What problem does it solve? Codebases are hard to navigate with grep alone, and ad-hoc knowledge graphs drift out of sync or get overwritten when machines run different tool versions. This Skill performs an end-to-end graphify setup for the current project: it extracts an AST-based knowledge graph, labels communities semantically, commits the graph into the repo so it travels via git pull, and installs refresh hooks and a union-merge driver to keep it consistent. ## Core Features & Use Cases - In-project committed graph: Builds graphify-out/graph.json inside the repo with .gitignore surgery so the graph is tracked and shared across machines. - Two refresh tiers: A free hook-driven AST structure rebuild on post-merge, plus an on-demand paid DeepSeek semantic pass that re-derives edges and community labels. - Version pinning and dedup patching: Installs a pinned graphifyy build via uv and applies a vendored dedup fix on pre-0.9 versions so all machines produce structurally identical graphs. - MCP query surface: Registers a user-scoped graphify MCP server for graph-first code questions, with the CLI as backup. - Use Case: After cloning a large Laravel monorepo, run the skill to generate a labeled knowledge graph, then ask "how does the order lifecycle work" and get an answer from the graph instead of running five greps. ## Quick Start Ask the agent to run /graphify-this-project to set up the graphify knowledge graph for the current repository.