Summary
Ask several models, then make one of them decide.
Fanning a prompt out is the easy half. The board-and-decision tool is the reason to reach for it: several models answer, one is nominated to weigh them, and the reasoning is saved with the answers.
What it is
A single interface to several LLM providers. Models are named with a provider prefix, so one call fans out across as many as you list; the responses come back together, can be written out as markdown files, and can be arbitrated by a model you designate.
What you get
- One prompt sent to several models at once, each named by its provider prefix
- The prompt read from a file, for anything longer than a chat message
- Responses written out as markdown files, one per model, into a directory you name
- A board of models answering and a designated model weighing them, with the decision saved alongside the responses
- The available providers listed, and the models available within one of them
- OpenAI, Anthropic, Google Gemini, Groq, DeepSeek and Ollama behind the same call
Requirements
An API key for each provider you intend to call, in the server's environment — Ollama runs locally and needs none. The file-based tools take absolute paths only.
Setup effort
Build from source — clone the repository and build it, then point your client at the binary