ProAgents

AI-Agnostic Development Workflow Framework

A portable, universal development workflow framework that works with any AI platform (Claude, ChatGPT, Gemini, Cursor, Copilot, etc.) and enables multi-AI collaboration on the same project.

📚 COMMANDS.md - Complete Command Reference (100+ Commands)

CLI commands, AI commands (pa:), workflow phases, debugging, testing, deployment, and more.

Why ProAgents?

• AI-Agnostic: Works with Claude, ChatGPT, Gemini, Cursor, Copilot - switch AI tools anytime without losing context
• Multi-AI Collaboration: Multiple AIs can work on the same project with shared context and conflict detection
• Guided Workflow: 10-phase development process from Analysis to Deployment with quality gates
• Learning System: AI learns from your corrections and never repeats the same mistakes

Installation

No install required. Just run:

npx proagents init

Quick Start

1. Initialize ProAgents

cd your-project
npx proagents init

What happens:

• Creates ./.proagents/ folder with workflow files
• Prompts for project name and type (auto-detected)
• Prompts for tech stack (API style, database, styling, auth)
• Creates AI instruction files (CLAUDE.md, .cursorrules, etc.)

2. Use Commands in Any AI

Open your AI assistant (Claude, ChatGPT, Cursor, etc.) and type:

pa:feature "Add user authentication"

What happens:

• AI loads project context automatically
• Guides you through 10-phase workflow
• Logs all changes for other AIs to see

3. See Progress

pa:status

Example output:

Feature: Add user authentication
Phase: Implementation (5/10)
Progress: ████████░░░░ 50%
Next: pa:test

Features

Multi-AI Collaboration

Multiple AIs can work on the same project with full context sharing:

| Feature | Description | |---------|-------------| | Auto Context Loading | AI loads project context automatically on first command | | Auto Change Logging | Every code change logged with AI model name | | Conflict Detection | Warns if files were modified by another AI | | Feature Changelogs | Per-feature and per-module change history | | Quick Undo | pa:undo-last reverts last AI's entire session | | Activity Tracking | Track what each AI does with timestamps | | Handoff Notes | pa:handoff creates notes for next AI session | | Test Status Sync | Auto-updates test status after every test run | | Progress Tracking | Auto-calculates feature progress from tasks |

pa:feature "add login"    # AI auto-loads context, then works
pa:undo-last              # Revert last AI's changes if needed

AI Learning & Feedback

• Feedback Log - AI learns from corrections, doesn't repeat mistakes
• Error Tracker - Log errors and solutions for future reference
• Decision Log - Track architectural decisions with reasoning
• Context File - Persistent project knowledge AI reads every session

Full Automation

ProAgents follows "ALWAYS DO, NEVER JUST TELL" principle:

| Instead of | ProAgents does | |------------|----------------| | "Run npm test" | Actually runs the tests | | "Create a test file" | Creates the file | | "Fix the bug by..." | Fixes the bug directly | | "Next steps: 1. 2. 3." | Does all steps automatically |

Auto-fix intelligence:

• Learns from previous fixes
• Applies known patterns automatically
• Stores successful fixes in .proagents/.learning/

File Protection

• Watch List - Critical files require confirmation before AI modifies
• Never auto-modify .env, config files, migrations, etc.

CLI Commands

# Initialize
npx proagents init                    # Initialize or update ProAgents

# Daily Use
proagents status                      # Show ProAgents status
proagents doctor                      # Health check installation
proagents stats                       # Show project & AI usage stats

# AI Platforms
proagents ai list                     # List installed AI platforms
proagents ai add                      # Add more platforms

# Release
proagents release                     # Generate release notes
proagents release --changelog         # Update CHANGELOG.md

📖 Want more? See COMMANDS.md for 30+ CLI commands (release notes, stats, config, backups, shell completions...)

Essential AI Commands (pa:)

Type these in any AI assistant (Claude, ChatGPT, Gemini, Cursor, etc.):

| Command | Description | |---------|-------------| | pa:feature "name" | Start new feature workflow (all phases) | | pa:fix "description" | Quick bug fix mode | | pa:project-setup | Interactive project setup wizard | | pa:rnd | Research & Development workflow | | pa:status | Show current progress | | pa:sync | Load project context (usually auto) | | pa:test | Run tests | | pa:review | Code review workflow | | pa:deploy | Deployment preparation | | pa:help | Show all commands | | pa:undo-last | Revert last AI's changes | | pa:handoff | Create notes for another AI |

Quick aliases: pa:f (feature), pa:s (status), pa:t (test), pa:h (help), pa:setup (project-setup), pa:research (rnd)

📖 Want more? See COMMANDS.md for 100+ AI commands (debugging, testing, refactoring, code generation, sprints...)

Supported AI Platforms

ProAgents works with any AI that has agentic capability (can read files and execute commands). Pre-configured instruction files are available for:

| Platform | Instruction File | |----------|-----------------| | Claude Code | CLAUDE.md | | Cursor | .cursorrules | | Windsurf | .windsurfrules | | GitHub Copilot | .github/copilot-instructions.md | | ChatGPT | CHATGPT.md | | Gemini | GEMINI.md | | Bolt | BOLT.md | | Lovable | LOVABLE.md | | Replit | REPLIT.md | | Kiro | KIRO.md | | Groq | GROQ.md |

Using a different AI? Point it to .proagents/AI_INSTRUCTIONS.md - it contains everything needed to work with ProAgents.

Documentation

| Document | Description | |----------|-------------| | 📚 COMMANDS.md | 100+ CLI & AI commands | | Getting Started Story | Narrative walkthrough | | Complete Workflow | 10-phase workflow guide | | AI Instructions | Full AI command reference | | Examples | Project-specific walkthroughs |

Contributing

Contributions are welcome! Please feel free to submit a Pull Request.

License

MIT License - see LICENSE file.

Built for developers who want AI collaboration without vendor lock-in.

Co-authored with Claude (Anthropic)