Actions7
- Knowledge Base (RAG) Actions
- Document Actions
- Memory Actions
Knowledge Base (RAG) → Search
AI-generatedSummary
Searches the GuidedMind knowledge base for chunks relevant to a query by sending a POST request to /rag/search. Use the limit and similarity threshold to control how many results are returned and how closely they must match; metadata is included by default.
Inputs
- Query (required) — Search text, sent as
query. The UI describes a length of 1–1000 characters. - Limit — Maximum number of chunks to return, sent as
k; defaults to 5 (UI range: 1–20). - Threshold — Minimum similarity score from 0 to 1; lower values can return more, less precise matches. Defaults to 0.7.
- Include Metadata — Whether to include metadata in the search response, sent under
options.includeMetadata. Defaults to true. - Request Options — Optional HTTP request settings: batching, accepting invalid SSL certificates, proxy, and timeout. Defaults include batches of 50 with a 1000 ms interval, SSL certificate validation enabled, no proxy, and a 10000 ms timeout.
Output shape
A single API response containing the search results.
The source pack declares the request method and endpoint but does not show the response schema or result fields.