A MIT-licensed collection of AGENTS.md rules and skills for AI coding agents, derived from software design books, with releases up to v0.5 and a large set of books listed.
Collecting history — the radar snapshots this repo daily. The trend line appears after 3 days of data (1 so far).
What it is
MIT licensed universal project rules for coding agents. This repository contains ready-to-use rule sets inspired by well-known books on software design, architecture, refactoring, legacy code, reliability, and data-intensive systems. Each rule set is released in three tool-agnostic Markdown versions: mini, nano, and full.
How it works
The project provides editor-specific setup guidance in USAGE.md for Codex, Claude Code, and Cursor, describing always-on vs on-demand usage, skills, scoped rules, MCP or RAG patterns, and preferred setup for each editor. Rule sets are organized per book and multiple versions; the README shows a release-matrix structure and references a CHANGELOG.md for history.
Getting started
Key usage directions are distributed across the USAGE.md file (linked in the README). The repository also offers a books list with per-book links to full, mini, and nano rule files.
Recent releases
- v0.5 v0.5 (2026-05-04): Full Changelog: https://github.com/ciembor/agent-rules-books/compare/v0.4...v0.5
- v0.4 v0.4 (2026-05-03): Full Changelog: https://github.com/ciembor/agent-rules-books/compare/v0.2...v0.4
- v0.3 v0.3 (2026-05-02): Full Changelog: https://github.com/ciembor/agent-rules-books/compare/v0.2...v0.3
- v0.2 v0.2 (2026-05-02): Release tool agnostic rules with mini i nano versions.
Full Changelog: https://github.com/ciembor/agent-rules-books/compare/v0.1...v0.2
- v0.1 v0.1 (2026-05-02): Publish first release with AGENTS.md for Coder, Cursor and Claude Code for 13 books.
Full Changelog: https://github.com/ciembor/agent-rules-books/commits/v0.1
Traction
2376 stars, 409 forks, 5 open issues as of the data provided.
Books List
A compilation of books including A Philosophy of Software Design, Clean Architecture, Clean Code, Code Complete, Designing Data-Intensive Applications, Domain-Driven Design, and others, each with full, mini, and nano rule sets.
Does it work?
The README describes an evaluation setup comparing two branches: aposd-refactor-mini-rules and aposd-refactor-no-rules, using AGENTS.md rules to guide refactoring plans and assess outcomes. The experiment discusses scoring differences and a qualitative assessment of architectural judgment.