HermesFace

What HermesFace does

HermesFace packages Hermes Agent as a Space you duplicate. After duplicating, you set repository secrets: HF_TOKEN, an access token with write permission, is required, along with at least one LLM API key such as OPENROUTER_API_KEY, OPENAI_API_KEY, ANTHROPIC_API_KEY, NOUS_API_KEY or GOOGLE_API_KEY. Setting AUTO_CREATE_DATASET to true makes HermesFace create a private backup dataset on first startup.

The README says the Space gets 2 vCPU and 16 GB of RAM from Hugging Face at no cost, and that conversations, skills, memories and config survive restarts through the Dataset sync. It lists Telegram, Discord, Slack, WhatsApp, Signal, WeChat and iMessage among 16 or more channels, 47 built-in tools, and a React-based web dashboard for config, API keys and session monitoring. After duplicating, you should edit the datasets field in the Space's README to point at your own dataset repo or remove it.

Key features

• Duplicate-the-Space deployment of Hermes Agent on Hugging Face Spaces
• State persistence through Hugging Face Dataset sync
• Optional automatic creation of a private backup dataset
• Support for OpenRouter, OpenAI, Claude, Gemini and Nous Portal keys
• Web dashboard for config, API keys and session monitoring

When to use it

• Run a Hermes agent on Telegram or Discord without renting a VPS
• Try Hermes Agent on hosted infrastructure before buying hardware
• Keep a hosted agent's memories and skills across Space restarts

Who it is for: People who want to host Hermes Agent on Hugging Face Spaces instead of their own machine or server.

How it fits with Hermes Agent

Built around Hermes Agent: it wraps the agent in a Docker Space and adds Dataset-based persistence.

Requirements: A Hugging Face account with an HF_TOKEN that has write permission, and at least one LLM API key.

Note: Persistence relies on syncing data to a Hugging Face Dataset, so an HF token with write permission is required.

FAQ