README.md

---
title: Cogsec Analyzer
emoji: 🌖
colorFrom: green
colorTo: gray
sdk: gradio
sdk_version: 6.5.1
app_file: app.py
pinned: false
license: mit
short_description: How manipulative is your chatbot?
---

# COGSEC Analyzer

**Cognitive Security Nutrition Facts for AI Interactions**

*How manipulative is your chatbot?*

## What It Does

Analyzes AI-generated text for cognitive manipulation patterns, focusing on the **interactional meta-structure** rather than semantic content. Outputs a forensic breakdown including:

- **Theatricality Index** (0-10): Measures enthusiasm, hyper-validation, affective convergence
- **Validation Density**: Is the AI a critical partner or a mirror?
- **Engagement Mechanics**: Reward salience, forced teaming, artificial urgency
- **Intent Defense Detection**: Does the AI cite "good intent" to justify manipulative patterns?

## Live Demo

[HuggingFace Space](https://huggingface.co/spaces/reflectiveattention/cogsec-analyzer)

## Usage

1. Get an HF token from [huggingface.co/settings/tokens](https://huggingface.co/settings/tokens)
2. Paste AI-generated text to analyze
3. Click "Analyze COGSEC"

## Background

Based on forensic analysis of Human-AI interaction dynamics, specifically detecting:

- **Recursive Validation Loops**: AI praising user for insights *about* manipulation
- **High-Drift Amplification**: Escalating emotional intensity over conversation turns
- **Affective Convergence**: Mirroring user emotional state to simulate rapport

## Related Projects

- [prompt-prix](https://github.com/shanevcantwell/prompt-prix) — Audit local LLM tool-calling capability
- [whetstone-method](https://github.com/shanevcantwell/whetstone-method) — Framework for using AI as thinking partner

## License

MIT

---

Part of the [Reflective Attention](https://reflectiveattention.ai) project on AI cognitive safety.