A query is a user request that guides retrieval of relevant data for an LLM prompt. In RAG, the query connects a testing need to supporting documents such as user stories. section 11.1 section 11.4
A useful query-driven RAG flow:
- Accept a focused testing request, such as asking for ideas about deleting bookings. section 11.1
- Retrieve documents judged relevant by a library, tool, or similarity algorithm. section 11.4 section 11.1
- Combine the query and retrieved content in a prompt so the LLM can produce more contextual testing risks. section 11.1
Testing an API payload
The query “I want test ideas for the payload portion of the PUT branding endpoint” retrieves a user story and acceptance criteria. The resulting suggestions include missing parameters, invalid data types, invalid JSON, absent cookies, and incorrect endpoint formatting. section 11.2.2
{
"name": "string",
"map": {
"latitude": 0,
"longitude": 0
},
"logoUrl": "string",
"description": "string",
"contact": {
"name": "string",
"address": "strings",
"phone": "string",
"email": "string"
}
}
Inspect the JSON payload structure retrieved as supporting context for the branding-endpoint query. section 11.2.2
