Getting access to raw response object in a Guardrail? #32
Comments
|
Hey @Manouchehri not today but I think it's a good feature request! We will have to see how to best do it since the chat completions response gets converted into a different format before it reaches any guardrail execution so it's not as easy as exposing it. |
|
Hmm, is it possible to do with a hook and add the raw response to the context? (I tried and failed to add that.) |
|
I think what we should do is move all the optional fields into providerData properties on the messages and then expose the full ModelResponse to the output guardrail as a separate field |
|
This issue is stale because it has been open for 7 days with no activity. |
|
Not stale. :P |
|
This should get fixed in the 0.0.8 release today. Change was done in #104. There is now an additional |
Inside a Guardrail, is there a way I can access the raw response object? e.g. I am trying to read
prompt_filter_resultsin my Guardrail. Example response object:{ "choices": [ { "content_filter_results": { }, "finish_reason": "tool_calls", "index": 0, "logprobs": null, "message": { "annotations": [], "content": null, "refusal": null, "role": "assistant", "tool_calls": [ { "function": { "arguments": "{}", "name": "transfer_to_History_Tutor" }, "id": "call_UsPG83icg4t0pZsf1uQWI97W", "type": "function" } ] } } ], "created": 1749158752, "id": "chatcmpl-BfCNsjM9bfc3k7Dj3vnGZlN6Zmp5V", "model": "gpt-4.1-nano-2025-04-14", "object": "chat.completion", "prompt_filter_results": [ { "prompt_index": 0, "content_filter_results": { "hate": { "filtered": false, "severity": "low" }, "jailbreak": { "filtered": false, "detected": true }, "self_harm": { "filtered": false, "severity": "safe" }, "sexual": { "filtered": false, "severity": "safe" }, "violence": { "filtered": false, "severity": "safe" } } } ], "system_fingerprint": "fp_68472df8fd", "usage": { "completion_tokens": 15, "completion_tokens_details": { "accepted_prediction_tokens": 0, "audio_tokens": 0, "reasoning_tokens": 0, "rejected_prediction_tokens": 0 }, "prompt_tokens": 200, "prompt_tokens_details": { "audio_tokens": 0, "cached_tokens": 0 }, "total_tokens": 215 } }The text was updated successfully, but these errors were encountered: