Navegador

Project Knowledge Graph for AI Coding Agents

https://navegador.dev | https://github.com/ConflictHQ/navegador

Overview

Navegador builds and maintains a queryable graph of your software project by combining static code analysis with business knowledge curation.

Two-Layer Architecture

Code Layer (automatically populated): * Repository, File, Module structures * Class, Function, Method definitions * Decorators, Imports, Variables * Call graphs and inheritance hierarchies

Knowledge Layer (manually curated): * Concepts and Domains * Rules and Decisions * People and Wiki pages * PlanOpticon meeting/video knowledge graphs

These layers interact through cross-layer relationships: ANNOTATES, GOVERNS, and IMPLEMENTS connections.

Quick Start

pip install navegador
navegador ingest ./my-repo
navegador explain AuthService

Agent Integration

• CLI: Direct command invocation from scripts
• MCP Protocol: Configure once and all navegador commands become callable tools
• Bootstrap: Automated setup supporting Claude, Gemini, and OpenAI agents

Technical Details

• Supported Languages: Python and TypeScript via tree-sitter
• Storage Backend: FalkorDB (SQLite local, Redis production)
• Requirement: Python 3.12+

MIT License -- CONFLICT LLC