DependencyMCP Server

DependencyMCP is an MCP server that turns a raw codebase into a structured, analyzable map of its internal relationships. By parsing source files across multiple languages—such as TypeScript, JavaScript, C#, and Python—it produces a dependency graph that highlights imports, exports, and architectural layers. This capability addresses the common developer pain point of navigating large or unfamiliar projects: understanding where components live, how they interact, and whether the architecture adheres to defined patterns.

The server’s core value lies in its ability to surface hidden coupling and architectural drift. When a new feature is added or a refactor occurs, developers can quickly run to see how the change ripples through the codebase. The resulting JSON or DOT graph can be visualized with standard tools, revealing modules that are tightly coupled or isolated. The tool then quantifies adherence to architectural rules, awarding points for compliance and deducting them for violations. This scoring mechanism turns subjective code reviews into objective metrics, helping teams maintain clean separation of concerns and preventing technical debt from accumulating unnoticed.

Key capabilities include:

• Multi‑language parsing that supports a wide range of file types without manual configuration.
• Customizable depth and exclusion patterns so analyses can focus on relevant parts of a repository.
• Architectural layer inference, which tags files (e.g., Domain, Infrastructure) based on naming conventions and import patterns.
• Metadata extraction for individual files, providing a quick lookup of imports, exports, and dependencies.
• Caching to speed up repeated analyses, with configurable limits on lines read and cache TTL.

Typical use cases span the software development lifecycle. During onboarding, new engineers can run a full dependency scan to get an instant map of the project’s structure. During code reviews, reviewers can pull the architectural score to verify that changes respect layer boundaries. Continuous integration pipelines may automatically flag violations, enforcing architectural contracts before code merges. Finally, architects can use the graph to identify candidate modules for microservice extraction or to plan a gradual migration toward a more modular architecture.

By integrating seamlessly with any MCP‑compliant AI assistant, DependencyMCP allows conversational agents to answer questions like “Which files import ?” or “Show me the dependency graph for the package.” This tight coupling between AI assistants and static analysis reduces context switching, accelerates troubleshooting, and embeds architectural governance directly into the developer’s workflow.