Skip to content

feat: [Prompt Scoring Engine with Multi-Dimension Quality Analysis #104

Description

@Moresh-Mishra

Feature Request

Problem Statement

Is your feature request related to a problem? Please describe.

Currently, PromptPilot enhances a user's raw prompt but gives zero feedback on
why the original prompt was weak or how much it improved after forging.
Users have no visibility into what makes their prompt good or bad — they're flying
blind. This makes it hard to learn better prompting habits over time, and it reduces
trust in the enhancement output since there's no before/after quality comparison.


Proposed Solution

Describe the solution or enhancement you'd like to see added to PromptPilot.

Implement a Prompt Scoring Engine that automatically evaluates a prompt across
multiple quality dimensions and displays the result as a visual score card inside
the extension popup.

Scoring Dimensions (suggested)

Dimension What it measures
🎯 Clarity How unambiguous and direct the prompt is
📐 Specificity Amount of concrete detail provided
🧩 Context Background information given to the AI
🏁 Goal Orientation How clearly the desired output is defined
🔧 Structure Logical flow and formatting of the prompt

How it works

  1. User types a raw prompt in the textarea
  2. On clicking "✦ Forge Prompt", the scoring engine evaluates the raw prompt
    first, then the enhanced version
  3. Both scores are displayed side-by-side as a radar chart or 5-bar score card
  4. Each dimension shows a score out of 10 with a one-line explanation
  5. An overall Prompt Grade (A / B / C / D) is shown as a badge

Scoring Approach

Call the existing AI provider (Groq/Gemini/OpenAI) with a structured evaluation
system prompt that returns JSON scores per dimension:

Metadata

Metadata

Assignees

Labels

Projects

No projects

Milestone

No milestone

Relationships

None yet

Development

No branches or pull requests

Issue actions