Edge
Graphify glossary
A typed, directed relationship between two nodes — calls, imports, defines, references — each carrying a provenance tag.
An edge is a typed, directed relationship between two nodes: PaymentService calls StripeClient, worker.py imports queue.py, migration 0042 defines the sessions table. The type says what kind of relationship it is; the direction says which way it runs. Edges are what turn a collection of entities into a graph — and they're the thing search indexes fundamentally don't have.
That's why edges, not nodes, are the real payoff. A text search can find every file mentioning a function's name — including comments, tests, and an unrelated variable that shadows it — but it can't tell you which of those mentions is an actual call. An edge exists because the relationship exists: the parser found the call in the AST, or the model connected a doc to the service it describes. Once edges are in place, questions become traversals: follow calls edges backwards to find every caller, keep following to find their callers, and you've computed a blast radius instead of guessing at one.
Every edge in a Graphify graph also records where it came from. EXTRACTED edges came straight out of the tree-sitter AST and are deterministic. INFERRED edges were proposed by your model and are labeled as judgment calls. AMBIGUOUS edges had evidence that couldn't be fully resolved and are kept but flagged. This provenance is attached at the edge level precisely because it's the edges an assistant reasons over — knowing which links were found and which were guessed is the difference between an answer you can check and one you have to take on faith.
Related terms
- Node — A single entity in the graph — a function, class, file, table, or doc page — carrying attributes like name, path, and kind.
- Confidence tags — Graphify's provenance labels — EXTRACTED, INFERRED, AMBIGUOUS — attached to every edge so you always know what was found versus what was guessed.
- Traversal / hop — Following edges from node to node — one hop is a single edge, a traversal is a path of them. How graphs answer multi-step questions.