Back to Portfolio

Atlas - Code Ingestion Pipeline with MCP

Sophisticated code ingestion pipeline combining semantic embeddings with relationship graphs, plus MCP server for persistent LLM memory. Features graph-aware RAG and intelligent context assembly.

Project Screenshot

Technologies Used

PythonMCPFAISSChromaDBNetworkXTree-sitterRAGFlask

About This Project

Atlas is a production-ready code understanding system with two key components: (1) Modular ingestion pipeline combining semantic vector search (ChromaDB/FAISS) with structural graph analysis (NetworkX) for hybrid retrieval, and (2) Model Context Protocol (MCP) server providing persistent memory for LLM coding assistants like Claude Code. Features Tree-sitter AST parsing, hierarchical summarization, multi-granularity code representation, hallucination detection, and token budget management. Includes web chatbot interface with conversation history. Built to handle enterprise-scale codebases with intelligent, context-aware code comprehension.

Key Features

  • • Feature 1: Add your project features here
  • • Feature 2: Describe the main functionality
  • • Feature 3: Highlight what makes it special

Challenges & Solutions

Describe the challenges you faced during development and how you solved them. This helps showcase your problem-solving skills.