rushikeshmore/CodeCortex

Overview

CodeCortex pre-builds a persistent knowledge layer for AI agents extracting architecture, dependency, coupling, and risk data from codebases. It serves this knowledge through the Model Context Protocol (MCP) supporting 27 languages and 13 analysis tools.

Best for

Best for
Developers using AI agents to navigate and understand large, multi-language codebases.

Use cases

• Feed an AI agent structured codebase context to answer architecture questions
• Map dependencies and coupling across a multi-language project
• Identify risk areas before making cross-cutting changes

Notes

CodeCortex pre-builds a persistent knowledge layer for AI agents extracting architecture, dependency, coupling, and risk data from codebases. It serves this knowledge through the Model Context Protocol (MCP) supporting 27 languages and 13 analysis tools.

5 stars on GitHub. Last updated 2026-04-21. Licensed MIT.

Use cases

• Feed an AI agent structured codebase context to answer architecture questions
• Map dependencies and coupling across a multi-language project
• Identify risk areas before making cross-cutting changes

Pros

• Reduces redundant re-analysis by caching knowledge persistently
• Broad language support and structured MCP interface for agents
• No runtime dependency on external APIs once built

Cons

• Requires a one-time setup and initial analysis run
• Knowledge can become stale if code changes out of sync with the pre-built layer
• Limited to the 13 predefined tools; custom analyses not supported

Indexed from awesome-mcp-servers-punkpeye and enriched against its public facts.