Skip to content

rosudrag/ai-praxis

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

18 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

AI Praxis

Make your AI coding assistant actually good at your codebase.

Get Started

1. Clone to a temp folder:

git clone https://github.com/rosudrag/ai-praxis.git /tmp/ai-praxis

2. Open your project in your AI coding tool and tell it:

Read /tmp/ai-praxis and use it to bootstrap my project

Your AI detects your stack, installs MCP tools, and generates everything. You can delete the cloned folder after.

Works with: Claude Code · Cursor · Windsurf · Cline · any MCP-compatible tool


What You Get

  • Custom AGENTS.md — AI instructions tailored to your actual tech stack and conventions
  • TDD & Code Quality Guides — Methodology docs your AI follows automatically
  • Semantic Code Navigation — Serena MCP gives your AI real understanding of your codebase
  • Up-to-Date Library Docs — Context7 MCP pulls current documentation into prompts
  • Structured Problem-Solving — Sequential Thinking MCP for complex debugging and design
  • Architecture Decision Records — Track and enforce architectural decisions with your AI

How It Works

AI Praxis has two modes that run automatically:

Mode When What it does
Environment Setup Once per machine Installs MCP servers (Serena, Context7, Sequential Thinking) and agents globally
Project Bootstrap Once per project Generates AGENTS.md, methodology guides, ADRs, and Serena config for your project

First-time users get both automatically. Your AI detects what's already done and skips it.

What Gets Created

In your project:

your-project/
├── AGENTS.md           # AI instructions for your codebase
├── ai-docs/            # TDD, security, research, code quality guides
├── docs/adrs/          # Architecture Decision Records
└── .serena/            # Semantic code understanding config

On your machine (global, shared across projects):

MCP servers configured in your AI tool — Serena, Context7, and Sequential Thinking.

Examples

See real-world AGENTS.md files generated for different stacks:

AGENTS.md Convention

This project follows the AGENTS.md convention — a standard for AI coding assistant configuration adopted by 20,000+ repositories. Claude Code users get an auto-generated CLAUDE.md forwarding file.

Customization

After bootstrapping, everything is yours to edit:

  • AGENTS.md — Add project-specific rules and conventions
  • ai-docs/ — Add or modify methodology guides
  • docs/adrs/ — Record architectural decisions
  • .serena/memories/ — Persist knowledge across sessions

Philosophy

  1. Methodology over configuration — A complete way of working, not just settings files
  2. Test-first thinking — Tests are hypotheses; state expectations, then prove them
  3. Research before assuming — Investigate systematically, don't guess
  4. Tool-agnostic — Works with any AI coding assistant, not just one vendor

Requirements

  • An AI coding assistant (Claude Code, Cursor, Windsurf, Cline, etc.)
  • Node.js 18+
  • Git

Contributing

See AGENTS.md for development instructions.

License

MIT

About

No description, website, or topics provided.

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors