converge-planninglisted

Design a Converge playbook end-to-end: extract the goal, gather requirements, decompose it into deliverable tasks, decide what's a static child vs. a dynamically spawned template, factor reusable "how-to" into skills, and write the TASK.md + playbook.yml contracts with deterministic shell-level checks. Use this skill whenever the user says "plan a project", "design a playbook", "decompose this goal into tasks", "scaffold a Converge workflow", "onboard this codebase to Converge", "restructure work into a playbook", or asks how to author TASK.md / playbook.yml / SKILL.md files. Also use it before invoking `converge add` or `/converge-control` when no playbook exists yet — planning is the prerequisite for execution.
openplaybooks-dev/converge · ★ 3 · AI & Automation · score 76

Install: claude install-skill openplaybooks-dev/converge

# Converge Planning ## When to use this skill Trigger this skill whenever the user is about to author or restructure a Converge playbook. Concrete signals: - A fresh project with no `.converge/` directory and a goal like *"build me a SaaS app"*, *"automate this multi-step research"*, or *"turn this monorepo into a buildable playbook"*. - An existing project where the user wants to add a playbook for a new workflow (a new fan-out, a new epoch loop, a new domain split). - An existing playbook that needs restructuring — tasks have grown middle-work, the goal tree feels flat, or instructions are duplicated across many TASK.md bodies and would benefit from skill extraction. - The user asks any of: *how do I split this into tasks?*, *what should the DAG look like?*, *should this be a static child or a runtime spawn?*, *where should this skill live?*, *what should TASK.md contain?* If `.converge/project.yaml` doesn't exist yet, the user should run `converge init --skills` first; this skill assumes a scaffolded project and produces the `playbooks/<name>/` structure inside it. After this skill finishes producing contracts, hand off to `/converge-control` for execution. ## 1. Core Mental Model **Start with the playbook. Work backwards from the finished result.** Converge is not trying to freeze work into a brittle static workflow. The playbook is the durable artifact: a living specification that can branch, spawn new work, adapt to the state of the repo, and keep