---
name: "typesense-instant-in-memory-search"
description: "Achieve sub-50ms instant search across millions of records using in-memory Typesense."
metadata:
  author: "Typesense"
  version: "1.0.0"
  runtime_compatibility:
    - "codex"
    - "claude-code"
    - "cursor"
    - "windsurf"
triggers:
  - "create typesense schema"
  - "query typesense facets"
  - "configure typesense hybrid search"
permissions:
  fs_read: true
  fs_write: "workspace"
  shell_exec: false
  net_access: false
---

# Typesense In-Memory Fulltext & Vector Search

Configures Typesense schemas: handling numerical filtering, faceting, geo-search, typo tolerance, hybrid vector search (OpenAI / SBERT), and multi-node Raft clustering.

## When to Invoke
Invoke this skill whenever you need to:
- create typesense schema
- query typesense facets
- configure typesense hybrid search

## Execution SOP & Rules
1. Define schema JSON specifying fields, types, and facet: true attributes.
2. Create collection: await client.collections().create(schema).
3. Import documents in JSONL batches using client.collections('items').documents().import(data).
4. Query with faceting: client.collections('items').documents().search({ q: 'laptop', query_by: 'name,desc', facet_by: 'brand' }).

## Verification Checklist
- Run local linting, syntax validation, and type-checks before reporting completion.
- Ensure all created or modified files adhere to workspace boundaries and avoid uncommitted secrets.
- Verify tests or build commands exit cleanly with return code 0.