Show HN: Book-skills – Agent Skills distilled from mgmt and programming books
**Show HN** introduced **book-skills**, an open-source collection of Claude Code skills that encode frameworks from management and programming books people…
By Dillip Chowdary • Aug 04, 2026 • Source: HN AI Agents
**Show HN** introduced **book-skills**, an open-source collection of Claude Code skills that encode frameworks from management and programming books people already use at work. Each skill maps to a single title—**The Lean Startup**, **High Output Management**, **The Pragmatic Programmer**, and similar books—rather than to a generic “productivity” prompt. The author built it after experimenting with Claude Code skills and asking whether those same skill files could carry the methods they reapply on real projects, not just loose advice.
The design is deliberate about what a skill is not. **book-skills** does not aim to summarize the books. Each skill turns a book’s methodology into something the agent can run: install the skill, then ask Claude in the usual way and get behavior shaped by that framework. The Lean Startup skill is the concrete example: once installed, prompts go through Lean Startup practice instead of a one-off recap of the text. That is skills as executable procedure, not as reading notes.
Advertisement
Tech Pulse Daily
Get tomorrow's pulse first
Join engineers who read Tech Pulse before stand-up. Free, weekday mornings.
For engineers and builders already in agent workflows, that matters because framework knowledge usually lives in memory, notes, or habit. Encoding it as a Claude Code skill makes the same method available in the tool loop where work actually happens—planning, iteration, prioritization—without re-explaining the book every session. It also treats skills as a packaging unit for domain process: one book, one installable behavior surface, reusable across tasks.
On the market side, this sits in the same Show HN / HN AI Agents lane as other agent tooling: open-source, install-and-prompt, built on Claude Code’s skills model rather than a closed product. Competitors and peers are other skill packs, prompt libraries, and agent playbooks; **book-skills** differentiates by grounding each pack in a named, widely used book methodology instead of ad hoc house rules. It is less “AI book club” and more “methodology as agent configuration.”
Practical next step for builders is narrow: try one skill that matches work you already do (Lean Startup for product experiments, High Output Management for management cadence, The Pragmatic Programmer for engineering practice), install it, and run real tasks against it. Watch whether the agent stays faithful to the book’s process under pressure, how install and discovery feel for a multi-book collection, and whether the pattern spreads beyond this set—more titles, tighter task-to-skill routing, or skills derived from internal playbooks the same way these books were distilled.
Advertisement