Multi-Model Advisor

Example Use Cases

• 1

  Combine perspectives from multiple local Ollama models to produce a synthesized, multi-view answer.

• 2

  Assign distinct personas or system prompts to each model to encourage diverse lines of reasoning.

• 3

  Seamless integration with Claude for Desktop to surface multi-model advice and tooling such as listing or querying available Ollama models.

Description

This Model Context Protocol (MCP) server sends a single user question to multiple Ollama models (each with configurable system prompts/personas), gathers their responses, and returns them so Claude can synthesize a comprehensive answer. It exposes tools to list available Ollama models and to query configured models in parallel. The project is configurable via environment variables, supports custom system prompts per model, and integrates with Claude for Desktop for an advisory "council of advisors" workflow.