Contributing to the kaboo stack
The kaboo stack is five repositories that share one contribution model. Each repo
has a CONTRIBUTING.md (for humans) and an AGENTS.md (for AI contributors);
this page is the shared entry point that links them all.
The repositories
| Repo | Role | Contributing | AI guide |
|---|---|---|---|
| kaboo-workflows | Python YAML multi-agent orchestration → AG-UI SSE | CONTRIBUTING | AGENTS |
| kaboo-runtime | CopilotKit runtime persistence (AgentRunner + ThreadStore) | CONTRIBUTING | AGENTS |
| kaboo-react | React agent-activity UI | CONTRIBUTING | AGENTS |
| kaboo-workflows-demo | Runnable end-to-end demo | CONTRIBUTING | AGENTS |
| kaboo-docs | This umbrella landing | CONTRIBUTING | AGENTS |
Shared principles
- Docs are proven. Library code snippets are type-checked/executed in CI; keep examples runnable and in sync with the source.
- Everything is documented. No undocumented public exports; the API-reference completeness gates fail otherwise.
- AI-native. Each library ships
AGENTS.mdplusllms.txt/llms-full.txtso agents can learn the APIs directly. - Conventional Commits and green CI (build, tests, docs gates, package/dry-run checks) on every PR.
- No secrets in git. The demo's
.envholds a live key and is gitignored; always redact keys in docs.
Where changes go
- A library's behavior, API, or its own docs → that library's repo.
- How the pieces run together, ports, env, startup → the demo.
- This landing (stack intro, library cards, this page) →
kaboo-docs.
Editing this site
pip install mkdocs-material
mkdocs serve # live preview at http://127.0.0.1:8000
mkdocs build --strict # must be clean (no warnings / broken links)