compose-anything/apps/nanobot

Nanobot

中文说明 | English

Nanobot is a lightweight, production-ready personal AI assistant with multi-channel support (Telegram, Discord, WhatsApp, Feishu), local model integration, and powerful tool capabilities.

Features

• 🤖 Multi-Provider LLM Support: OpenRouter, Anthropic, OpenAI, DeepSeek, Groq, Gemini, and more
• 🖥️ Local Models: Run your own models with vLLM or any OpenAI-compatible server
• 💬 Multi-Channel: Telegram, Discord, WhatsApp, and Feishu (飞书) integration
• 🛠️ Powerful Tools: File operations, shell execution, web search, and custom skills
• 📅 Scheduled Tasks: Cron-like job scheduling with natural language
• 🎯 Memory & Skills: Persistent memory and extensible skill system
• 🔒 Security: Sandbox mode, access control, and safe command execution

Quick Start

Prerequisites

• Docker and Docker Compose installed
• At least one LLM provider API key (recommended: OpenRouter)

Setup

1. Copy the example environment file:

   cp .env.example .env
   

2. Edit .env and configure at least one LLM provider:

   # For OpenRouter (recommended for global access)
   OPENROUTER_API_KEY=sk-or-v1-xxxxx
   
   # Or use any other provider
   ANTHROPIC_API_KEY=sk-ant-xxxxx
   OPENAI_API_KEY=sk-xxxxx
   

3. Start the service:

   docker compose up -d
   

4. Initialize configuration (first time only):

   docker compose exec nanobot nanobot onboard
   

5. Check status:

   docker compose exec nanobot nanobot status
   

Usage

CLI Mode

Chat directly with nanobot:

docker compose exec nanobot nanobot agent -m "What is 2+2?"

Interactive mode:

docker compose exec nanobot nanobot agent

Gateway Mode (Default)

The default docker compose up starts the gateway server which enables:

• Telegram bot integration
• Discord bot integration
• WhatsApp messaging (requires additional setup)
• Feishu/Lark integration
• HTTP API access (port 18790)

Channel Setup

Telegram

1. Create a bot via @BotFather on Telegram

2. Get your user ID from @userinfobot

3. Configure in .env:

   TELEGRAM_ENABLED=true
   TELEGRAM_TOKEN=your_bot_token
   

4. Restart the service

Discord

1. Create a bot at Discord Developer Portal

2. Enable MESSAGE CONTENT INTENT in bot settings

3. Configure in .env:

   DISCORD_ENABLED=true
   DISCORD_TOKEN=your_bot_token
   

4. Restart the service

WhatsApp

Requires Node.js and additional setup. See official documentation for details.

Feishu (飞书)

1. Create an app at Feishu Open Platform

2. Enable Bot capability and add im:message permission

3. Configure in .env:

   FEISHU_ENABLED=true
   FEISHU_APP_ID=your_app_id
   FEISHU_APP_SECRET=your_app_secret
   

4. Restart the service

Configuration

Environment Variables

See .env.example for all available configuration options.

Key settings:

Variable	Description	Default
NANOBOT_MODEL	LLM model to use	anthropic/claude-opus-4-5
NANOBOT_COMMAND	Command to run (gateway, agent, status)	gateway
RESTRICT_TO_WORKSPACE	Sandbox mode - restrict tools to workspace	false
BRAVE_API_KEY	API key for web search tool	(empty)
TELEGRAM_ENABLED	Enable Telegram channel	false
DISCORD_ENABLED	Enable Discord channel	false

LLM Provider Priority

When multiple providers are configured, nanobot will:

1. Match provider based on model name (e.g., gpt-4 → OpenAI)
2. Fall back to first available API key

Security

For production deployments:

• Set RESTRICT_TO_WORKSPACE=true to sandbox all file and shell operations
• Configure allowFrom lists in the config file for channel access control
• Use dedicated user accounts for channel integrations
• Monitor API usage and set spending limits
• Keep credentials in environment variables, never in code

Scheduled Tasks

Run tasks on a schedule:

# Add a daily reminder
docker compose exec nanobot nanobot cron add \
  --name "morning" \
  --message "Good morning! What's on the agenda?" \
  --cron "0 9 * * *"

# List scheduled jobs
docker compose exec nanobot nanobot cron list

# Remove a job
docker compose exec nanobot nanobot cron remove <job_id>

Local Models (vLLM)

Run nanobot with your own local models:

1. Start a vLLM server:

   vllm serve meta-llama/Llama-3.1-8B-Instruct --port 8000
   

2. Configure in .env:

   VLLM_API_KEY=dummy
   VLLM_API_BASE=http://host.docker.internal:8000/v1
   NANOBOT_MODEL=meta-llama/Llama-3.1-8B-Instruct
   

3. Restart the service

Volumes

• nanobot_config: Configuration files and credentials
• nanobot_workspace: Agent workspace and files

Ports

• 18790: Gateway HTTP API (configurable via NANOBOT_PORT_OVERRIDE)

Resource Limits

Default resource limits:

• CPU: 1.0 cores (limit), 0.5 cores (reservation)
• Memory: 1GB (limit), 512MB (reservation)

Adjust via environment variables: NANOBOT_CPU_LIMIT, NANOBOT_MEMORY_LIMIT, etc.

Troubleshooting

Check logs

docker compose logs -f nanobot

Verify configuration

docker compose exec nanobot nanobot status

Test LLM connection

docker compose exec nanobot nanobot agent -m "Hello!"

Common issues

No API key configured:

• Ensure at least one provider API key is set in .env
• Restart the service after updating environment variables

Channel not responding:

• Check that the channel is enabled in .env
• Verify bot tokens are correct
• Check logs for connection errors

File permission errors:

• Ensure volumes have proper permissions
• Try running with RESTRICT_TO_WORKSPACE=false for debugging

License

Nanobot is an open-source project. See the official repository for license details.

Links

• Official Repository: https://github.com/HKUDS/nanobot
• Documentation: https://github.com/HKUDS/nanobot#readme
• Issues: https://github.com/HKUDS/nanobot/issues