graphify

I’m seeing Graphify-Labs/graphify trending on GitHub, and the concrete pitch is practical: take an existing repo (not a prompt) and convert it into a knowledge graph that links application code, database schema, and infrastructure into one structure you can query.

This is the kind of thing I’d reach for when I need an agent to answer engineering questions with minimal hand-holding—for example: “Where is this table written, what service owns it, what migrations/SQL views affect it, and which deployable references it?” If you feed the assistant only text snippets, it usually misses cross-file and cross-layer context. A repo graph is a better substrate for RAG/agent retrieval.

What I’d look at first: - The entry point in the README for generating the graph from a local folder - Examples showing how SQL schemas and app code get connected - The output format / query interface (how you actually ask questions) - Any flags or config knobs for scoping what gets indexed (so you don’t graph the whole universe by default)

If you run Claude Code or similar workflows, the useful pattern is to build the graph once per repo (or per commit), then let the agent query it during planning and tool-calling. That turns “turn context into text” into “turn context into structured retrieval.”