Nuanced MCP Server

Nuanced MCP Server is an MCP server that provides static call graph analysis and dependency mapping for Python codebases using the nuanced library. Designed for software engineers, code reviewers, and AI coding assistants, it connects large language models directly to the structural relationships within a project. By generating and indexing function-level call graphs, the server allows coding agents to trace execution paths, inspect upstream and downstream callers, and evaluate the blast radius of proposed edits. Developers and agents can navigate complex repositories without relying on inaccurate text search or guesswork. The server exposes dedicated tools to initialize repositories, switch between multiple codebases, query specific function call graphs, and list intra-module dependencies. In addition, it provides standardized MCP resources and dynamic prompt templates tailored for impact analysis, enabling context-aware architectural reviews, bug diagnosis, and refactoring within compatible development environments.

Category: Developer Tools & Code Intelligence

Tags: call graph, code analysis, source-code, static analysis

Visit Nuanced MCP Server

How to install and configure Nuanced MCP Server

To run Nuanced MCP Server with Claude Desktop using uv, follow these steps: 1. Ensure you have uv and Python installed on your system. 2. Clone the repository locally or place the project files in a designated path. 3. Open your Claude Desktop configuration file (claude_desktop_config.json). 4. Add the server under the mcpServers configuration key: json { "mcpServers": { "nuanced": { "command": "uv", "args": [ "--directory", "/path/to/nuanced-mcp", "run", "nuanced_mcp_server.py" ] } } } 5. Replace /path/to/nuanced-mcp with the absolute path to your local repository clone, save the file, and restart Claude Desktop.

What you can do with Nuanced MCP Server

• Tracing function execution flow: Generate explicit call graphs for specific Python functions to identify direct callers and downstream dependencies across files. - Evaluating change impact: Run the change impact tool before refactoring a function to discover which dependent modules and entry points will be affected. - Analyzing repository dependencies: Map module-level and file-level import relationships to understand architectural coupling in large or unfamiliar Python codebases. - Switching across multiple projects: Initialize and swap between several active project repositories in a single agent session to inspect shared architecture.

Key facts

• https://github.com/MattMorgis/nuanced-mcp
• Developer Tools & Code Intelligence
• call graph, code analysis, source-code, static analysis

Part of MCP Servers

What is Nuanced MCP Server?

Nuanced MCP Server is a Model Context Protocol service that connects large language models to the nuanced library. It enables AI coding assistants to generate, inspect, and query function call graphs and architectural dependencies across Python code repositories.

How do I install Nuanced MCP Server?

You can configure Nuanced MCP Server in your Claude Desktop configuration file using uv. Add the nuanced server block under mcpServers, specifying the uv command, the working directory path to your nuanced-mcp directory, and the script nuanced_mcp_server.py.

What tools are included in Nuanced MCP Server?

The server includes initialize_graph to scan a repository, switch_repository and list_repositories for multi-project management, get_function_call_graph to inspect callers and callees, analyze_dependencies to list module imports, and analyze_change_impact to measure code modification effects.

What resources does Nuanced MCP Server provide?

Nuanced MCP Server provides URI-based resources including graph://summary for loaded codebase overviews, graph://repo/{repo_path}/summary for specific repository metrics, and graph://function/{file_path}/{function_name} for granular structural information on specific functions.

Which MCP clients work with Nuanced MCP Server?

The server works with any client supporting the Model Context Protocol standard, including Claude Desktop, Cursor, and custom autonomous agents capable of interacting with standard MCP tools, prompts, and resources.