Hyperspell icon

Hyperspell

Add documents to Hyperspell and run semantic search or grounded generation from any workflow

Actions10

Search → Search

AI-generated

Summary

Search the Hyperspell index with a natural-language query to retrieve relevant document chunks ranked by semantic and lexical relevance.

Inputs

  • Query (required) — The natural-language query to run against Hyperspell.
  • Sources — Optional list of sources to search within. Leave empty to search all connected sources including Vault (recommended for agents).
  • Simplify — Whether to return a simplified version of the response instead of the raw data.
  • Options — Additional options including Max Results (number of documents to return, default 10) and Effort (level of pre-processing on the query before retrieval: Minimal, Low, Medium, High).
  • Act as User — The user ID this request runs as, overriding the credential's Act as User. Leave empty to use the credential’s setting or default to the app's user with the most documents.

Output shape

a list of matching document chunks from the Hyperspell index

The output items contain the documents matching the query, optionally simplified. The number of results is limited by Max Results (default 10, max 200). If no Sources are specified, all sources are searched. Errors are properly raised for API failures. This operation performs a POST request to /memories/query with answer set to false and returns documents ranked by relevance.

Examples

Example 1: Search all sources with a natural language query

Provide a 'Query' string and optionally leave 'Sources' empty to search all sources. Adjust 'Max Results' and 'Effort' to control response size and pre-processing level.

Links

Discussion