| name | book-distiller |
| description | Distills books (PDF, EPUB, TXT) into structured AI Agent skills and Gemini Gems. Use when you need to transform a book's methodology into an actionable AI persona. |
| origin | firstsun-dev/skills |
Book Distiller
This skill provides a workflow for transforming long-form content (books, manuals, guides) into high-signal AI Agent instructions.
Dependencies
The parsing scripts require the following Python libraries:
pymupdf (for PDF)ebooklib (for EPUB)beautifulsoup4 (for EPUB cleaning)
Install them via: pip install pymupdf ebooklib beautifulsoup4
Core Workflow
1. File Parsing
Use the scripts/parse_book.py script to extract raw text from the source file.
python3 scripts/parse_book.py path/to/book.epub --output temp_raw.txt
2. Knowledge Distillation
Read the extracted text and use the prompt templates in references/PROMPT_TEMPLATES.md to distill the content.
Distillation Strategy:
- Persona Extraction: Identify the "Expert Role" the author represents.
- Principle Extraction: Identify the "Hard Truths" and "Axioms".
- Workflow Extraction: Identify "Step-by-Step" implementations.
- Constraint Extraction: Identify "Mandates" and "Anti-patterns".
3. Synthesis & Export
Follow the skill-manager SOP to save and register the results.
- Taxonomy: Ensure the new skill follows the "Flattened Design" and is placed in
custom/books/<book-slug>/.
- Refactoring: If the distilled content is large (>100 lines), apply the Progressive Disclosure pattern as defined in REFACTORING.md.
- Registration: Run
./setup.sh and update SKILLS_LIST.md.
- Gem Export: Follow the Gem Export Workflow in
yao-skill-manager to generate gem/<book-slug>.txt.
Guidance for AI Agents
When distilling, focus on Actionable Utility over narrative description.
- Good: "Workflow: Before starting task X, check for condition Y."
- Bad: "The book discusses the importance of Y in the context of X."
If the book is extremely large, distill chapter by chapter and consolidate the findings at the end.
