graphify: a knowledge graph generator for codebases that replaces grepping with path-tracing and semantic queries

What it solves

Graphify transforms a codebase and its associated documentation into a searchable knowledge graph. This replaces the need to manually grep through files or rely solely on vector-based retrieval, allowing developers to trace exact paths between concepts and understand the architecture of a project through structured relationships.

How it works

Graphify parses source code using tree-sitter ASTs to deterministically map relationships (like calls, imports, and inheritance) across approximately 40 languages locally. For non-code assets like PDFs, images, and videos, it uses a semantic pass via an LLM or API. It then generates a graph.json file containing the full graph, a graph.html interactive visualization, and a GRAPH_REPORT.md summarizing key architectural highlights. Users can query this graph using commands like explain for a specific concept or path to find the shortest connection between two nodes.

Who it’s for

Developers using AI coding assistants (such as Claude Code, Cursor, GitHub Copilot, or Aider) who need a more precise way to navigate complex codebases and understand how different components interact.

Highlights

• Local-first code parsing: Uses tree-sitter AST for deterministic mapping without needing an LLM or sending code off-machine.
• Multi-modal support: Maps code, documentation, PDFs, images, and audio/video into a single unified graph.
• Explicit provenance: Every connection is tagged as either EXTRACTED (explicit in source) or INFERRED (derived by resolution).
• AI Assistant Integration: Can be registered as a skill across 20+ AI coding platforms to nudge assistants toward graph-based queries.
• Architectural Insights: Automatically identifies "God nodes" (most-connected concepts) and detects subsystems using the Leiden community detection algorithm.