A set of human-AI collaboration playbooks that can be installed, migrated, and iterated continuously.
What is agent-project-playbook?
This skill provides a structured playbook for human-AI collaborative project management, enabling consistent handoffs, state preservation, and document governance across development cycles. It works with any AI agent that supports custom skills (Claude Code, Cursor, Codex, etc.) by defining clear workflows, document boundaries, and linting steps. The playbook ensures that agents read the latest handoff document (`CURRENT_HANDOFF.md`) at task start, verify claims against actual repository state, and update only the necessary stable documents (`BLUEPRINT.md`, `ARCHITECTURE.md`) when relevant changes occur. It also enforces line limits on key documents to keep them concise, uses Git for history instead of appending logs, and includes a lint script (`docs_lint.py`) to validate document compliance before substantial deliveries. This reduces confusion, prevents drift, and makes project collaboration predictable and scalable.