Skip to main content
Glama
Dakraid

mcp-consensus

by Dakraid
OverviewSchemaRelated ServersScoreDiscussions
JavaScript
Remote

MCP Consensus Server

A Model Context Protocol (MCP) server that implements an advisor-based consensus mechanism for collaborative problem-solving using multiple AI models.

Features

  • Multi-Advisor Consensus System: Engages 5 specialized AI advisors in structured discussion

  • Collaborative Problem-Solving: Advisors work together to find optimal solutions through debate and analysis

  • Tool Integration: Advisors can request additional information through available tools

  • Configurable Discussion Parameters: Adjustable rounds and consensus thresholds

  • Real-time Discussion Logging: Colorful console output showing the consensus process

  • Multiple AI Models: Utilizes different models (OpenAI, Anthropic, DeepSeek, Moonshot, Z-AI) for diverse perspectives

Related MCP server: Sequential Thinking Multi-Agent System

Installation

npm install @dakraid/mcp-consensus

Prerequisites

  • Node.js 18+

  • OpenRouter API key (set as OPENROUTER_API_KEY environment variable)

Usage

MCP Server Configuration

Add to your MCP client configuration:

{
  "mcpServers": {
    "consensus": {
      "command": "npx",
      "args": [
        "-y",
        "@dakraid/mcp-consensus"
      ],
      "env": {
        "OPENROUTER_API_KEY": "",
        "CONSENSUS_MAX_ROUNDS": "5",
        "CONSENSUS_THRESHOLD": "0.8"
      }
    }
  }
}

Available Tools

consensus

A multi-advisor consensus system that facilitates structured discussion and debate among general-purpose AI advisors to reach optimal solutions.

Parameters:

  • problem (required): Detailed description of the problem to solve

  • availableTools (required): Array of tool names available for research

Example Usage:

// Basic consensus
{
  "problem": "Should we adopt a remote-first work policy for our tech company?",
  "availableTools": ["web_search", "read_file"]
}

How It Works

  1. Problem Presentation: The problem is presented to all 5 advisors simultaneously

  2. Initial Analysis: Each advisor provides their analysis and proposed solution

  3. Tool Requests: Advisors can request additional information through available tools

  4. Multi-Round Discussion: Advisors engage in structured debate, considering each other's perspectives

  5. Consensus Detection: The system monitors for agreement based on the configured threshold

  6. Result Delivery: Returns the final consensus with complete discussion history

Advisors

The system includes 5 pre-configured advisors, each using different AI models:

  • Advisor Alpha: Moonshot AI Kimi-k2

  • Advisor Beta: DeepSeek Chat v3

  • Advisor Gamma: Z-AI GLM-4.5

  • Advisor Delta: OpenAI GPT-4.1

  • Advisor Epsilon: Anthropic Claude Sonnet 4

Each advisor follows core principles of objectivity, collaboration, thoroughness, adaptability, and clarity.

Tool Request Format

Advisors can request additional information using this format:

TOOL_REQUEST: {"tool": "web_search", "parameters": {"query": "remote work productivity statistics"}, "reason": "I need current data on remote work effectiveness"}

Response Structure

The consensus tool returns:

{
  "status": "consensus_reached" | "max_rounds_reached" | "tool_requests_needed",
  "finalConsensus": "The agreed-upon solution",
  "totalRounds": 3,
  "discussionHistory": [...]
}

Development

# Clone the repository
git clone https://github.com/dakraid/mcp-consensus.git
cd mcp-consensus

# Install dependencies
npm install

# Build the project
npm run build

# Watch for changes
npm run watch

Configuration

Environment Variables

  • OPENROUTER_API_KEY: Required API key for OpenRouter

  • CONSENSUS_MAX_ROUNDS: Maximum number of discussion rounds (default: 5, range: 1-10)

  • CONSENSUS_THRESHOLD: Agreement threshold for consensus detection (default: 0.8, range: 0.0-1.0)

  • DISABLE_CONSENSUS_LOGGING: Set to "true" to disable console logging (default: false)

Customization

You can modify the advisor configurations in index.ts to:

  • Change system prompts

  • Use different AI models

  • Add or remove advisors

  • Adjust model parameters

License

MIT License - see LICENSE file for details.

Contributing

  1. Fork the repository

  2. Create a feature branch

  3. Make your changes

  4. Add tests if applicable

  5. Submit a pull request

Support

For issues and questions, please visit the GitHub Issues page.

Author

Created by @dakraid

Available Tools

1 tool
consensusA

A multi-advisor consensus system that facilitates structured discussion and debate among general-purpose AI advisors to reach optimal solutions.

Important instruction for the requesting model: If your request includes or references source code, provide the full source code without truncation (do not omit sections). If the consensus requests a code artifact, include the complete implementation.

This tool uses a team of 5 general-purpose AI advisors (Alpha, Beta, Gamma, Delta, Epsilon) who are trained to:

  • Analyze any type of problem from multiple relevant perspectives

  • Consider technical, practical, strategic, ethical, and risk dimensions as appropriate

  • Work collaboratively toward consensus through objective analysis

  • Adapt their approach based on the specific problem domain

  • Request additional information when needed to make informed decisions

Each advisor follows core principles of objectivity, collaboration, thoroughness, adaptability, and clarity, making them suitable for any problem type without being bound to specific roles.

