🚀 Feature: Guardrails

[dartpain]

🔖 Feature description

Add guardrails dropdown with checkboxes (turn guardrails on / off) in agent builder + abstraction with guardrails implementations in the code to handle input and output.

🎤 Why is this feature needed ?

Safety

✌️ How do you aim to achieve this?

Create abstraction and possibility to handle input and output separately

🔄️ Additional Information

No response

👀 Have you spent some time to check if this feature request has been raised before?

• I checked and didn't find similar issue

Are you willing to submit PR?

None

[matthewschaney]

Hello, could I please take this issue?

[matthewschaney]

Hi @dartpain, I am still finishing this up - thank you for assigning it to me. 😊

[dartpain]

Amazing, if you have any questions or want to run some questions by me, please don't hesitate to ask.

[shanumas]

If I understand correctly, there was zero-guardrails consideration until now. You guys are going to add it now ?

[dartpain]

@shanumas So that agents can be deployed more safely in production environments.
We had this issue on our planned list for many months now, we decided to create this issue recently.

Also since we have a more stable version of agents - I think its one of the first improvements that comes to attention.

[shanumas]

@dartpain Thanks for your reply! I'm a big fan of DocsGPT and would love to contribute by adding guardrails. Do you think that would be helpful?