Ability to run completely without subscriptions

[Bandit253]

Description

I would like to be able to have the whole system thing to be run locally, with no reliance on external APIs, and use self-hosted LM Studio and Ollama models.

I am happy to accept it may already be possible, if so, a simple guide would be fantastic.

I love the goal of the project, Rust is an excellent choice.

Alternatives Considered

I have tried removing all references to subscription services, yet I still get "Streaming setup failed: Boot failed: Agent LLM driver init failed: Missing API key: Set GROQ_API_KEY environment variable for provider 'groq'" messages. I have never had nor referenced Groq, in this project or ever.

Additional Context

No response

[technerd-official]

I would love to have a preset where all the online stuff is hidden. I tried to run it with LM Studio as the server with Qwen3.5-9b but seems like something is still broken, OpenFang runs for a couple of minutes fine than out of a sudden OpenFang has connection errors and I have to delete sessions and restart the agents. Sometimes also reloading the model helps but this is frustrating. A pity because I really love the simplicity and the dashboard of OpenFang. But Im sure most of the things will get fixed.

[jaberjaber23]

OpenFang already supports fully local operation with Ollama, LM Studio, vLLM, and any OpenAI-compatible local server. Set your config.toml like this:

[default_model]
provider = "ollama"
model = "llama3"
base_url = "http://localhost:11434/v1"
api_key_env = ""

The empty api_key_env means no API key needed. We also fixed a bug where empty api_key_env was incorrectly handled, and the dashboard now works from localhost without requiring an api_key in config.toml. No subscriptions or external APIs required.