Structured prompt templates for Claude Code. PR reviews, release notes, audits, blog posts, and more.
You use Claude for repetitive tasks—reviewing PRs, writing release notes, auditing code—but get inconsistent results because you write prompts from scratch each time.
Pre-built, battle-tested prompt templates that produce structured, actionable output every time.
| What You Get | Why It Matters |
|---|---|
| 6 specialized skills | Right template for each task type |
| Structured output | Same input → same format, always |
| Professional artifacts | Audit reports, changelogs, blog posts |
| Zero configuration | Install and start using immediately |
Inside Claude Code, run:
/plugin install andreolf/playbooks-plugin
Done! Start using any skill immediately.
| Skill | Command | What It Does |
|---|---|---|
| PR Review | /review-pr |
Structured code review with High/Medium/Low risk ratings and test plan |
| Release Notes | /release-notes |
Changelog, upgrade notes, and announcement copy for multiple channels |
| Ship Blog | /ship-blog |
Complete blog post with 5 titles, outline, draft, TLDR, and X thread |
| Audit Contract | /audit-contract |
Security audit with severity ratings, checklist, and Go/No-Go recommendation |
| Token Launch | /token-launch |
Launch checklist with T-21 to T+7 timeline and failure mode analysis |
| Panel Questions | /panel-questions |
Moderator guide with 12 sharp questions and follow-ups |
You: "/review-pr" + paste diff
↓
┌─────────────────────────────────────────┐
│ Playbooks Plugin │
│ → Loads review-pr skill template │
│ → Applies structured output rules │
│ → Enforces consistent format │
└─────────────────────────────────────────┘
↓
Claude: Structured review with:
• Summary (3 bullets)
• High/Medium/Low risk issues
• Test plan
• Follow-up suggestions
Key features:
- Each skill enforces a specific output schema
- Deterministic: same input produces consistent structure
- Never forgets key sections (test plans, upgrade notes, etc.)
- Works with any codebase or context
/review-pr
Review this authentication change:
diff --git a/src/auth.ts b/src/auth.ts
+ import bcrypt from 'bcrypt';
+ const hash = await bcrypt.hash(password, 10);
...
Output includes:
- Summary bullets
- Risk-categorized issues with fixes
- Test plan (unit, integration, edge cases)
- Suggested follow-ups
/release-notes
Product: MyApp
Version: 2.0.0
Changes:
- Added dark mode support
- Fixed login timeout after 30 mins
- Deprecated /api/v1 endpoints
- Breaking: Changed auth token format
Output includes:
- Executive summary
- Categorized changelog (Added/Fixed/Breaking/etc.)
- Upgrade steps and gotchas
- Announcement copy (Twitter, blog, dev-focused)
/ship-blog
Topic: Why we rewrote our auth system in Rust
Audience: Backend engineers
Angle: Performance gains with real benchmarks
Output includes:
- 5 title options
- Hook line (140 chars)
- Full outline (H2/H3)
- 800-1200 word draft
- TLDR bullets
- 7-tweet X thread
- 3 CTA variants
✅ Use for:
- PR reviews that need consistent structure
- Release notes with proper changelog format
- Blog posts with full distribution package
- Security audits with Go/No-Go decisions
- Any task where you want repeatable quality
❌ Skip for:
- Simple one-off questions
- Tasks without a standard output format
- Creative writing that needs to be freeform
Each skill is a markdown file in skills/*/SKILL.md. Fork the repo to:
- Modify output schemas
- Add new skills
- Adjust rules for your workflow
git clone https://github.com/andreolf/playbooks-plugin
cd playbooks-plugin
# Test locally
claude --plugin-dir /path/to/playbooks-pluginPRs welcome! To add a new skill:
- Create
skills/your-skill/SKILL.md - Follow the existing skill format (frontmatter + rules + output schema)
- Test with Claude Code
- Submit PR
MIT — see LICENSE
Inspired by claude-playbooks, the CLI version of this concept.
Get consistent, professional output from Claude — every time.