Skip to content
MCP server

Tokonomix Council

Ask several LLMs at once and get one judged answer through MCP

Runs locallyStatus and history
Install
Pricing
Paid
Platforms
API, CLI
Runs
Locally
Version
1.14.0
Listing updated
1 day ago

About Tokonomix Council

Tokonomix Council is a paid MCP server for ai models and llm tooling. It runs locally on your machine and is started by your AI client.

Instead of trusting a single model, this server sends a prompt to between two and six frontier models that answer blind to each other. A separate judge from a different model family then reconciles the answers into one verdict, and the response also shows what each model said and what the round cost.

The README is open about what it is not: it is a decision tool, not a chatbot or a voting system, and it does not pick a winner by majority. It stresses grounding, meaning checking a claim against the real artifact, as the strongest lever for decision quality. Some features, such as large shared context packs, are gated by account, and reranking and a web-search plugin are marked as planned.

It runs locally over stdio from npm, and the operator also publishes a hosted connector at mcp.tokonomix.ai. Billing is prepaid and metered by tokens plus a small consensus margin, and keyless onboarding is available for a first run.

Install

Run in a terminal

claude mcp add council --env TOKONOMIX_API_KEY='<TOKONOMIX_API_KEY>' -- npx -y tokonomix-council-mcp

Add to claude_desktop_config.json

{
  "mcpServers": {
    "council": {
      "command": "npx",
      "args": [
        "-y",
        "tokonomix-council-mcp"
      ],
      "env": {
        "TOKONOMIX_API_KEY": "<TOKONOMIX_API_KEY>"
      }
    }
  }
}

Add to ~/.cursor/mcp.json

{
  "mcpServers": {
    "council": {
      "command": "npx",
      "args": [
        "-y",
        "tokonomix-council-mcp"
      ],
      "env": {
        "TOKONOMIX_API_KEY": "<TOKONOMIX_API_KEY>"
      }
    }
  }
}

Add to .vscode/mcp.json

{
  "servers": {
    "council": {
      "type": "stdio",
      "command": "npx",
      "args": [
        "-y",
        "tokonomix-council-mcp"
      ],
      "env": {
        "TOKONOMIX_API_KEY": "<TOKONOMIX_API_KEY>"
      }
    }
  }
}

Add to ~/.codeium/windsurf/mcp_config.json

{
  "mcpServers": {
    "council": {
      "command": "npx",
      "args": [
        "-y",
        "tokonomix-council-mcp"
      ],
      "env": {
        "TOKONOMIX_API_KEY": "<TOKONOMIX_API_KEY>"
      }
    }
  }
}

Add to your client's MCP settings (most use this shape)

{
  "mcpServers": {
    "council": {
      "command": "npx",
      "args": [
        "-y",
        "tokonomix-council-mcp"
      ],
      "env": {
        "TOKONOMIX_API_KEY": "<TOKONOMIX_API_KEY>"
      }
    }
  }
}

Configuration

  • TOKONOMIX_API_KEYTokonomix API key (tok_live_...). Optional - keyless onboarding works without it.

Tools this server provides (11)

Taken from the project documentation; the running server may differ.

tokonomix_consensus_ask
Ask several models and get one judged answer (consensus, diff, best_of, raw or full)
tokonomix_single_ask
Single-model passthrough
tokonomix_list_models
List available models
tokonomix_get_balance
Show account balance
tokonomix_skill_version
Report the skill version
tokonomix_get_skill
Fetch the usage skill
tokonomix_onboard
Keyless first-run onboarding
tokonomix_onboard_verify
Verify onboarding
tokonomix_rate_consensus
Rate a consensus result
tokonomix_relay_human_feedback
Relay a human verdict
tokonomix_upload
Upload a large shared context pack

What it does

Multi-model consensus
Fans one prompt out to 2-6 models and has an independent judge produce a single answer.
Several answer modes
The ask tool supports consensus, diff, best_of, raw and full modes.
Single-model passthrough
Calls one model directly when a council is not needed.
Vision input
Images can be attached to an ask.
Itemised cost
Each model's answer, tokens and cost are reported per round.
Keyless onboarding
Onboarding tools let a first run work without a pre-issued API key.

Good for

  • Reviewing a high-stakes technical decision with several models
  • Surfacing disagreements between models that one answer would hide
  • Comparing how different models answer the same prompt and what each costs
  • Grounding a review against code or documents supplied as context

Pricing

Paid

Prepaid pay-as-you-go: tokens plus a small consensus margin; no subscription or minimum

Prices change. Check tokonomix.ai before you buy.

Tokonomix Council FAQ

Does it need an API key?

The registry entry marks TOKONOMIX_API_KEY as optional because keyless onboarding works without it.

How is it billed?

Prepaid balance, no subscription: you pay the tokens spent plus a small margin on multi-model consensus checks.

Can I use my own provider keys?

Yes. With BYOK the provider bills you directly and you pay only the per-package consensus fee; single-model BYOK calls are free.

Where do the models run for personal data?

The pricing page says an EU-only council can be pinned so the request stays in-region.

Is Tokonomix Council free?

No. Tokonomix Council is a paid product.

How do I add Tokonomix Council to Claude Code?

Run this command in your terminal: claude mcp add council --env TOKONOMIX_API_KEY='<TOKONOMIX_API_KEY>' -- npx -y tokonomix-council-mcp

What are the best alternatives to Tokonomix Council?

The closest alternatives to Tokonomix Council are Claude Code Ultimate Guide, signafAIde, Toolkien, Lugh and HypePaper.

Alternatives to Tokonomix Council

  1. MCP server for searching the Claude Code Ultimate Guide from any MCP client

    Runs locally
    • Open source
    • 6.5K GitHub stars
    • AI models and LLM tooling
  2. signafAIde

    MCP server

    Local MCP server to set up, verify and check coverage of signed agent action records

    Runs locally
    • 5 tools
    • Freemium
    • Security
  3. Toolkien

    MCP server

    One MCP gateway for all your MCP servers, with code mode, permissions and audit

    Not checked yet
    • 3 tools
    • Remote
    • Open source
    • AI models and LLM tooling
  4. Lugh

    MCP server

    Write, check and dry-run Lugh agent programs with policy, budgets and approvals

    Runs locally
    • 5 tools
    • Open source
    • AI models and LLM tooling
  5. HypePaper

    MCP server

    Research papers, benchmarks, citations and topic experts for AI assistants

    • 58 tools
    • Remote
    • Freemium
    • AI models and LLM tooling