Nuanced

A Model Context Protocol (MCP) server that provides call graph analysis capabilities to LLMs through the nuanced library

python

17

GitHub Stars

python

Language

6 months ago

First Indexed

2 months ago

Catalog Refreshed

Documentation & install

Readme and setup notes from the catalogue, plus a client-ready config you can copy for your MCP host.

Installation

Add the following to your MCP client configuration file.

Configuration

View docs

{
  "mcpServers": {
    "mattmorgis-nuanced-mcp": {
      "command": "uv",
      "args": [
        "--directory",
        "/path/to/nuanced-mcp",
        "run",
        "nuanced_mcp_server.py"
      ]
    }
  }
}

Nuanced MCP Server provides call graph analysis capabilities to help LLMs understand code structure. It exposes tools to initialize, explore, and analyze function call graphs across Python repositories, enabling more contextually aware code assistance.

How to use

You interact with the Nuanced MCP Server through the supported MCP client tools. Start by loading a repository’s call graph, then inspect specific functions to see their callers, callees, and dependencies. You can switch between repositories, list what you have loaded, and generate prompts that incorporate the call graph for deeper analysis. Common workflows include initializing a graph for a repo, retrieving a function’s detailed call relationships, and analyzing how changes to a function would ripple through the codebase.

How to install

Prerequisites: you need Python installed to run the Nuanced MCP Server locally, plus a runtime environment capable of executing the server entry point via your chosen MCP runner.

Additional content

The server exposes a set of tools and resources that let you initialize and inspect call graphs, explore function details, and analyze dependencies or change impact. Key capabilities include initializing graphs for repositories, listing repositories, switching active repositories, and retrieving detailed information about specific functions and their relationships. You can also generate prompts specifically tailored to analyze a function or understand dependency and impact contexts. Use the provided resources to obtain summaries of loaded graphs or per-repo graphs, and consult function-level details for targeted AI-assisted code understanding.

Available tools

initialize_graph

Initialize a code graph for the given repository path. You provide a repo_path string and the server loads the function call graph for that repository.

switch_repository

Switch to a different initialized repository by providing a repo_path string. The active repository updates for subsequent queries.

list_repositories

List all initialized repositories with their statuses. No input required.

get_function_call_graph

Return the call graph for a specific function given a file_path, function_name, and optional repo_path. If repo_path is not provided, the active repository is used.

analyze_dependencies

Identify all module or file dependencies in the codebase. You can provide file_path or module_name to scope the analysis.

analyze_change_impact

Analyze the impact of changing a specific function by providing its file_path and function_name. The analysis outlines potential ripple effects across the graph.

graph://summary

Get a summary of the currently loaded code graph.

graph://repo/{repo_path}/summary

Get a summary of a specific repository's code graph by providing repo_path.

graph://function/{file_path}/{function_name}

Get detailed information about a specific function, including its callers, callees, and related metadata.

analyze_function

Create a prompt to analyze a function with its call graph by specifying file_path and function_name.

impact_analysis

Create a prompt to analyze the impact of changing a function by specifying file_path and function_name.

analyze_dependencies_prompt

Create a prompt to analyze dependencies of a file or module by specifying file_path and/or module_name.