I would like to use Rules for AI to have general instructions for its responses, as well as, have project/app specific instructions that stay with the app.
In Settings/Rules_for_AI I pointed the AI to /REQUIREMENTS.md and in that file I assigned the AI the name “Jarvis”. To test this in chat I asked using ‘no context’, ‘chat’, and ‘with codebase’ but the AI did not find the instruction of its name being “Jarvis”.
Is there a way to make this work?
If not then I will change this to a Feature Request.
I can reference the file in chat with @REQUIREMENTS.md for app specific requirements or I could use the Settings/Rules for AI and have sections for ‘general instructions’ and for each of my apps. This seems like overkill and makes the prompts larger than they need to be for all projects.
It would be great to have the chat @ functionality in the Rules for AI setting.
This way I could add tech docs and an app-specific doc to the prompts.
Possibly with the ability to flag each reference for either all submit buttons or specify which ones to use with each reference.
1 Like
second
+1 I agree, it would be beneficial to have the ability to add custom context to guide the LLM’s responses in the right direction. For instance, when working with Python code generated by the LLM, it would be helpful if the code adhered to PEP 8 standards or allowed for specific PEP rules to be excluded. This could be achieved by loading a simple text file with these instructions into the context before sending it to the LLM.
1 Like