Runs a semantic or full-text search against a single asset's indexed documents
Executes a vector, full-text, or hybrid (RRF) query restricted to one asset and returns ranked chunks with scores. The asset must have completed at least one successful ingestion (`/operations` reports `Completed`) before results are meaningful. Primarily a debugging and tuning surface: agents at runtime issue cross-asset searches through the agent pipeline, not this endpoint. `VectorWeight`, `FtsWeight`, and `RrfK` are only consulted in hybrid mode; `ScoreThreshold` filters results post-ranking. Subject to per-tenant concurrency limits.
Authorization
Bearer JWT Authorization header using the Bearer scheme. Enter 'Bearer' [space] and then your token.
In: header
Path Parameters
The RAG asset ID.
uuidThe tenant identifier
Search query and optional parameters.
The search query text.
1 <= lengthMaximum number of results to return. Defaults to the service setting.
int32Minimum relevance score (0–1). Results below this threshold are excluded.
double"UNSPECIFIED" | "VECTOR" | "FULL_TEXT" | "HYBRID"Weight for vector (semantic) results in hybrid search (0–1). Only used when SearchMode is hybrid.
doubleWeight for full-text search results in hybrid search (0–1). Only used when SearchMode is hybrid.
doubleReciprocal Rank Fusion constant for hybrid search. Higher values dampen rank position effect.
int32Response Body
curl -X POST "https://api.fruxon.com/v1/tenants/string/assets/497f6eca-6276-4993-bfeb-53cbbbba6f08:search" \ -H "Content-Type: application/json" \ -d '{ "query": "string" }'{
"results": [
{
"content": "string",
"assetId": "00000000-0000-0000-0000-000000000000",
"sourceType": "string",
"chunkId": "00000000-0000-0000-0000-000000000000",
"chunkIndex": 0,
"source": "string",
"sourceUrl": "string",
"score": 0,
"startCharIdx": 0,
"endCharIdx": 0,
"pageNumber": "string",
"startLine": 0,
"endLine": 0,
"documentId": "00000000-0000-0000-0000-000000000000",
"headingPath": "string",
"contextText": "string",
"totalChunks": 0,
"windowChunks": [],
"sourceLabel": "string",
"codeIntel": {}
}
],
"totalResults": 0,
"searchMode": "VECTOR"
}{
"type": "string",
"title": "string",
"status": 0,
"detail": "string",
"instance": "string",
"property1": null,
"property2": null
}Fetches one sample record from a tool-source's list tool so the configure UI can offer the record's fields as a picker
Invokes the configured list tool once (single page) against the referenced integration config and returns the first record it yields, so the asset-configuration UI can present the record's fields as a field picker before the producer asset is created. The probe is read-only: nothing is persisted and no asset is provisioned. Supply `recordsPath` when the list tool returns the records nested inside a page envelope rather than as the top-level array. Returns a null `record` when the tool yields no rows; responds 400 when the integration config or list tool is missing.
Lists available asset source types with their required and optional properties.
Returns the descriptors that drive asset creation: each entry lists the source type key (for example `google_drive`, `url`, `file`), its required and optional config parameters, and whether it supports OAuth. Use this to build the asset-creation form and to validate the `config` payload sent to `POST /assets`. Source types that report OAuth support must be authorized via `POST /assets:authorize` before creation.