recommendation-engine

Files (1)

• SKILL.md

SKILL.md (first lines)

# Recommendation Engine

## Goal

Turn findings into practical, mechanism-based actions that improve both AI readability and user engagement.

## Responsibilities

- map each finding to a specific root cause
- explain why the issue matters
- rank fixes by severity and recoverability
- provide action steps grounded in the evidence
- avoid generic advice

## Output format

Each recommendation should include:

- summary
- rationale
- steps
- priority
- expected impact

## Good recommendation qualities

- tied to the actual evidence
- clearly explains the mechanism
- prioritizes the highest-impact fixes
- avoids vague SEO language
- focuses on machine readability and user clarity

## Example

- finding: product facts are only generated by JavaScript
- mechanism: the page is not machine-readable in initial HTML
- fix: expose critical facts in semantic HTML or JSON-LD and keep them visible in the DOM
- priority: high

## Final implementation status

The recommendation engine now uses tighter, context-aware guidance and avoids generic suggestions that ignore page type, site purpose, or evidence quality. Recommendations are grounded in the actual audit findings and are aligned with the final validated output paths and benchmark findings in this repository.

From Haloninja2605/Adobe_Hackathon, MIT.

More from Haloninja2605/Adobe_Hackathon