CognitiveBiasLabs-MCP
# Cognitive Bias Labs - MCP Server
Official [Model Context Protocol (MCP)](https://modelcontextprotocol.io) Server for [Cognitive Bias Labs](https://cognitivebiaslabs.com).
This server acts as an AI-powered diagnostic engine, empowering Claude, Cursor, and other LLMs to detect cognitive biases in text and decision-making processes.
## Features
- **Analyze Text for Bias**: A diagnostic engine tool that analyzes arguments or decisions for logical fallacies and cognitive biases.
- **Get Supported Biases**: Lists all cognitive biases currently understood by the diagnostic engine (e.g., [Dunning-Kruger Effect](https://cognitivebiaslabs.com/dunning-kruger-test/), [Confirmation Bias](https://cognitivebiaslabs.com/confirmation-bias-test/)).
- **Test Recommendations**: Provides direct links to interactive psychology tests on Cognitive Bias Labs based on the diagnosed bias.
## Installation
### For Claude Desktop
Add the following to your `claude_desktop_config.json`:
```json
{
"mcpServers": {
"cognitivebiaslabs": {
"command": "npx",
"args": ["-y", "cognitivebiaslabs-mcp"]
}
}
}
```
### For Cursor
1. Go to Cursor Settings > Features > MCP Servers
2. Add a new server
3. Name: `CognitiveBiasLabs`
4. Type: `command`
5. Command: `npx -y cognitivebiaslabs-mcp`
## Important Links
- [Cognitive Bias Labs Website](https://cognitivebiaslabs.com)
- [All Psychology Tests](https://cognitivebiaslabs.com/tests/)
- [Cognitive Bias Index](https://cognitivebiaslabs.com/biases/)
## License
MIT License
TDQS
Scored across 3 tools
Each tool serves a distinct function: listing supported biases, analyzing text, and retrieving test recommendations for a specific bias. There is no overlap or ambiguity between them.
All tool names follow a consistent verb_noun snake_case pattern (get_supported_biases, analyze_text_for_bias, get_test_recommendation). The verbs and nouns are clear and predictable.
With only 3 tools, the set is tightly scoped to the server's core purpose of bias diagnosis. There are no redundant or superfluous tools, and the count is reasonable for a specialized diagnostic engine.
The tools cover the main workflow: list, analyze, and get test details. A minor gap is the lack of a general 'get bias details' endpoint, but for the apparent purpose of recommending tests, the coverage is sufficient.