AuditEngine — Development skill for Claude Code

Ethics Engine

A web application for psychometric assessment of Large Language Models. Ethics Engine applies validated psychological instruments (authoritarianism scales, moral foundations questionnaires, personality inventories) to LLMs, measuring how models respond across different ideological framings.

**Live Demo**: [ethicsengine.eduba.io](https://ethicsengine.eduba.io)

Features

Text Assessment

• 10 Built-in Psychometric Scales: RWA, LWA, MFQ, NFC, BFI-10, SDO-7, RSES, GSE, LOT-R, and more
• 7 LLM Providers: OpenAI, Anthropic, xAI, Google Gemini, DeepSeek, Groq, and self-hosted Llama
• Custom Personas: Create manual personas or use AI to generate persona variations
• Custom Scales: Upload your own psychometric instruments with guided validation

Visual Stimulus Assessment

• Image-based Assessment: Upload images (Rorschach inkblots, visual stimuli) for multimodal analysis
• Vision-Capable Models: Automatically filters to models that support image input (GPT-5.x, GPT-4o, Claude 4.x, Gemini 2.x, etc.)
• Custom Personas: Create or AI-generate personas specifically for visual assessment
• Open-ended Responses: Get detailed model interpretations of visual stimuli

Common Features

• Real-time Progress: Live updates during assessment runs
• Export Results: Download data as CSV or JSON for analysis

Tech Stack

• Frontend: Next.js 14+ (App Router, TypeScript, Tailwind CSS)
• Backend: FastAPI (Python, async-native)
• Deployment: Vercel (frontend) + Railway/Fly.io (backend)

Quick Start

Prerequisites

• Node.js 18+
• Python 3.11+
• API keys for at least one LLM provider

Backend Setup

cd backend
python -m venv venv

# Windows
venv\Scripts\activate
# macOS/Linux
source venv/bin/activate

pip install -r requirements.txt
uvicorn app.main:app --reload

Backend runs at `http://localhost:8000`

Frontend Setup

cd frontend
npm install
npm run dev

Frontend runs at `