playbook-usage
Lists /playbook:* commands and when to use each. Use when asked what the plugin does or which command fits a task.
Instal·la
npx skillscat add pragmatic-engineer/playbook/playbook-usage Instal·la des del registre de SkillsCat.
Playbook Usage
pragmatic-engineer/playbook is a Claude Code plugin: slash commands under /playbook:* plus supporting skills and hooks. The main flow is a planning pipeline: /playbook:plan (or /playbook:adr for a hard-to-reverse decision), then /playbook:implement, which takes a raw idea through a verified plan to delivered code. /playbook:plan runs the moment the intent matches. /playbook:adr and /playbook:implement only offer in one line and wait for a yes. Each command's own description lists its trigger phrases.
Planning pipeline
| Command | Use when |
|---|---|
/playbook:plan |
An idea needs to become a plan, raw or settled. One interview-driven session that challenges the premise, weighs 2-3 approaches, then produces a verified plan of Work Units grouped into PR-sized Segments. |
/playbook:adr |
Choosing between named alternatives, a decision that is expensive to undo, or recording why a choice was made. Saves a fact-checked ADR with an optional blueprint to docs/adr/. |
/playbook:implement |
An approved plan or blueprint exists. Delegates Work Units to subagents, commits each, delivers PR-sized Segments, ends with a refinement pass and an adversarial review. --boundary=land merges each Segment before starting the next. Never designs new scope. |
Review
| Command | Use when |
|---|---|
/playbook:review |
Reviewing a PR or your own branch. Start here. It recommends a quick or a deep review from the diff, asks you to pick, then runs it. --quick and --deep skip the question. |
/playbook:quick-review |
A routine PR review, or a quick check of your own branch. Single pass, asks which findings to post as a pending GitHub review, or only reports with --self or --no-post. |
/playbook:deep-review |
Direct entry for the deep review. A substantial, risky or cross-cutting PR. A parallel swarm of specialist reviewers, fact-checked into one pending review. |
/playbook:address-pr-comments |
Working through open review feedback on a PR, one comment at a time. |
Delivery
| Command | Use when |
|---|---|
/playbook:commit-and-push |
Committing staged changes with a generated, signed message and pushing, instead of hand-running git commit and git push. |
/playbook:create-pull-request |
Opening a PR with pre-flight checks, a conventional-commit title and the team template, instead of hand-running gh pr create. Then run the playbook:finish-pull-request skill. |
/playbook:fix |
A small, well understood bug. A failing test, the smallest fix, one PR. Escalates to /playbook:plan when the fix is not small. |
Utilities
| Command | Use when |
|---|---|
/playbook:repo-audit |
Assessing an unfamiliar or long-lived repo without changing it. Read-only, every claim cited to file:line. |
/playbook:learn-project |
Building durable project knowledge in the memory store from git history, PRs, Jira and Confluence. |
/playbook:doctor |
Checking the seven playbook layers after install or update, or when hooks seem not to run. |
/playbook:session-start |
After /clear, when the saved handoff did not load. The /playbook:session-handoff command writes the handoff. |
/playbook:setup |
First install, or repairing the plugin's own settings. |
Delivery pattern
/playbook:commit-and-push, /playbook:create-pull-request and /playbook:repo-audit run in an isolated forked context (context: fork). /playbook:implement delegates each Work Unit to implementer subagents, some in isolated git worktrees. A subagent's return value alone is not a reliable signal of what it did. See playbook:delegating-subagents for the file-based handoff discipline.
The memory store, its scopes and its edge types are described in docs/guides/03-decisions-and-memory.md.
Categories
Instal·la
npx skillscat add pragmatic-engineer/playbook/playbook-usage Instal·la des del registre de SkillsCat.