audit_management_system
Audit an AI management system against ISO/IEC 42001 clauses 4-10.
LABSCO SUMMARY
Billing is opt-in, and one token covers the whole fleet.
The metering is the unusual part: leave MEOK_PAYG_KEY unset and the server runs as it always did; set it and every call draws on a prepaid balance that works across the vendor's other compliance servers too. The server code registers ten tools, each introduced in a single line, so run one against a system you know before building a review process on top of it.
What it is
A Python server for ISO 42001 compliance work: AI management-system assessment, risk management, AI lifecycle governance and certification readiness. It belongs to a fleet of compliance servers that share one pay-per-call token.
What you get
- An audit against ISO/IEC 42001 clauses 4-10, a certification-readiness checklist and the certification steps with typical timelines —
audit_management_system,create_certification_checklist,certification_timeline - Risk checks: an Annex B risk assessment, an Annex A controls check, a neural risk prediction with its aggregate insights, and a quick gap scan from a one-line description —
assess_ai_risk,check_annex_controls,predict_risk_neural,neural_insights,quick_scan - AI policy documents generated per ISO 42001, and a map from its clauses and Annex A controls to EU AI Act articles —
generate_policy_template,crosswalk_to_eu_ai_act - Optional metering: with
MEOK_PAYG_KEYset, each call draws down a prepaid balance and the tool returns a top-up URL when it reaches zero
Requirements
Install with pip install iso_42001_ai_mcp; your client then starts it with the iso_42001_ai_mcp command, over stdio. With MEOK_PAYG_KEY unset, behaviour is unchanged and nothing is deducted. With it set, calls cost £0.05 each by default and MEOK_PAYG_RATE_GBP changes that; MEOK_X402_RECEIVER enables paying in USDC on Base L2.
Setup effort
One command — pip install iso_42001_ai_mcp
USE CASES
01
Getting a first read on ISO 42001 readiness
Audit an existing management system and see which parts of the standard it does not yet answer.
02
Assessing the risk of a model you are about to ship
Run the AI risk assessment as part of the review rather than as a separate document exercise.
03
Starting a policy from something rather than nothing
Generate a template and edit it, instead of writing governance documentation from a blank page.
10 tools
Listed in project code.
assess_ai_risk
Perform ISO 42001 Annex B risk assessment for AI systems.
generate_policy_template
Generate AI policy documents per ISO 42001 requirements.
check_annex_controls
Evaluate AI system against ISO 42001 Annex A controls.
crosswalk_to_eu_ai_act
Map ISO/IEC 42001 clauses and Annex A controls to EU AI Act articles.
create_certification_checklist
Generate ISO 42001 certification readiness checklist with pass/fail.
predict_risk_neural
Neural network-based risk prediction that improves from every compliance check.
neural_insights
Get aggregate learning insights from the neural compliance model.
quick_scan
One-line system description to instant ISO 42001 gap assessment. No API key needed.
certification_timeline
Returns ISO 42001 certification steps and typical timelines. No parameters needed.
BEFORE YOU INSTALL
What to know first
Tools you get
10 tools registered in the project's code.
Requires
Python 3.10 or later
INSTALLATION
Connect your client
Add the block to your client, or run the command directly.
Server setup
Configuration file
JSON
{
"mcpServers": {
"iso-42001-ai-mcp": {
"command": "iso_42001_ai_mcp"
}
}
}pip
TERMINAL
pip install iso_42001_ai_mcp