System Requirements
PicoClaw has a very low hardware footprint:
- OS: Linux / macOS / Windows (WSL2)
- Architecture: x86_64 / ARM64 / RISC-V
- Minimum Memory: 64MB (512MB+ recommended)
- Dependencies: LLM API Key (required for AI chat functionality)
Supported platforms: macOS, Linux, Windows (WSL2), Raspberry Pi, Docker
Step 1: Clone the Repository
Open your terminal and run the following command:
git clone https://github.com/sipeed/picoclaw.gitOnce cloned, navigate to the project directory:
cd picoclawStep 2: Initialize Configuration
Run the onboard command to initialize:
picoclaw onboardFollow the prompts to complete the setup:
- LLM API Key: Enter your LLM provider's API Key.
- Model Preferences: Select your preferred AI model.
- Messaging Platform Token (Optional): If you want to enable Telegram, Discord, QQ, or DingTalk integration, enter the corresponding Bot Token.
Configuration files are stored locally; all data remains on your device and is never uploaded to the cloud.
Step 3: Start Chatting
Once configured, you can start chatting with your AI.
Interactive Mode
picoclaw agentLaunches an interactive chat interface where you can send messages and receive replies continuously.
Single Query Mode
picoclaw agent -m "Hello, please introduce yourself"Sends a single message, receives a reply, and exits—ideal for use in scripts.
Start Gateway Service
To use PicoClaw on messaging platforms like Telegram or QQ:
picoclaw gatewayOnce started, you can interact with PicoClaw via your platform's bot.
Multi-Platform Bot Configuration
Telegram Bot
Find @BotFather on Telegram, send /newbot to create a bot and get your Token, enter the Token into your configuration file, and start picoclaw gateway.
Discord Bot
Create an application in the Discord Developer Portal, obtain the Token from the Bot settings, invite the bot to your server, and enter the Token into your configuration.
QQ Bot
Refer to the QQ Open Platform documentation to register a bot, obtain your AppID and Token, and enter them into your configuration file.
DingTalk Bot
Create a robot application on the DingTalk Open Platform, obtain the Webhook URL and secret, and enter them into your configuration file.
Docker Deployment
PicoClaw supports Docker deployment, perfect for servers or headless environments:
docker pull ghcr.io/sipeed/picoclaw:latestdocker run -d --name picoclaw -v /path/to/config:/config -e PICOCLAW_API_KEY=your_key ghcr.io/sipeed/picoclaw:latestCron Job Configuration
PicoClaw has built-in cron support, managed via the configuration file.
One-time Reminder
tasks: - type: reminder time: "1h" message: "Meeting reminder: Team weekly meeting"Periodic Tasks
tasks: - type: cron schedule: "0 8 * * *" message: "Generate daily work summary"FAQ
Does PicoClaw require an internet connection?
PicoClaw itself can run offline, but an internet connection is required to call LLM APIs and use messaging platform integrations.
Which LLM providers are supported?
It supports OpenRouter, Zhipu AI, Anthropic, OpenAI, DeepSeek, Groq, and more, which can be switched freely via the configuration file.
How do I update PicoClaw?
Navigate to the project directory, run git pull to fetch the latest code, and recompile.
Where is the data stored?
All chat history and configuration data are stored on your local file system, giving you full control.
How large is the Docker image?
The PicoClaw Docker image is approximately 15MB, significantly smaller than the 350MB+ of competing products.
Can it run on a Raspberry Pi?
Yes, PicoClaw supports the ARM64 architecture and runs smoothly on Raspberry Pi 4 (4GB+).
How do I configure multiple bots to work simultaneously?
Simply add the tokens and configuration items for each platform into your configuration file and start picoclaw gateway to activate all bots at once.