Graphify: Turn Your Codebase into a Queryable Knowledge Graph for Claude Code

Your codebase is too large for grep and too expensive to dump into an AI context window every time you have a question. Graphify converts code, docs, PDFs, images, and videos into a single knowledge graph you can query instantly — saving 71.5x tokens per query compared to reading raw files. In this video, you'll learn: SETUP & USAGE How to install Graphify with uv and register it with your AI assistant in three commands Building your first knowledge graph with a single slash command Query commands: graphify query, graphify path (with DFS flag), and graphify explain Output formats: interactive HTML visualization, markdown report, and JSON export UNDER THE HOOD The six-stage pipeline: detect → extract → build-graph → cluster → analyze → output Three extraction passes: free local tree-sitter parsing (28+ languages), free local whisper transcription, and LLM-powered doc/PDF/image analysis Leiden algorithm for community detection that groups related concepts across file types and languages Shared node ID mechanism that automatically merges references across code and documentation God nodes and surprising connections: identifying architectural spines and non-obvious cross-module dependencies PRACTICAL GUIDANCE When to use Graphify vs. grep vs. reading files directly Token economics: build once, query forever, with incremental updates for changed files Confidence tags (extracted, inferred, ambiguous) for knowing what's ground truth vs. LLM inference Integration with Claude Code, Codex, Cursor, Gemini CLI, GitHub Copilot, Aider, and more Export to Obsidian, SVG, GraphML (Gephi), and Neo4j Honest trade-offs: LLM costs for non-code files, 1-3 min first build, static analysis limitations GitHub: github.com/safishamsi/graphify #Graphify #KnowledgeGraph #CodebaseAnalysis #TreeSitter #AIAssistant #ClaudeCode #DevTools #CodeNavigation #StaticAnalysis #LeidenAlgorithm #CommunityDetection #TokenOptimization #NetworkX #GraphAnalysis #DeveloperProductivity #CodeUnderstanding #OpenSource #Python #RAG #CodeIntelligence