Graphify: Turn an Entire Codebase into a Queryable Knowledge Graph

Install Graphify via uv tool install graphifyy, then run graphify install to register a /graphify skill across 20+ assistants including Claude Code, Cursor, Codex, Gemini CLI, and GitHub Copilot (Codex uses $graphify). It outputs three files: graph.html for clicking through nodes and communities, GRAPH_REPORT.md summarizing key concepts and cross-file links, and graph.json as the full graph that can be queried anytime without rereading source files. Code is parsed locally with tree-sitter AST—deterministic, no LLM calls; semantic parsing of docs, PDFs, images, and video requires the assistant's own model or a configured API key. Every edge is labeled EXTRACTED (explicit in source) or INFERRED (parsing inference), distinguishing direct reads from derived results. For developers doing code search and refactoring in daily repos, direct graph queries can replace some grep and full-file reading; official benchmarks show LOCOMO recall@10 of 0.497 with 0 LLM credits, versus 0.048 for mem0 and 0.149 for supermemory. ERPNext cross-tool key fact coverage is 82.0%, above the 70.8% grep/read baseline. Graphify Cloud is positioned as a continuously running graph service supporting monorepos and cross-repo, formal verification, SDLC code review, plus Sentry and Jira connectors; it claims to cut agent token usage by 80% and offers a 14-day free trial. The local CLI suits teams running private repos, while the cloud trial fits teams wanting to plug graphs into CI and ticketing workflows.