This repository offers a curated collection of AI coding agent rules, directly inspired by foundational software engineering books. It aims to empower developers using tools like Codex, Cursor, and Claude by making established principles of software design, architecture, refactoring, and reliability readily applicable during agent-assisted development. The project provides practical, ready-to-use instructions that translate complex book concepts into actionable guidance for AI coding assistants.

How It Works

The project translates core principles from influential software engineering texts into specific, operational instructions tailored for AI coding agents. Rules are organized by book and provided in formats compatible with Codex (AGENTS.md), Cursor (.cursor/rules/*.mdc), and Claude (.claude/rules/*.md). A key feature is the unified-software-engineering rule set, which synthesizes guidance from multiple books, resolving potential conflicts and establishing context-aware decision-making priorities for a comprehensive default experience.

Quick Start & Requirements

Installation involves copying the relevant rule files (e.g., AGENTS.md, .mdc, .md) into the appropriate directories within your project structure, such as the project root for Codex or .cursor/rules/ and .claude/rules/ for Cursor and Claude respectively. No specific software prerequisites beyond access to the target AI coding tools are detailed.

Highlighted Details

• Covers principles from over a dozen seminal software engineering books, including "Clean Code," "Domain-Driven Design," "Refactoring," and "Designing Data-Intensive Applications."
• Provides distinct rule formats optimized for Codex, Cursor, and Claude Code.
• Features a synthesized "unified-software-engineering" rule set designed as a broad, coherent default, merging and prioritizing guidance from various sources.
• Allows users to select specific rule sets based on task context, such as "release-it" for production systems or "working-effectively-with-legacy-code" for maintenance.

Maintenance & Community

The project is authored by Maciej Ciemborowicz. While the README references a Reddit discussion for community feedback, it does not provide direct links to active community channels (e.g., Discord, Slack) or detail a specific roadmap or ongoing maintenance schedule.

Licensing & Compatibility

The repository and its contents are released under the MIT License. This permissive license allows for unrestricted use, modification, distribution, and sublicensing, ensuring broad compatibility with commercial and closed-source projects.

Limitations & Caveats

Significant user-raised concerns include the absence of benchmarks to objectively measure the effectiveness of the rules, the potential for increased token consumption and context dilution when using numerous rules, and the risk of AI models exhibiting overengineering or pseudo-compliance. Conflicts between principles from different books and the reliance on book-derived rules over project-specific failure analysis are also noted as potential drawbacks.