents — create a new specialist agent or refine an existing one.
- Improve docs — fix unclear language, add examples, or improve the README.
- Create a new
.mdfile inagents/ - Follow the existing format — see any agent in that directory for the pattern
- Register it in
.claude-plugin/marketplace.jsonunder theagentsarray - Test it by running
/team-upand verifying the lead discovers and uses your agent - Open a PR
Every agent definition needs:
- YAML frontmatter with
model: opus(orsonnetfor lighter tasks) - Role — what the agent does and doesn't do
- Boundaries — explicit in-scope and out-of-scope
- Process — step-by-step workflow
- Quality standards — checklist before reporting done
- Communication — status codes:
DONE,DONE_WITH_CONCERNS,NEEDS_CONTEXT,BLOCKED
The team lead playbook lives in skills/team-up/SKILL.md. If you're changing how the lead coordinates, update this file. Keep the 7-phase structure unless you have a good reason to change it.
- Keep PRs focused — one change per PR
- Test with a real
/team-upinvocation before submitting - Update the README if user-facing behavior changes
By contributing, you agree that your contributions will be licensed under the MIT License.