🏥 AutoHeal AI

Autonomous Self-Healing Software Engine — Install once, self-heal forever.

AutoHeal AI is a language-agnostic CLI tool that wraps any command, detects errors in real-time, and automatically diagnoses and fixes them using AI.

✨ Features

• 🔍 Real-time Error Detection — Monitors stdout/stderr for errors across 8+ languages
• 🧠 AI-Powered Diagnosis — Uses LLM (GPT-4o, Claude, Gemini, Ollama) for root cause analysis
• 🔧 Auto-Fix — Restarts crashed processes, installs missing dependencies
• 💡 Smart Suggestions — Generates code patches with diff preview (user-approved)
• 📚 Learning Memory — Remembers past errors and solutions for instant future fixes
• 🔒 Safety-First — Never auto-applies code changes; secrets redacted before LLM calls
• 🌍 Language Agnostic — Works with Python, Node.js, Go, Rust, Java, Ruby, PHP, and more

🚀 Quick Start

Install

pip install autoheal-ai

Setup

# Initialize in your project
cd your-project/
autoheal init

# Set your LLM API key
autoheal config set llm.api_key sk-proj-your-key-here

# (Optional) Use a different provider
autoheal config set llm.provider anthropic  # or google, ollama

Run

# Monitor any command
autoheal run "python app.py"
autoheal run "npm start"
autoheal run "go run main.go"
autoheal run "cargo run"

📖 How It Works

You run:  autoheal run "python app.py"

┌─────────────┐     ┌──────────┐     ┌─────────────┐
│  Sentinel    │────>│   CECE   │────>│ Diagnostics  │
│  (Detect)    │     │ (Context)│     │ (Root Cause) │
└─────────────┘     └──────────┘     └──────┬───────┘
                                            │
┌─────────────┐     ┌──────────┐     ┌──────▼───────┐
│   Memory     │<────│ Harness  │<────│  Resolution  │
│   (Learn)    │     │ (Safety) │     │    (Fix)     │
└─────────────┘     └──────────┘     └──────────────┘

1. Sentinel watches your process output for errors
2. CECE builds full context (source code, environment, dependencies)
3. Diagnostics determines root cause via pattern matching + AI
4. Resolution generates the appropriate fix
5. Harness checks safety gates before applying
6. Memory stores the resolution for future instant fixes

🛠 CLI Commands

⚙️ Configuration

AutoHeal stores configuration in .autoheal/config.toml:

[llm]
provider = "openai"           # openai, anthropic, google, ollama
model = ""                    # Auto-selects best model per provider
api_key = "sk-proj-..."       # Your API key
temperature = 0.2

[resolution]
confidence_threshold = 0.75   # Min confidence for auto-fix
code_patch_threshold = 0.90   # Min confidence to suggest code patch
create_backup = true          # Git backup before patches

[general]
mode = "auto"                 # auto, suggest, manual
verbose = false

🤖 Supported LLM Providers

📝 License

MIT License — see LICENSE for details.