What Is graphify?
graphify is an AI agent skill that ingests any folder of files — code, documentation, research papers, images, audio, video — and converts them into a queryable, interactive knowledge graph. It identifies entities, relationships, and communities across your entire corpus and renders the result as a navigable HTML graph. Key capabilities:
25 programming languages
Code ingestion with language-aware entity extraction
Multimodal
Processes images, video, audio, and documents alongside code and text
Leiden community detection
Automatically clusters related content into communities
Interactive HTML graph
Navigable visualization of the entire knowledge structure
Queryable
Ask questions against the graph; answers are grounded in the connected structure
Supported File Types
Code and Documentation
25 programming languages with language-aware parsing: Python, JavaScript/TypeScript, Rust, Go, Java, C/C++, Ruby, PHP, Swift, Kotlin, and more. Documentation in Markdown, RST, AsciiDoc, and HTML is parsed for concepts, references, and structure.
Multimodal
Images — OCR and visual content description, embedded in the graph as nodes with descriptive metadata.
Video and Audio — Transcription-based ingestion; content becomes searchable text nodes linked to their source timestamps.
Research Papers — PDF parsing with entity extraction, citation graph construction, and concept clustering.
Data Files
CSV, JSON, and structured data files are parsed for schema, key values, and relationships to other files in the corpus.
How It Works
Step 1: Ingestion — graphify scans the target folder and processes each file with format-appropriate parsers.
Step 2: Entity extraction — Named entities, code symbols, concepts, and references are extracted from each file.
Step 3: Relationship mapping — Cross-file references, shared concepts, and semantic similarity are used to build edges between nodes.
Step 4: Leiden clustering — The Leiden community detection algorithm groups related nodes into clusters, revealing the natural structure of the corpus.
Step 5: HTML rendering — The graph is rendered as an interactive HTML file with zoom, pan, search, and filter.
How to Install graphify?
npx skills add Graphify-Labs/graphify Then run against a folder: Use graphify to map the relationships in /my-project Or download from GitHub https://github.com/Graphify-Labs/graphify