Playbook Format
Playbooks are end-to-end learning guides with a human-agent split: humans make decisions, agents handle execution.
Structure
Each playbook is an MDX file in playbooks/<locale>/playbook-<name>.mdx.
Sections
1. Overview
- Title and one-line description
- Target audience
- Estimated time
- Difficulty level (Beginner / Intermediate / Advanced)
2. The Pain
What problem does this playbook solve? What's the user struggling with?
3. The Learning Logic
Evidence-based reasoning for why this approach works.
4. The Human-Agent Split
Clear division of responsibilities:
- Human does — Decision-making, goal-setting, quality judgment
- Agent does — Research, drafting, analysis, execution
5. Steps (3-5)
Each step includes:
- Clear objective
- Human actions
- Agent actions
- Expected output
6. Skill Pack
Required skill packages:
clawford install @clawford/google-search @clawford/summarizer7. Expected Output
What the user should produce by the end of the playbook.
Categories
- General — Learning, writing, coding, research
- Crypto — Trading, DeFi, KOL operations, token launches
Localization
Playbooks are available in English (playbooks/en/) and Chinese (playbooks/zh/). The system automatically selects the right version based on the user's locale.