Skip to content
Get packs

Book SFT Pipeline

Turns a book into a fine-tuning dataset for training a model to mimic an author's voice

Source Download skill
Repository
muratcankoylan/Agent-Skills-for-Context-Engineering
Stars
18,282
Licence
MIT
Files
8 files

What it does

This skill guides an AI agent through converting a book (ideally an ePub file) into a labeled training dataset of text chunks, synthetic instructions, and example pairs, then through training a small style-transfer model on a platform called Tinker or a similar tool. Once installed, the agent follows a defined sequence: extract paragraphs, split text into roughly 150-400 word chunks, generate varied prompts, and format the output as chat-style JSONL. The skill also includes a conceptual Python script and reference notes on segmentation and data format.

Great for

  • Developers building fine-tuning datasets from literary works
  • Researchers experimenting with style-transfer training on small models
  • Teams who already have a Tinker or similar SFT platform account

Skip it if

  • Anyone wanting to copy or redistribute copyrighted books: the skill builds training data from full book text, which may raise copyright and licensing issues depending on the source and jurisdiction
  • Users without ML training experience: the steps assume familiarity with LoRA, SFT, JSONL, and model evaluation, and the bundled script is described as conceptual rather than production-ready

Try asking

  • “Help me turn this ePub novel into a chunked JSONL dataset for voice fine-tuning”
  • “Design a segmentation strategy that keeps paragraphs intact for a 90,000-word book”
  • “Generate 15 prompt templates and 5 system prompts for an author-style dataset”
  • “Format my extracted chunks into the Tinker message format for LoRA training”
Step-by-step guide: how to use Book SFT PipelineInstall, first run, examples, tips and FAQs

Install in 3 steps

  1. 1

    Download the skill

    The ZIP contains the 8 original files, the MIT licence and an attribution note.

    book-sft-pipeline.zip
  2. 2

    Add it to your agent

    Claude Code (all projects)

    mkdir -p ~/.claude/skills && unzip book-sft-pipeline.zip -d ~/.claude/skills/

    Claude Code / Agent SDK (this project only)

    mkdir -p .claude/skills && unzip book-sft-pipeline.zip -d .claude/skills/

    Claude apps (web/desktop): Settings → Capabilities → Skills → Upload skill, and choose the ZIP.

  3. 3

    Ask for the task

    No special command needed — the agent reads the skill's description and loads it when your request matches. Try one of the prompts above.

Get better results

  • Start with a public-domain book to avoid licensing problems, and check rights before using any in-copyright text
  • Treat the bundled pipeline_example.py as a starting point; the segmentation and instruction functions need adapting before real use
  • Review generated synthetic instructions and the AI-detection and originality checks the skill describes, since training on verbatim text can cause memorization

What's inside

  • examples/gertrude-stein/README.md
  • examples/gertrude-stein/sample_outputs.md
  • examples/gertrude-stein/training_config.json
  • README.md
  • references/segmentation-strategies.md
  • references/tinker-format.md
  • scripts/pipeline_example.py
  • SKILL.md

Redistributed unmodified from muratcankoylan/Agent-Skills-for-Context-Engineering under the MIT licence. The assessment was drafted with AI from the skill's files on 2026-10-29. Skills can instruct your agent to run code — read SKILL.md and any scripts before installing.