mcp-rocq
RoCQ (Coq Reasoning Server)
claude mcp add --transport stdio angrysky56-mcp-rocq uv --directory F:/GithubRepos/mcp-rocq run mcp_rocq --coq-path F:/Coq-Platform~8.19~2024.10/bin/coqtop.exe --lib-path F:/Coq-Platform~8.19~2024.10/lib/coq \ --env COQ_PATH="path to coqtop binary" \ --env COQ_LIB_PATH="path to coq library directory"
How to use
MCP-RoCQ is a Model Context Protocol server that provides advanced reasoning capabilities by integrating with the Coq proof assistant. It exposes tooling for dependent type checking, inductive type definitions, and property proving, enabling structured and automated interaction with Coq through an MCP-compatible interface. The server communicates via a structured protocol (XML-based in this project as per description) and surfaces detailed feedback for type errors and proof attempts, helping you reason about terms, types, and properties within Coq contexts.
To use the server, configure the MCP host (as shown in the configuration example) and submit requests for type checking, inductive type definitions, or property proofs. Type checking requests send a term and an expected type to verify compatibility within a given context. Inductive type requests define constructors and verify their well-formedness within Coq. Property proving requests supply a statement and optional tactics or automation cues to drive a proof attempt. The tooling is designed to provide structured responses with success/failure signals, proof obligations, and diagnostic messages to guide refinement of terms or proof scripts.
How to install
Prerequisites:
- Coq Platform 8.19 (2024.10) installed on your system
- Python and uv (uvx) available for the environment you will use
- Git for cloning the repository
Step-by-step installation:
-
Install Coq Platform 8.19 (2024.10)
- Follow the Coq Platform installation instructions: https://github.com/coq/platform
-
Clone the MCP-RoCQ repository
- git clone https://github.com/angrysky56/mcp-rocq.git
- cd mcp-rocq
-
Set up the Python/uv environment (uvx)
- On Windows with Python: uv venv ./venv/Scripts/activate uv pip install -e .
- On Unix-like systems, adapt paths accordingly: uv venv source venv/bin/activate uv pip install -e .
-
Validate the installation
- Ensure the mcp_rocq entry point is discoverable by uv and that the Coq paths in the configuration point to your Coq Platform installation.
-
Run the MCP server
- Use the provided mcp_config example as a basis to start the server via your MCP host manager (Claude App, mcphost, etc.).
Additional notes
Notes and tips:
- Ensure the Coq Platform paths (coqtop.exe and lib/coq) match your actual installation directory. Adjust the --coq-path and --lib-path values accordingly.
- If you encounter syntax or protocol issues, verify that the MCP host client uses the expected XML/JSON messaging structure the RoCQ server expects.
- When updating Coq, some tactics or Verward proofs may require updates to the coercions or library imports; keep your Coq libraries in sync with your proof code.
- If you observe import or module resolution errors, double-check the Coq library path provided to the server and ensure Coq can locate the standard library and any required user libraries.
- For debugging, enable verbose logging in the MCP host and capture responses from RoCQ to pinpoint where syntax or type mismatches occur.
Related MCP Servers
mcp-pinecone
Model Context Protocol server to allow for reading and writing from Pinecone. Rudimentary RAG
mcp-python-interpreter
MCP Python Interpreter: run python code. Python-mcp-server, mcp-python-server, Code Executor
pfsense
pfSense MCP Server enables security administrators to manage their pfSense firewalls using natural language through AI assistants like Claude Desktop. Simply ask "Show me blocked IPs" or "Run a PCI compliance check" instead of navigating complex interfaces. Supports REST/XML-RPC/SSH connections, and includes built-in complian
mcp-images
## MCP-Images Looking for a powerful image processing server? MCP Server-Image provides enterprise-grade image handling with just a few lines of code. Perfect for AI applications, web services, and data processing pipelines. [Get Started](#installation) | [Support Us](https://www.buymeacoffee.com/blazzmocompany)
google-search-console
It connects directly to your Google Search Console account via the official API, letting you access key data right from AI tools like Claude Desktop or OpenAI Agents SDK and others .
coder_db
An intelligent code memory system that leverages vector embeddings, structured databases, and knowledge graphs to store, retrieve, and analyze code patterns with semantic search capabilities, quality metrics, and relationship modeling. Designed to enhance programming workflows through contextual recall of best practices, algorithms, and solutions.