@@ -429,7 +429,7 @@ Feature: Responses endpoint API tests
429429 And The body of the response contains Unable to connect to Llama Stack
430430
431431
432- Scenario : Check if responses endpoint with tool_choice none answers knowledge question without file search usage
432+ Scenario : Responses endpoint with tool_choice none answers knowledge question without file search usage
433433 Given The system is in default state
434434 And I set the Authorization header to Bearer eyJhbGciOiJIUzI1NiIsInR5cCI6IkpXVCJ9.eyJzdWIiOiIxMjM0NTY3ODkwIiwibmFtZSI6Ikpva
435435 And I capture the current token metrics
@@ -572,4 +572,27 @@ Scenario: Check if responses endpoint with tool_choice none answers knowledge qu
572572 """
573573 Then The status code of the response is 200
574574 And The responses output should not include an item with type "file_search_call"
575- And The token metrics should have increased
575+ And The token metrics should have increased
576+
577+ Scenario : Required allowed_tools with invalid filter returns no tool invocations on knowledge question
578+ Given The system is in default state
579+ And I set the Authorization header to Bearer eyJhbGciOiJIUzI1NiIsInR5cCI6IkpXVCJ9.eyJzdWIiOiIxMjM0NTY3ODkwIiwibmFtZSI6Ikpva
580+ And I capture the current token metrics
581+ When I use "responses" to ask question with authorization header
582+ """
583+ {
584+ "input": "What is the title of the article from Paul?",
585+ "model": "{PROVIDER}/{MODEL}",
586+ "stream": false,
587+ "instructions": "You are an assistant. You MUST use the file_search tool to answer. Answer in lowercase.",
588+ "tools": [],
589+ "tool_choice": {
590+ "type": "allowed_tools",
591+ "mode": "required",
592+ "tools": [{"non-existing": "tool"}]
593+ }
594+ }
595+ """
596+ Then The status code of the response is 200
597+ And The responses output should not include any tool invocation item types
598+ And The token metrics should have increased
0 commit comments