Plans large, ambiguous, or multi-file changes by decomposing into verifiable steps with TodoWrite breakdown, then hands off to an execution skill.
How this skill is triggered — by the user, by Claude, or both
Slash command
/agent-coding-playbook:acp-plan-taskThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
Use this skill when the work is too big or too unclear to start coding directly.
Use this skill when the work is too big or too unclear to start coding directly.
The goal is not to write a plan for its own sake.
The goal is to produce the smallest plan that makes execution safe — clear scope, clear non-goals, clear verification, clear handoff.
Use this skill when ANY of these are true:
If the task is small and obvious, skip this skill and go straight to acp-bug-fix, acp-refactor, or acp-feature-add.
In one or two sentences, write what success looks like from the user's perspective.
If you cannot do this, the task is not ready for planning — go back and ask the user.
Use the Explore subagent to identify, without polluting the main session:
Quote file paths in the plan so reviewers can verify your reading is correct.
Break the work into steps that each:
Bad: "rewrite the auth module". Good: "1. add failing test for X; 2. extract helper Y; 3. swap implementation; 4. delete old path".
Record the steps in a TodoWrite list so progress is visible during execution.
For each step, write down how it will be verified:
pnpm test, cargo check, ...).A step without a verification path is not a step — it is a wish.
Present the plan via ExitPlanMode so the user can approve or push back before any code is written.
Do not start editing until the plan is approved.
Once approved:
acp-bug-fix skillacp-refactor skillacp-feature-add skillacp-test-first skillThis skill ends at handoff.
Goal:
- ...
Non-goals:
- ...
Affected files / modules:
- ...
Risks:
- ...
Steps:
1. <step> — verified by <how>
2. <step> — verified by <how>
3. ...
Handoff:
- This will be executed under the <skill-name> skill.
Stop and ask the user before continuing if:
npx claudepluginhub bravekingzhang/agent-coding-playbook --plugin agent-coding-playbookUse when you have a spec or requirements for a multi-step task, before touching code
Structures multi-step work into small, verifiable tasks with clear dependencies and verification criteria. Useful for implementing features, refactoring, or any multi-step work.
Generates detailed, step-by-step implementation plans from specs or requirements, with bite-sized tasks, file paths, and test-first guidance.