Measured on this repo, graphify cost more than it returned. `graphify query`
answers with a keyword-seeded BFS neighbourhood, not a location: asking where
CORS origin reflection lives returned 73 nodes, mostly api_test.go helpers
plus an unrelated golang-performance doc section matched on the word
"reflection", and never named httpmw/middleware.go:135 or main.go:121 — which
grep gives in 40ms. `graphify explain` on a known symbol is sound but
duplicates what the LSP already answers against live files.
Against that, the PreToolUse hooks injected a "run graphify query first"
paragraph on every grep and every source read, the map cost 685k input tokens
across five builds, and graph.json plus graph.html carried 3.4MB through 11
commits of churn. AGENTS.md is the better orientation artifact: it holds the
CDP and clock findings, the adapter list, and the security invariants, none of
which an AST graph can derive.
Removes the committed map, the AGENTS.md rules block, the .gitignore entries,
both hooks, and the stale `graphify query` instruction in the
implement-tickets skill. The CLI, its skill directory, and the post-commit /
post-checkout git hooks were uninstalled outside the tree.
sulthan
d168cf1ad0