Openclaw Huggingface

🦞 OpenClaw: Autonomous AI Agent Space

[](LICENSE)

🚀 Free 24/7 Cloud Version

[!CAUTION] This file is named `README.md` to avoid conflicts with the standard `README.md` required by Hugging Face do not commit this file to hugging face.

**DO NOT COMMIT SECRETS DIRECTLY TO GITHUB OR HUGGING FACE.** Always use the **Settings -> Variables & secrets** dashboard to manage your API keys, tokens, and passwords. Committing secrets to your repository will expose them to the public or Hugging Face logs.

OpenClaw is a premium, autonomous AI agent environment designed to run for **free, 24/7 in the cloud** via Hugging Face Spaces. It leverages the power of multimodal LLMs to browse the web, control interfaces, and maintain a persistent memory across sessions.

🚀 Quick Start (Hugging Face Deployment)

1. Duplicate this Space (or push this code to a new Docker Space).
2. Configure Secrets: Go to Settings -> Variables & secrets and add the following:

🔑 Required Secrets

Secret	Description
MODEL_KEY_1	Primary Google Gemini API Key
MODEL_KEY_2	Second Gemini Key (for rotation/quota)
MODEL_KEY_3	Third Gemini Key (for rotation/quota)
MODEL_KEY_4	Fourth Gemini Key (for rotation/quota)
TELEGRAM_BOT_TOKEN	Token from @BotFather for your bot
TELEGRAM_ALLOWED_USER_ID	Your numeric Telegram ID for private access
HF_GITHUB_TOKEN	GitHub Personal Access Token (for memory sync)

📁 Required Variables

Variable	Value
GIT_MEMORY_REPO	https://github.com//.git
MODEL_NAME	Name of the primary model (e.g., google/gemini-3.1-flash-lite-preview)
MODEL_FALLBACK	Backup model name (default: openrouter/google/gemma-4-31b-it)
MODEL_PROVIDER	Provider of the model (default: google)

Example Model Configuration

# Set your primary engine
MODEL_