[Bug]: Unwanted params attached to guardrails

Please select the area the issue is related to

Area/Management (Management API or Management Portal UI), Area/AIGateway (AI Gateway runtime/control plane)

Please select the aspect the issue is related to

Aspect/API (API backends, definitions, contracts, interfaces, OpenAPI), Aspect/AI (AI/LLM integration, MCP, AI readiness)

Description

There are some unwanted params attached to the guardrails deployed in the gateway. This can be observed through the policy engine config dump. This does not seem to impact the functionality.

{
            "name": "word-count-guardrail",
            "version": "v1",
            "enabled": true,
            "execution_condition": null,
            "parameters": {
              "attachedTo": "route",
              "completionTokens": {
                "identifier": "$.usage.completion_tokens",
                "location": "payload"
              },
              "promptTokens": {
                "identifier": "$.usage.prompt_tokens",
                "location": "payload"
              },
              "remainingTokens": {
                "identifier": "x-ratelimit-remaining-tokens",
                "location": "header"
              },
              "request": {
                "enabled": true,
                "invert": false,
                "jsonPath": "$.messages[-1].content",
                "max": 5,
                "min": 3,
                "showAssessment": false
              },
              "requestModel": {
                "identifier": "$.model",
                "location": "payload"
              },
              "response": {
                "enabled": false,
                "invert": false,
                "jsonPath": "$.choices[0].message.content",
                "max": 0,
                "min": 0,
                "showAssessment": false,
                "streamingJsonPath": "$.choices[0].delta.content"
              },
              "responseModel": {
                "identifier": "$.model",
                "location": "payload"
              },
              "totalTokens": {
                "identifier": "$.usage.total_tokens",
                "location": "payload"
              }
            }
          },

Steps to Reproduce

Deploy an LLM provider with a guardrail (or any policy) attached.

Severity Level of the Issue

Severity/Trivial (Cosmetic issues. Can be fixed in future releases)

Environment Details (with versions)

No response