fix: respect guardrail mock_response during during_call to return blo…

[uc4w6c]

Title

fix: respect guardrail mock_response during during_call to return blocked output

Relevant issues

Fixes #15092

Pre-Submission checklist

Please complete all items before asking a LiteLLM maintainer to review your PR

• I have Added testing in the tests/litellm/ directory, Adding at least 1 test is a hard requirement - see details
• I have added a screenshot of my new test passing locally
• My PR passes all unit tests on make test-unit
• My PR's scope is as isolated as possible, it only solves 1 specific problem

Type

🐛 Bug Fix
✅ Test

Changes

Example configuration:

  - guardrail_name: "bedrock-guard"
    litellm_params:
      guardrail: bedrock
      mode: "during_call"
      guardrailIdentifier: xxxxx
      guardrailVersion: "DRAFT"
      aws_region_name: us-east-1
      disable_exception_on_block: true

Example curl test:

curl -i  -X POST http://localhost:4001/v1/chat/completions \
-H "Content-Type: application/json" \
-H "Authorization: Bearer sk-1234" \
-d '{
    "model": "gpt-4o",
    "messages": [
        {
            "role": "user",
            "content": "My name is Yuta. Say my name."
        }
    ], "guardrails": ["bedrock-guard"]
}'
HTTP/1.1 200 OK
date: Fri, 28 Nov 2025 05:17:54 GMT
server: uvicorn
content-length: 316
content-type: application/json
x-litellm-call-id: 0b9dc18c-87f0-4aeb-bf62-87f19c41668c
x-litellm-version: 1.80.0
x-litellm-response-cost-original: 0.00016250000000000002
x-litellm-response-cost-discount-amount: 0.0
x-litellm-key-spend: 0.0
x-litellm-applied-guardrails: bedrock-guard

{"id":"chatcmpl-6ffa39bd-c3f4-4932-a555-b6e6a63fed06","created":1764307087,"model":"bedrock-guardrail","object":"chat.completion","choices":[{"finish_reason":"stop","index":0,"message":{"content":"The name cannot be returned.","role":"assistant"}}],"usage":{"completion_tokens":0,"prompt_tokens":0,"total_tokens":0}

[vercel]

@uc4w6c is attempting to deploy a commit to the CLERKIEAI Team on Vercel.

A member of the Team first needs to authorize it.