Problem. graphify explain "<symbol>" on a high-degree node (e.g. a logging helper with degree 692) prints ~20 connections plus "... and 672 more". The question that leads people to explain ("who calls this, what's the impact?") is exactly the one the truncation leaves unanswered — in our sessions the agent then falls back to a repo-wide grep, which is what the graph was meant to replace.
Proposal. For nodes above a degree threshold, group incoming edges by directory/file with counts (e.g. inc/email-inbox: 84 callers across 12 files), and/or add --callers --all / --group-by=dir flags. Full flat output behind a flag is fine; the default should just stop hiding 97% of the answer.
Environment. graphifyy 0.8.36, code-only graph, ~22k nodes / ~33k edges.
Problem.
graphify explain "<symbol>"on a high-degree node (e.g. a logging helper with degree 692) prints ~20 connections plus "... and 672 more". The question that leads people toexplain("who calls this, what's the impact?") is exactly the one the truncation leaves unanswered — in our sessions the agent then falls back to a repo-wide grep, which is what the graph was meant to replace.Proposal. For nodes above a degree threshold, group incoming edges by directory/file with counts (e.g.
inc/email-inbox: 84 callers across 12 files), and/or add--callers --all/--group-by=dirflags. Full flat output behind a flag is fine; the default should just stop hiding 97% of the answer.Environment. graphifyy 0.8.36, code-only graph, ~22k nodes / ~33k edges.