repo-graph

Start from a failure, not a filename — turn a stacktrace, failing test, or diff into the exact code that matters, then read it. Use when you have an error and need to know where to look.

Locate where a feature or symbol lives — match nodes by name, or pull the related cluster from a seed via spreading activation. Use to scope before reading.

Blast radius before a change — everything a node affects (forward) or that depends on it (backward), ranked and located, with likely-dead code flagged. Also the one-hop view of direct connections.

Set up or rebuild the structural graph for a repo — on first use, after a big refactor, or to wire repo-graph into a project permanently.

Get oriented in a codebase before reading files — repo shape, entry points, where the graph is blind, and the full structural map. Use at the start of any task in an unfamiliar or large repo.

Follow the code across boundaries — a feature end-to-end, or the shortest path between two nodes (e.g. a frontend action to its backend handler). Links across the stack in one call instead of grepping both sides.