Skip to content
IntelliRag

Introduction

IntelliRag is a code intelligence platform that makes large legacy codebases intelligible to developers and AI coding assistants through RAG (Retrieval-Augmented Generation).

What it does

Section titled “What it does”

IntelliRag indexes your codebase using tree-sitter AST parsing, builds a knowledge graph of symbols, call relationships, and data flow, then exposes that knowledge through 20 MCP tools that AI coding assistants call automatically during development sessions.

For developers

Section titled “For developers”
  • Ask your AI assistant questions about any part of the codebase and get accurate, grounded answers
  • Trace call graphs, find entry points, and understand impact before making changes
  • Discover technical debt, dead code, and complexity hotspots

For teams

Section titled “For teams”
  • Onboard new developers faster with AI-powered codebase understanding
  • Reduce the blast radius of changes by understanding dependencies
  • Track and triage technical debt across repositories

Architecture overview

Section titled “Architecture overview”
┌─────────────┐     ┌──────────────┐     ┌──────────────┐
│   Indexer    │────>│  API Server  │<────│  MCP Server  │
│ (your infra) │     │  (hosted)    │     │ (AI assistant)│
└─────────────┘     └──────────────┘     └──────────────┘
                           │
              ┌────────────┼────────────┐
              │            │            │
         ┌────▼───┐  ┌─────▼────┐  ┌───▼────┐
         │Postgres│  │  Neo4j   │  │ Qdrant │
         │metadata│  │  graphs  │  │vectors │
         └────────┘  └──────────┘  └────────┘
  • Indexer - Runs on your infrastructure. Parses code with tree-sitter, detects frameworks, extracts symbols, and sends results to the API server.
  • API Server - Hosted platform. Stores indexed data, serves queries, manages tenants and billing.
  • MCP Server - Runs alongside your AI assistant. Calls the API to answer questions about your code.

Key concepts

Section titled “Key concepts”
  • Workspace - A logical grouping of repositories (e.g. a team or product)
  • Repository - A git repository connected to IntelliRag
  • Symbol - A function, class, method, or other named code element
  • FQN - Fully qualified name (e.g. com.example.UserService.findById)

Next steps

Section titled “Next steps”
QuickstartGet up and running in 5 minutes
MCP ToolsExplore the 20 tools available to AI assistants