The advisors work collaboratively to:

  1. Analyze problems comprehensively from whatever angles are most relevant

  2. Engage in constructive debate and evidence-based discussion

  3. Request additional information through tools when needed

  4. Work toward consensus on the objectively best solution

  5. Provide well-reasoned justification for their conclusions

When to use this tool:

  • Any complex problem that would benefit from multiple perspectives

  • Decisions requiring thorough analysis and debate

  • Problems where objective consensus is valuable

  • Situations where different dimensions need to be considered

  • Scenarios requiring comprehensive evaluation and risk assessment

  • Any decision where you want to validate through multiple AI perspectives

Key features:

  • 5 general-purpose advisors adaptable to any problem domain

  • Multi-round discussion with automatic consensus detection

  • Tool request integration for research and information gathering

  • Configurable discussion limits and consensus thresholds

  • Detailed discussion history and reasoning documentation

  • Advisors consider all relevant dimensions without role limitations

Parameters explained:

  • problem: Detailed description of the problem or request to solve

  • availableTools: Array of tool names available to the LLM for research

  • maxRounds: Maximum number of discussion rounds (configured via CONSENSUS_MAX_ROUNDS env var)

  • consensusThreshold: Consensus agreement threshold (configured via CONSENSUS_THRESHOLD env var)

Tool request format for advisors: TOOL_REQUEST: {"tool": "tool_name", "parameters": {"param": "value"}, "reason": "explanation of why this tool is needed"}

The system will:

  1. Present the problem to all 5 general-purpose advisors simultaneously

  2. Collect initial responses and any tool requests

  3. If tools are requested, pause for LLM execution and continue with results

  4. Facilitate multiple discussion rounds until consensus or max rounds reached

  5. Return final consensus solution with complete discussion history

Example usage scenarios:

  • Business decisions requiring comprehensive analysis

  • Technical problems with multiple possible solutions

  • Strategic planning with various considerations

  • Ethical dilemmas needing balanced evaluation

  • Any complex problem where multiple perspectives would be beneficial

ParametersJSON Schema
NameRequiredDescriptionDefault
problemYesDetailed description of the problem or request to solve
maxRoundsNoMaximum number of discussion rounds (configured via CONSENSUS_MAX_ROUNDS env var)
availableToolsYesArray of tool names available to the LLM for research
consensusThresholdNoAgreement threshold for consensus (configured via CONSENSUS_THRESHOLD env var)

TDQS

A4/5.0
Behavior4/5

Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?

With no annotations provided, the description carries the full burden of behavioral disclosure. It explains the multi-round discussion process, the role of the five advisors, the optional tool-request mechanism, and the return of the final consensus with discussion history. It also includes the important instruction about providing full source code when code artifacts are involved, which is unusual but useful context. Some internal behaviors (e.g., failure modes) are not covered, so it stops short of a 5.

Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.

Conciseness3/5

Is the description appropriately sized, front-loaded, and free of redundancy?

The description is lengthy and contains redundancy: 'Key features' and 'The system will' sections overlap, and 'Example usage scenarios' reiterates the 'When to use' list. Some sentences, like the advisor 'core principles,' are non-actionable and add length without substance. However, the opening sentence is a strong concise summary, and the structure with clear headings helps navigation.

Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.

Completeness4/5

Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?

Given the tool's complexity and the absence of an output schema or annotations, the description provides a fairly complete picture. It covers the input parameters, the step-by-step system process (from initial presentation to final consensus), the tool-request format, and the return of discussion history. It does not specify the exact output schema, but it explicitly states the tool returns 'final consensus solution with complete discussion history,' which is sufficient for an agent to invoke it correctly.

Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.

Parameters3/5

Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?

The schema already covers all four parameters with descriptions, and the description's 'Parameters explained' section mostly repeats that information. It adds minimal value by noting that maxRounds and consensusThreshold are configured via environment variables, but it does not provide default values or further clarify the expected input format beyond the schema. The baseline of 3 is appropriate since schema coverage is 100%.

Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.

Purpose5/5

Does the description clearly state what the tool does and how it differs from similar tools?

The description clearly states the tool's purpose with a specific verb and resource: 'facilitates structured discussion and debate among general-purpose AI advisors to reach optimal solutions.' It also enumerates the five advisors by name, giving a precise scope of what the tool does.

Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.

Usage Guidelines4/5

Does the description explain when to use this tool, when not to, or what alternatives exist?

A dedicated 'When to use this tool' section provides a clear list of appropriate scenarios, such as 'Any complex problem that would benefit from multiple perspectives' and 'Decisions requiring thorough analysis.' However, it does not explicitly state when not to use the tool or name alternatives, likely because no sibling tools exist.

Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.

Install Server
RayrunBuild, deploy, and operate MCP servers without becoming an infrastructure team.ray.runAd

TDQS

A4.1/5.0
Disambiguation5/5

Since there is only one tool, there is no possibility of confusing it with other tools. The tool's purpose is clearly defined as a consensus system, so agents will know exactly what it does.

Naming Consistency5/5

The tool is named 'consensus', which directly reflects its function. With a single tool, there is no naming pattern to break, so consistency is perfect.

Tool Count3/5

The server exposes only one tool, which is below the typical 3-15 range. However, the tool is a comprehensive consensus system that encapsulates the entire workflow, making the count slightly thin but arguably sufficient for its narrow purpose.

Completeness5/5

The tool provides a complete lifecycle for consensus-based problem solving, including multi-round discussion, tool integration, and consensus detection. No obvious gaps exist within the described domain.

Maintenance

ActivityInactive
ResponsivenessSyncing

Resources

Unclaimed servers have limited discoverability.

Looking for Admin?

If you are the server author, to access and configure the admin panel.

Related MCP Connectors

Related MCP Servers

Latest Blog Posts

MCP directory API

We provide all the information about MCP servers via our MCP API.

curl -X GET 'https://glama.ai/api/mcp/v1/servers/Dakraid/mcp-consensus'

If you have feedback or need assistance with the MCP directory API, please join our Discord server