quick_scanLABSCO SUMMARY
Named to the two rules that ask for the evidence.
Fairness tooling is easy to find and hard to point at a specific obligation; this one names EU AI Act Article 10 and NYC Local Law 144 up front, which tells you whether it is the right shape for your audit before you install anything. The free self-hosted build gives you the assessment; signing the result is what the paid tiers add.
What it is
A Python server for AI bias detection and fairness assessment. It is aimed at the two regimes that actually demand this evidence — EU AI Act Article 10 and New York City Local Law 144 — so the output is meant to be shown to someone, not just read once.
What you get
Five tools: quick_scan for a first pass, detect_bias for demographic bias and stereotyping in text, fairness_metrics for the numbers from prediction data, mitigation_recommendations for how to fix a given kind of bias, and regulatory_check against EU AI Act Article 10 and the NIST AI RMF.
Requirements
pip install bias_detection_mcp, or point your client at uvx with the argument bias-detection-mcp (package version 1.2.7, stdio transport). Self-hosting is MIT-licensed and free; the paid tiers add HMAC-signed attestations verifiable at verify.meok.ai.
Setup effort
One command — pip install bias_detection_mcp
USE CASES
01
Getting a first read on a model you inherited
Run the quick scan before committing to a full assessment.
02
Producing a bias assessment for an audit
Run the detection pass and keep the output as the record.
03
Backing a fairness claim with numbers
Pull the metrics rather than describing the result qualitatively.
5 tools
Listed in project docs. Your client may show a different list.
detect_biasfairness_metricsmitigation_recommendationsregulatory_checkBEFORE YOU INSTALL
What to know first
Tools you get
5 tools named in the project's documentation.
Requires
uvx on PATH
Where it runs
The documented client runs the server on your machine.
INSTALLATION
Connect your client
Server setup
TERMINAL
pip install bias_detection_mcpMCP Clients
Claude Desktop
CLAUDE DESKTOP · JSON
{
"mcpServers": {
"bias-detection-mcp": {
"command": "uvx",
"args": [
"bias-detection-mcp"
]
}
}
}
