715 defence successfully configured message should not disappear after a delay

dhinrichs-scottlogic commented 7 months ago

Description

Previously, when a user changed or reset the text or number inputs of the defence and model configurations, a message would pop up for 3 seconds to indicate to the user that the change has been made. This can easily be missed, which is why I moved this message to the chat, where the user already gets updates when defences get toggled on/off. Additionally, I added an info message for the GPT model selection and the model sliders

Screenshots

Before:

After:

Notes

changed two key/value pairs of the enum DEFENCE_ID: FILTER_USER_INPUT to INPUT_FILTERING and FILTER_BOT_OUTPUT to OUTPUT_FILTERING
added the info messages to the chatHistory
removed the old pop up info messages

Concerns

Due to the naming of the Information Defence, "defence" is repeated in the chat, there is already a ticket to address this.
There is currently no error messages for the user if the backend fails while we are changing a configuration. There's a separate ticket to address this.
The defence.name and the DEFENCE_ID are slightly different q/a LLM versus qa LLM. I couldn't change the enum to include the "/" in the enum key (which is what's being used in the function to update the config) and I didn't want to create a special conditional to change the output to the UI. Personally I think the difference between q/a and qa is acceptable for now. There is a general question about how useful the DEFENCE_ID enum is given that the defence names and the enum is almost identical, so this could become a bigger refactor.

Checklist

Have you done the following?

[x] Linked the relevant Issue
[ ] Added tests
[x] Ensured the workflow steps are passing

pmarsh-scottlogic commented 7 months ago

this change interacts amusingly with this bug #742 Not something to block this PR, but we should prioritise the bug once this is in

dhinrichs-scottlogic commented 7 months ago

This all turned out much more complicated than anticipated, so after some discussion decided to approach it slightly differently and started again from scratch. New PR: https://github.com/ScottLogic/prompt-injection/pull/802

ScottLogic / prompt-injection