Prompt optimizations

[akatsoulas]

No description provided.

[Copilot]

Pull Request Overview

This PR optimizes the prompt instructions used for content moderation and topic classification in Mozilla's "{product}" support forums. Key changes include:

• Updating setup.cfg to ignore E501 in the prompt file.
• Overhauling the prompt in kitsune/llm/questions/prompt.py to clarify and improve the instructions for spam detection and classification.

Reviewed Changes

Copilot reviewed 2 out of 2 changed files in this pull request and generated no comments.

File	Description
setup.cfg	Adds per-file ignore for E501 specifically for the prompt file
kitsune/llm/questions/prompt.py	Refines the prompt for both spam and topic classification with updated instructions

Comments suppressed due to low confidence (1)

kitsune/llm/questions/prompt.py:36

• [nitpick] Consider defining a common constant for JSON formatting instructions instead of using the same placeholder in multiple locations to improve maintainability and reduce duplication.

{format_instructions}

[escattone]

r+wc

I really like your changes, especially the adjustments to the topic prompt. My only recommendation is to remove the formatting instruction preambles that you added, since they're not needed.

[escattone]

The StructuredOutputParser.get_format_instructions() takes care of adding the following preamble (prior to the JSON schema):

The output should be a markdown code snippet formatted in the following schema, including the leading and trailing "```json" and "```":

so this line can be removed.

[escattone]

Same comment as above.

[escattone]

Perhaps also remove this extra empty line?

[escattone]

Perhaps keep this blank line to add some separation from the next section?