MCP-CoConuT
by MarceloAssis123
MCP-CoConuT is an MCP (Model Context Protocol) server implementing the CoConuT tool, which facilitates structured chain-of-thought reasoning. It includes automatic cycle detection, branch management, and guided interaction.
Last updated: N/A
What is MCP-CoConuT?
MCP-CoConuT is a server that provides the CoConuT tool for structured problem-solving using a continuous chain of thought. It helps users explore different lines of reasoning, detect cyclical patterns, and document conclusions.
How to use MCP-CoConuT?
First, clone the repository and install the dependencies using npm install. Then, you can run the server using npm run dev for development or npm run build and npm start for production. The server accepts JSON requests with parameters for the CoConuT, CoConuT_Analyser, and CoConuT_Storage tools, allowing you to initiate thought chains, analyze them, and save structured conclusions.
Key features of MCP-CoConuT
Continuous Chain of Thought (CoConuT) implementation
Cycle detection using various similarity metrics (Levenshtein, Jaccard, Cosine)
Branch management for exploring different lines of thought
Automatic reflection for evaluating progress
Automated analysis of the thought chain
Structured conclusion logging
Integrated persistence
Multiple response formats (JSON, Markdown, HTML)
Modular architecture with dependency injection
Integrated documentation
Internationalization support
Flexible templates for customizing conclusions
Use cases of MCP-CoConuT
Structured problem-solving
AI-assisted reasoning
Knowledge management
Documentation of changes and conclusions
FAQ from MCP-CoConuT
What is the purpose of the CoConuT tool?
What is the purpose of the CoConuT tool?
The CoConuT tool is designed to facilitate structured chain-of-thought reasoning, helping users solve complex problems by breaking them down into smaller, interconnected thoughts.
How does cycle detection work?
How does cycle detection work?
The system uses advanced algorithms to detect cyclical reasoning patterns by comparing thoughts using different similarity metrics like Levenshtein, Jaccard, and Cosine.
Where are the data files saved?
Where are the data files saved?
Data files are saved in a folder named coconut-data within the path provided by the model through the projectPath parameter.
What input parameters does the CoConuT tool accept?
What input parameters does the CoConuT tool accept?
The CoConuT tool accepts parameters like thought, thoughtNumber, totalThoughts, nextThoughtNeeded, isRevision, revisesThought, branchFromThought, branchId, needsMoreThoughts, score, inputType, problemStatus, options, numberArray, and Call_CoConuT_Analyser.
How can I contribute to the project?
How can I contribute to the project?
You can contribute by forking the project, creating a feature branch, implementing your feature with tests, committing your changes, pushing to the branch, and opening a pull request.