Grafema, developed by Disentinel, is a free MCP server tool designed for graph-based static analysis. It transforms codebases, infrastructure, knowledge, and workflows into a single queryable graph. This graph supports interaction by both humans and AI agents, providing a structured way to understand complex systems. Grafema aims to enhance development practices by offering deep insights into project structures and dependencies. It operates on the mcp platform, focusing on analytical capabilities.

What does Grafema connect to?

Grafema functions as an MCP server and a graph-based static analysis tool. It establishes connections to a codebase, infrastructure, knowledge bases, and existing workflows. The primary goal is to transform these disparate elements into a unified, queryable graph. This graph then serves as a central data model for both human developers and integrated AI agents, enabling them to explore and understand the relationships and structures within a project.

Core features and tools

Grafema exposes several callable tools for AI agent integration. These include find_nodes, find_calls, trace_dataflow, get_file_overview, describe, and query_graph. It supports stdio and Streamable HTTP transport protocols, with Streamable HTTP replacing HTTP+SSE from protocol version 2024-11-05. Compatible AI clients include Claude Code, Cursor, Codex, Aider, and other MCP-compatible agents, ensuring broad integration possibilities.

Who is it for?

Grafema is intended for developers, engineers, and platform engineering teams. It provides a structured approach to understanding complex systems and codebases. Its capabilities are particularly useful for those needing to analyze project ontology, trace data flows, or define architectural guarantees. Skip this if your primary need is not deep structural analysis or AI-assisted codebase interaction.

How to get started

No credentials are required. For local use, a configuration is added to ".mcp.json" or ".cursor/mcp.json" specifying the command "npx grafema-mcp --project .".

Use cases

• Onboarding onto a new codebase by installing the CLI, configuring and indexing the repository, then querying the generated graph to understand project ontology and structure.
• Debugging and tracing, as well as answering complex L4 architecture questions by using structural graph queries.
• Defining guarantees, tracking dependencies, and identifying violations in the codebase before deployment.

Grafema offers a unique graph-based approach to static analysis for MCP servers, supporting both human and AI interaction with codebases. Its ability to unify diverse project elements into a single queryable graph provides significant analytical power for development and platform engineering teams. The tool's local, trust-based model simplifies initial access. Try querying the generated graph to understand a new project's ontology and structure.