angel-claw 0.7.0

Angel Claw personal AI agent framework

🪽 Angel Claw

Your Personal AI Agent Framework with memory, skills, and multi-channel superpowers.

---

What Is Angel Claw?

Angel Claw is a powerful, extensible AI agent framework inspired by OpenClaw designed to give you:

• 🧠 Long-term, evolvable memory
• 🛠 Self-generating skills & load skills from clawhub
• 🔌 Model Context Protocol (MCP) Host: standardized external tool interoperability
• 📅 Proactive scheduling & automation
• 🌍 Web browsing & tool usage
• 📱 Telegram & WhatsApp integration

All powered by:

• angel-recall for agent-native memory
• litellm for multi-model LLM support
• FastAPI as the gateway layer

$ pip install angel-claw
# replace .env values
$ angel-claw chat

---

🎬 Quick Demo

The agent learns, remembers, schedules, and extends itself in plain English.

> remember that I'm allergic to peanuts
> what should I avoid at the Thai restaurant?

Angel Claw retrieves stored memory and responds intelligently.

> remind me to check the oven in 20 minutes

It schedules the task and executes it proactively.

> create skill to shut down pc
> shut down pc

It writes and installs a new capability.

> search clawhub for frontend skills
> use x skill

---

⭐ Why Angel Claw?

Designed for Developers Who Want More Than Chat

Feature	What It Means
🧠 Agent-Native Memory	Persistent long-term memory per session using angel-recall.
⚡ Multi-Model Support	Connect to OpenAI, Anthropic, Ollama & more via litellm.
🔁 Proactive Tasks	Built-in cron, every, and at scheduling engine.
🌐 Web Search & Automation	Browse, extract, and summarize websites.
🛠 Self-Generating Skills	Agent writes tools for itself dynamically.
📱 Telegram & WhatsApp	Control your AI from your phone.

---

🏁 Get Started in 30 Seconds

🔧 Prerequisites

• Python 3.10+
• An LLM API key (OpenAI, Anthropic, Ollama, etc.)

---

📦 Installation

pip install angel-claw

---

⚙️ Configure Environment

On running angel-claw chat, a .env is automatically created

Add your keys:

MODEL_KEY=your_llm_api_key
TELEGRAM_TOKEN=optional
WHATSAPP_ENABLED=False

---

▶ Run the Gateway

angel-claw

You're live.

---

💬 Minimal Example (HTTP API)

Send a message to your agent:

curl -X POST http://localhost:8000/chat \
     -H "Content-Type: application/json" \
     -d '{"session_id": "user-123", "message": "Hi, I am Alex. Remember that I like Python.", "user_id": "alex"}'

The agent:

• Stores the fact
• Associates it with the session
• Uses it in future conversations

---

🧠 Core Capabilities

Long-Term Memory

Remember that I'm allergic to peanuts.

Later:

What should I avoid at the Thai restaurant?

✔ Retrieves stored memory
✔ Applies contextual reasoning

---

Proactive Scheduling

Remind me to check the oven in 20 minutes.

Every day at 9 AM check the weather in London.

Schedule task 'daily-report' with cron: 0 18 * * *

Angel Claw executes autonomously.

---

Web Search & Automation

Search for the latest news about SpaceX and summarize it.

Go to https://news.ycombinator.com and tell me the top story.

---

Custom Skill Generation

Create a skill called currency_converter using an API.

The agent:

• Writes tool code
• Registers it
• Immediately uses it

---

ClawHub Community Skills

Search ClawHub for frontend skills.
Install the 'slopwork-marketplace' skill.

⚠ Security Warning
ClawHub skills are community-contributed and unvetted. Always review skill code before use.

---

🔌 Model Context Protocol (MCP)

Angel Claw acts as a robust MCP Host, allowing you to consume external tools from any MCP-compliant server (local or remote).

Configuration

Add your servers to .env:

# Example: Local Node server and Remote Zapier server
MCP_SERVERS='{"everything": {"command": "npx", "args": ["-y", "@modelcontextprotocol/server-everything"]}, "zapier": {"url": "https://mcp.zapier.com/api/v1/connect"}}'

# Example: Bearer token for Zapier
MCP_AUTH='{"zapier": {"token": "your_zapier_key"}}'

Management Commands

# List all discovered MCP tools
angel-claw mcp list

# Test connections to MCP servers
angel-claw mcp test

Features

• Standardized Interoperability: Use any tool from the growing MCP ecosystem.
• Resilient Process Management: Automatic restarts (max 3) with exponential backoff.
• Concurrency Control: Per-server semaphores to prevent saturation.
• Security Guardrails: 1MB output size capping and 30s timeouts.

---

📱 Multi-Channel Bridges

---

Telegram Bridge

1. Create bot via @BotFather
2. Add TELEGRAM_TOKEN to .env
3. Start Angel Claw
4. Pair session:

/pair cli-default

Now your AI lives in Telegram.

---

WhatsApp Bridge

Enable:

WHATSAPP_ENABLED=True

Login:

angel-claw login-whatsapp

Scan QR → Pair session:

/pair cli-default

Now Angel Claw is on WhatsApp.

---

🧪 Testing

pytest

---

🤝 Contributing

We welcome contributions.

1. Fork the repository
2. Create feature branch
3. Submit PR

See CONTRIBUTING.md for full guidelines.

---

💬 Community & Support

• 🐛 Issues: GitHub Issues
• 💡 Feature Requests: Open a Discussion
• 🌍 Skills: Explore ClawHub
• ⭐ If you find this useful, give it a star!

---

📜 License

Apache 2.0 License.

---

🪽 Angel Claw

Not just a chatbot. A persistent, evolving AI agent.