| name | search |
| description | Enterprise search across all connected data sources. Provides hybrid semantic + keyword search, similar document finding, and search analytics. |
| category | search |
| version | 1.0.0 |
| tools | ["search_hybrid","search_semantic","find_similar_documents"] |
| triggers | {"keywords":["search","find","look for","query","where is","locate"]} |
| examples | [{"input":"Find documents about Q4 planning"},{"input":"Search for emails from last week about the budget"},{"input":"Where are the engineering specs?"}] |
Search Skill
You have access to powerful enterprise search capabilities across all connected data sources.
Available Tools
-
search_hybrid - Combined semantic + keyword search (best for most queries)
- Uses both vector embeddings and text matching
- Automatically reranks results for relevance
-
search_semantic - Pure semantic/vector search (best for conceptual queries)
- Finds conceptually related content even without keyword matches
- Ideal for questions like "how does authentication work"
-
find_similar_documents - Find documents similar to a given document
- Takes a document ID and finds related content
- Useful for exploring related topics
Best Practices
- Use hybrid search for most queries (balances precision and recall)
- Use semantic search when looking for conceptually related content
- Always respect access control - results are filtered by user permissions
- Include relevant filters (date range, connector type) to improve precision
Query Optimization
- Be specific: "Q4 2024 sales projections" > "sales"
- Include entity names when known: "John's performance review"
- Use date filters for time-sensitive queries
- Specify connector types when searching specific sources
Response Format
Always cite sources when presenting search results. Include:
- Document title
- Relevant excerpt
- Source connector
- Last updated date