MbarQ's shared playbook for agentic development with Claude Code — one task loop, a catalog of skills, and a contribution loop that compounds.
click a step to expand →
Reusable workflow modules Claude invokes. Filter by where they sit in the loop, who they're for, and how they fire.
A playbook is a use-case-specific recipe — owned by the specialist who knows that use case best — encoding end-to-end delivery from requirements through verification. The vision: at least one playbook for every frequent use case.
Once a use case has a playbook, delivering it becomes routine — the agent builds it end-to-end and a human just approves.
Prefer the vendored mattpocock skill over its superpowers:* twin.
The QA gate is the highest-leverage move — nothing leaves until it passes. The lifecycle is driven by the M-Workflow desktop app.
Capture friction once, ratify it together, land it where it can’t rot — then everyone pulls the stronger baseline.
Every ratified learning gets encoded — at the strongest feasible rung. The higher it climbs, the less anyone has to remember. A rule a machine enforces can’t rot.
Your move: push every fix as high up the ladder as it’ll go. Climb past the line and it stops being something to remember.