Graphify is an AI coding assistant skill that transforms various data types like code, documents, images, and videos into queryable knowledge graphs. You invoke it using a command like /[graphify](/lucasrosati/claude-code-memory-setup) in compatible AI coding tools (Claude Code, Codex, Gemini CLI, etc.). It analyzes your input, identifies relationships, and presents them as a visually navigable graph, revealing connections you might otherwise miss. The core of Graphify's functionality relies on leveraging LLMs and local processing to build a structured representation of your information.
-
Multimodal Input: Graphify accepts code, PDFs, images, audio, and video for graph creation. It handles files in 25 languages via tree-sitter AST.
-
Deterministic Graph Construction: The tool constructs graphs in three passes: AST analysis, audio/video transcription, and LLM-based relationship extraction.
-
No Embedding Dependency: Graphify utilizes graph-based community detection for clustering, eliminating the need for separate vector databases or embedding steps. Edges are tagged to show extraction confidence.
-
Flexible Output Options: The output is available as an interactive HTML graph, a JSON file for querying, and a plain-language report, enabling various use-cases.
-
Source Provenance: Graphify distinguishes between directly extracted relationships, inferences, and ambiguous connections, ensuring transparent and auditable results.
-
Code Analysis: Extracts code structure, function calls, comments, and dependencies for knowledge graph creation.
-
Document Understanding: Processes text-based documents (PDFs, Markdown) to identify key concepts and their relationships.
-
Multimedia Integration: Extracts information from images, audio, and video files, including transcribed audio/video segments.
-
Interactive Exploration: Provides an interactive HTML visualization allowing users to explore graph nodes, filters, and search functionality.
-
Cross-Repository Merging: Combines graphs from multiple repositories for comprehensive dependency mapping.
-
CI/CD Integration: Supports CI integration for automated graph generation and updates against source code changes.
-
Configurable Globals: Allows setting global config configurations for enabling/disabling experimental functionality.
Graphify is an actively developed project with regular updates and bug fixes. Recent changes focus on improved merging capabilities, deduplication of nodes, and improved prompt engineering for better transcriptions using Whisper. It incorporates feedback and addresses user concerns. The continuous development and active community engagement are indicators of its ongoing refinement.
Graphify benefits developers, researchers, and anyone working with complex information. It helps understand codebase architecture, explore relationships in documents, and uncover hidden connections in multimedia. It provides a clear overview of data dependencies and facilitates efficient information discovery by presenting data in a graphical format. Instead of manual analysis, graphify offers a fast and insightful way to grasp complex relationships that are otherwise difficult to discern.