CodeGraphy v2.1: 11 MCP tools, interactive architecture graph, -85% LLM tokens

CodeGraphy v2.1 introduces a comprehensive solution for modern AI-assisted development challenges. The tool transforms complex codebases into interactive, visual living graphs while significantly reducing token waste and preventing model hallucinations. Developers can now navigate their architecture more efficiently and feed clean, precise context directly to AI models.

AST Context Extraction

Instead of sending entire file bodies to AI models, CodeGraphy's built-in AST parser strips internal implementation blocks while preserving function and class definitions, parameter names, type annotations, and docstrings. This approach reduces token consumption by up to 85%, allowing a 20-file module to be processed using only ~1,500 tokens instead of 25,000.

Living Architecture Canvas

The platform features an Obsidian-style interactive canvas built with React Flow, supporting smooth physics, drag-and-drop organization, and status markers. Developers can pair multiple independent or interconnected modules simultaneously, with the chat compiling unified context across all attached files and dependency edges in real time.

Industrial Architecture Analysis

CodeGraphy implements Tarjan O(V+E) cycle detection for instant drift validation and coupling metrics calculation. It flags problematic modules and exports architecture reports in multiple formats including Mermaid, JSON, and HTML, helping maintain clean architectural boundaries.

Integrated AI Ecosystem

The tool supports over 340 AI models including Google Gemini, OpenAI's GPT-4o, Anthropic's Claude 3.5, and local models via Ollama and LM Studio. It features sub-second streaming through Server-Sent Events with intelligent fallback mechanisms and persistent local chat history stored in the workspace.

Security & IP Protection

The application employs multi-layer security including Python-to-native compilation via Nuitka and Zig, cryptographic frontend obfuscation using RC4, and ASAR integrity checks. These measures protect against reverse engineering and decompilation, ensuring intellectual property protection for developers using the tool.