Add knowledge

Add searchable project content for agents and applications, then verify that retrieval returns it.

Steps

  1. Add or import the source content.
  2. Create or update RAG documents.
  3. Connect knowledge to the agents or apps that need it.
  4. Search for a known passage and inspect the returned content.

Try it with REST

Search project knowledge:

POST /projects/{project_reference}/knowledge/search

bash
curl -X POST https://api.veryfront.com/projects/support-assistant/knowledge/search \
  -H "Authorization: Bearer $VERYFRONT_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "query": "refund policy"
  }'

Import a file from connected storage:

POST /projects/{project_reference}/external-files/{provider}/{file_id}/import

bash
curl -X POST https://api.veryfront.com/projects/support-assistant/external-files/drive/<FILE_ID>/import \
  -H "Authorization: Bearer $VERYFRONT_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "folder_path": "knowledge"
  }'

Verify

The import response contains an upload with its project, path, and visibility, plus a source identifying the provider and file. Confirm those fields and find the asset in the project’s Uploads.

Importing an external file creates an uploaded asset. It does not index the file as project knowledge. The search example requires knowledge already added to the project. Add the required retrieval documents or source content before expecting a phrase from that file to appear in search.

Knowledge surfaces

SurfaceUse it for
Knowledge manifestProject-level retrieval configuration.
RAG documentSearchable document context.
Agent connectionRuntime access to retrieval context.

API reference