book-to-skill

What book-to-skill does

You point book-to-skill at a file, folder or glob, for example /book-to-skill ./my-book.pdf. It distills the material into frameworks, decision rules, anti-patterns and per-chapter files rather than a summary. The generated skill has a core SKILL.md of about 4,000 tokens holding mental models and a chapter index, chapter files of about 1,000 tokens each that load only when a topic comes up, and glossary, patterns and cheatsheet files.

The output follows the open Agent Skills SKILL.md format, so one skill can be read by GitHub Copilot CLI, Amp, Claude Code, Hermes Agent, OpenCode and OpenClaw. By default it writes to the cross-agent directory ~/.agents/skills/, but Hermes Agent partitions personal skills by category and does not scan that root, so a Hermes install is written into the Hermes skills directory under a category folder instead. The README reports 24x to 51x fewer tokens than putting a whole book in context, and says the same method works for internal docs, brand guidelines and research collections.

Key features

• Accepts a PDF, folder, glob or list of files as input
• Generates a SKILL.md with mental models and a chapter index
• Writes per-chapter files that load on demand, plus glossary, patterns and cheatsheet files
• Output works with hosts that read the Agent Skills format, including Hermes Agent
• Separate install location for Hermes Agent

When to use it

• Turning a technical book into a skill you query with a slash command such as /your-book-slug replication
• Folding a docs folder of runbooks and architecture decision records into one skill
• Turning a brand or voice guide into a skill a team can query

Who it is for: Developers and learners who own technical books or internal docs and want their coding agent to answer from them.

How it fits with Hermes Agent

Hermes Agent is one of six named host agents, and the README handles it separately because Hermes organizes personal skills by category and does not scan the cross-agent skills root.

FAQ