graphify: a project-wide knowledge graph generator that turns code and docs into a queryable map for AI assistants

What it solves

Graphify transforms a project's disparate files—including code, documentation, PDFs, images, and videos—into a queryable knowledge graph. This eliminates the need to manually grep through files or read through entire directories to understand a codebase's architecture and connections.

How it works

Graphify uses a combination of local AST extraction (via tree-sitter for code) and AI model APIs for non-code files. It maps the project into a graph structure and generates three primary outputs: an interactive HTML visualization, a detailed Markdown report highlighting key concepts and "god nodes," and a JSON file for programmatic querying. It can be integrated as a "skill" into various AI coding assistants (like Claude Code, Cursor, and GitHub Copilot), allowing the assistant to query the graph directly rather than reading raw files.

Who it’s for

Developers and architects who need to navigate complex codebases, understand the relationships between different modules, and quickly onboard to new projects.

Highlights

• Broad File Support: Handles 36 tree-sitter grammars for code, as well as PDFs, Office documents, Google Workspace files, and video/audio transcriptions.
• AI Assistant Integration: Installs as a native skill across dozens of platforms (e.g., Claude Code, Cursor, Aider, Trae) to nudge assistants toward graph-based queries.
• Deep Insights: Identifies "god nodes" (most-connected concepts) and "surprising connections" between distant modules.
• Confidence Tracking: Marks every relationship as EXTRACTED, INFERRED, or AMBIGUOUS to distinguish between hard facts and AI guesses.
• Architecture Export: Can generate Mermaid call-flow diagrams for architecture documentation.