Graphify is an open-source tool that converts any folder of code, documentation, papers, or images into a structured, queryable knowledge graph. Designed as a skill for AI coding assistants including Claude Code, Codex, OpenCode, OpenClaw, and Factory Droid, it gives developers a bird's-eye view of codebases and document collections that would otherwise require hours of manual exploration. The tool supports 19 programming languages through tree-sitter AST parsing including Python, JavaScript, TypeScript, Go, Rust, Java, C, C++, Ruby, Swift, Kotlin, and more, extracting functions, classes, imports, and their relationships automatically.
Graphify operates in two passes to maximize coverage. The first pass is a fast, deterministic AST extraction that maps code structure without any LLM calls. The second pass deploys Claude subagents in parallel to process documents, PDFs, markdown files, screenshots, diagrams, and whiteboard photos using vision capabilities, extracting concepts and relationships from non-code content. The results from both passes merge into a single NetworkX graph, which is then clustered using Leiden community detection to reveal natural module boundaries and conceptual groupings across the entire project.
The output includes interactive HTML visualizations for exploring the knowledge graph in a browser, queryable JSON for programmatic access, and a plain-language audit report summarizing the project structure. With over 6,000 GitHub stars, Graphify has gained traction among developers who work with large or unfamiliar codebases and need rapid orientation. Recent updates added support for Zig and PowerShell, expanding its language coverage for systems and DevOps-oriented projects.