Graphify maps code into knowledge graphs to fix vector search's blind spot for coding agents
JeremyCMorgan · x · 2026-11-05
Graphify argues vector search finds semantically relevant code but can't reliably explain dependency paths. Their post walks an end-to-end worked example: for "where does an expired token get rejected," vector top-5 returns plausible-but-wrong functions, while the code knowledge graph recovers the real dependency chain via CALLS hops.
It surveys published failure cases: the orphan function (semantic distant, structurally close), the bug three hops down that token-based retrieval never reaches, and near-identical preprocessing functions vector search can't distinguish.
Graphify maps function calls and exceptions into a graph, distinguishing relationships extracted from source code from those inferred, and labels every hop so agent retrieval is inspectable. All examples run on a public repo you can clone yourself.
More from coding & agent
- Next-gen task management: agents should book human attention via MCP — lxfater · 2026-11-05
- GraphRAG in real projects: 3 practical patterns from Daan Di Scala's webinar — Al_Grigor · 2026-11-05
- Orderain ships an MCP server that lets Claude run your ecommerce store on autopilot — Slight_Guest3459 · 2026-11-05
- OpenAI 'Presence' page surfaces, pushing trusted AI agents into customer channels — juberti · 2026-11-05
- RL teaches agents to use their own notes; training survives 23M tokens of memory edits — simran_s_arora · 2026-11-05
- Dev forks Ollama into Ollamapaca, adding plugins with Chrome and computer control — RichardsonDx · 2026-11-05