AI Council MCP Server
Allows querying local LLMs served by Ollama for parallel response generation.
Allows querying OpenAI models (e.g., GPT-4o) as part of parallel consensus generation.
Allows querying Perplexity AI models (e.g., llama-3.1-sonar-large) via custom API endpoint.
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@AI Council MCP Server What are the best arguments for and against universal basic income?"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
AI Council MCP Server
Multi-AI Consensus Tool: Query multiple AI models in parallel, synthesize responses for better accuracy, and reduce AI bias through ensemble decision-making.
AI Council is a powerful MCP (Model Context Protocol) server that harnesses the "wisdom of crowds" by consulting multiple AI models simultaneously. Get more reliable, comprehensive answers by combining insights from OpenAI, Claude, Gemini, and any OpenAI-compatible API.
β¨ What is AI Council?
AI Council transforms how you interact with AI by:
π Parallel Processing: Queries multiple AI models simultaneously (not sequentially)
π― Bias Reduction: Uses anonymous code names to prevent synthesis bias
β‘ Smart Synthesis: One model synthesizes all responses into a comprehensive answer
π§ Universal Compatibility: Works with OpenAI, OpenRouter, and any OpenAI-compatible API
π‘οΈ Robust Error Handling: Graceful degradation when individual models fail
Perfect for: Research questions, complex analysis, creative projects, technical decisions, and any task where multiple AI perspectives add value.
Related MCP server: polydev-ai
π Requirements
Python 3.10+
uv installed (installation guide)
alternatively:
pipx installed (installation guide), update config
"command": "pipx", "args": ["run", "ai-council"]or manual install wiht
pip install ai-council, and update config"command": "ai-council", "args": []
π Quick Start
Get your OpenRouter api key
Cursor IDE Setup
Open Cursor Settings β MCP
Add new MCP server and set your api key:
{
"ai-council": {
"command": "uvx",
"args": ["ai-council"],
"env": {
"OPENROUTER_API_KEY": "..."
}
}
}Claude Desktop Setup
Edit
~/.claude_desktop_config.jsonand set your api key:
{
"mcpServers": {
"ai-council": {
"command": "uvx",
"args": ["ai-council"],
"env": {
"OPENROUTER_API_KEY": "..."
}
}
}
}That's it! Ask any complex question and the AI Council tool will automatically engage multiple models.
By default it will use OpenRouter with Claude Sonnet 4, Gemini 2.5 Pro, and DeepSeek V3.
CLI Arguments
Use command-line arguments for quick setup, add any of these to the args in you mcp config:
Available CLI Arguments:
--openai-api-key: Your OpenAI API key--openrouter-api-key: Your OpenRouter API key--max-models: Maximum models to query (default: 3)--parallel-timeout: Timeout in seconds (default: 60)--log-level: Logging level (DEBUG, INFO, WARNING, ERROR)--config: Path to custom config file
βοΈ Advanced Configuration
For advanced setups, create a config.yaml file and link to it with --config path/to/config.yaml:
# config.yaml
openai_api_key: "your_openai_key_here"
openrouter_api_key: "your_openrouter_key_here"
max_models: 3
parallel_timeout: 90 # in seconds
synthesis_model_selection: "random" # or "first"
models:
# use OpenAI API
- name: "GPT-4o"
provider: "openai"
model_id: "gpt-4o"
enabled: true # optional, defaults to true
# use OpenRouter API
- name: "Claude Sonnet"
provider: "openrouter"
model_id: "anthropic/claude-3.5-sonnet"
code_name: "Bob" # optional, auto assigned otherwise
# or any custom OpenAI compatible API
- name: "Perplexity"
provider: "custom"
model_id: "llama-3.1-sonar-large-128k-online"
base_url: "https://api.perplexity.ai"
api_key: "your_perplexity_key_here"
# Local LLM (Ollama)
- name: "Local Llama"
provider: "custom"
model_id: "llama-3b"
base_url: "http://localhost:11434"
api_key: "key-if-needed"π How It Works
AI Council uses a sophisticated three-phase approach:
1. Parallel Consultation
Simultaneously queries your configured AI models
Maintains the same context and question for each model
Handles failures gracefully (continues with successful responses)
2. Anonymous Analysis
Assigns code names (Alpha, Beta, Gamma, etc.) to each model's response
Prevents synthesis bias toward specific brands or providers
Preserves response quality while removing model identity
3. Smart Synthesis
Randomly selects one model to act as the synthesizer
Synthesizer analyzes all anonymous responses
Produces a comprehensive answer combining the best insights
π€ Acknowledgments
This project was inspired by Cognition Wheel.
AI Council extends these ideas with:
Enhanced configuration flexibility
OpenRouter support for many model options with a single api key
Support for custom API endpoints
Improved error handling and logging
Using Python
Available Tools
1 toolai_councilA
A tool that consults multiple AI models in parallel, then uses one of them to synthesize the results into a single, high-quality answer. Use this for complex questions requiring deep analysis and verification.
Inputs: context (required), question (required).
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelogβ¦
TDQS
Scored across 1 tool
Only one tool exists, so there is no possibility of ambiguity between tools.
With a single tool (ai_council), naming consistency is not an issue; the name follows a clear snake_case pattern.
Having only one tool feels thin for a server named 'AI Council MCP Server', as it limits functionality to a single action without auxiliary or configuration tools.
The tool covers a specific synthesis task, but the server lacks other potentially useful tools like model configuration, result handling, or iterative analysis, leaving notable gaps.
Maintenance
Related MCP Connectors
Multi-LLM AI Research & Analysis β smart routing, consensus analysis, due diligence reports
Multi-LLM council: 25+ frontier models in parallel, consensus scoring, verdict-first code review.
Ask GPT, Claude, Gemini and Grok at once: one consensus answer plus where they agree and split.
Fan out deep research across multiple AI providers, synthesize into one unified report.
Related MCP Servers
- AlicenseAqualityDmaintenanceProvides access to multiple frontier LLM models (GPT, Claude, Gemini, Grok, DeepSeek) for consulting a "conclave" of AI perspectives, enabling peer-ranked evaluations and synthesized consensus answers for important decisions.81MIT
- AlicenseNot gradedqualityCmaintenanceQuery multiple AI models (GPT-4, Claude, Gemini, Grok) in parallel for diverse perspectives. Get different expert viewpoints when stuck or need enhanced reasoning.1MIT
- AlicenseAqualityDmaintenanceEnables querying multiple AI models in parallel (Claude, Gemini, O3) and synthesizing their responses using anonymous analysis to reduce bias, providing a comprehensive answer.129 npm193MIT

polydev-aiofficial
AlicenseNot gradedqualityDmaintenanceQueries multiple AI models simultaneously (GPT 5.2, Claude Opus 4.5, Gemini 3, Grok 4.1) via a single API call to provide diverse expert perspectives and consensus recommendations.MIT