fix: added RAI check in user clarification endpoint

[Dhruvkumar-Microsoft]

Purpose

This pull request introduces updates to the rai_success function and its usage across the codebase to enhance Responsible AI (RAI) checks. Key changes include adding a new parameter to distinguish between task creation and other contexts, refining the RAI evaluation logic, and updating endpoints to handle RAI failures more robustly.

Enhancements to RAI functionality:

• src/backend/utils_kernel.py: Updated rai_success to accept a new is_task_creation parameter, allowing for context-specific evaluation. The function now uses a dynamic content_prompt tailored to either general input validation or task creation. [1] [2] [3]

Updates to endpoints for RAI integration:

• src/backend/app_kernel.py: Modified input_task_endpoint to pass True for the is_task_creation parameter when calling rai_success.
• src/backend/app_kernel.py: Enhanced human_clarification_endpoint to invoke rai_success with False for non-task-related clarification. Added robust error handling and event tracking for RAI failures.

Does this introduce a breaking change?

• Yes
• No

How to Test

• Get the code

git clone [repo-address]
cd [repo-name]
git checkout [branch-name]
npm install

• Test the code

What to Check

Verify that the following are valid

• ...

Other Information

[github-actions]

🎉 This PR is included in version 2.2.0 🎉

The release is available on GitHub release

Your semantic-release bot 📦🚀