README.md

metadata

title: PromptWork
emoji: 🔬
colorFrom: indigo
colorTo: purple
sdk: docker
pinned: false
license: mit

PromptWork: Trauma-Informed Prompt Assessment Hub

A professional tool for assessing and refining chatbot system prompts through a clinical UX lens.

Features

• Prompt Editor: Enter and refine system prompts with pre-loaded templates
• Conversation Simulator: Test prompts with realistic multi-turn conversations using Claude API
• Assessment Panel: Evaluate prompts across safety, trauma-informed design, cultural humility, and technical effectiveness
• Spotlight System: Mark and annotate important moments in conversations
• Reference Library: Quick access to clinical UX patterns and best practices
• Export: Generate assessment reports in Markdown, PDF, or JSON

Usage

1. Enter your Anthropic API key (stored only in your browser session, never logged)
2. Load or create a system prompt
3. Generate test conversations with various personas
4. Assess responses using the clinical framework
5. Spotlight important moments and add notes
6. Export your assessment report

Framework

This tool implements a trauma-informed clinical UX assessment framework including:

• Safety Rails: Crisis detection, escalation protocols, hard limits
• Youth Appropriateness: Reading level, tone calibration, parasocial risk
• Trauma-Informed Design: Agency, containment, validation practices
• Cultural Humility: Assumptions, economic sensitivity, institutional awareness
• Technical Effectiveness: Clarity, consistency, appropriate scope

Author

Jocelyn Skillman, LMHC - Clinical AI Ethics Consultant

License

MIT