ModelRefs / Vector Search Tool — Tool Pattern
Vector Search Tool — Tool Pattern
Expose semantic search over an embedded corpus as a typed tool the model can call to ground its answers.
Overview
Wrap an embedding store behind a search(query, top_k, filters) tool. The runtime embeds the query, performs ANN search, optionally reranks, and returns ranked snippets with citations.
When to use it: You need the model to ground answers in your private corpus.
Pattern details
| Pattern class | retrieval |
|---|---|
| Difficulty | intermediate |
| Invocation mode | synchronous |
| Also known as | semantic search tool, rag tool |
| Last reviewed | 2026-06-07 |
Known failure modes
- Low recall — Relevant chunk missing from top_k. Mitigation: Increase top_k; add reranker; tune chunking.
- Stale index — Corpus changed but index not refreshed. Mitigation: Schedule reindex; track corpus version.
When not to use it
- Returning raw documents instead of chunked, cited snippets.
Continue your research
Use these connected ModelRefs sections to compare alternatives, inspect implementation paths, and review the evidence and governance boundaries relevant to Vector Search Tool — Tool Pattern.
Frequently asked questions
When should I use the Vector Search Tool tool pattern?
You need the model to ground answers in your private corpus.
What are common failure modes of Vector Search Tool?
Low recall • Stale index
Is Vector Search Tool production-ready?
Yes when paired with the safety controls and observability hooks documented on the pattern page.