iso-27001-ai-mcp 1.0.7

AI-powered iso 27001 ai MCP server for agents. Supports audit isms, risk assessment, gap analysis. By MEOK AI Labs.

ISO 27001 AI MCP

ISO/IEC 27001:2022 compliance assessment for AI systems — 93 Annex A controls across 4 themes, ISO 27005 risk assessment, Statement of Applicability generation, incident classification, and ISO 42001 bridge.

What This Does

ISO/IEC 27001:2022 is the international standard for Information Security Management Systems (ISMS). Its Annex A contains 93 controls organized into 4 themes: Organizational (37 controls), People (8), Physical (14), and Technological (34). Certification requires demonstrating that your ISMS meets clauses 4–10 and that your Statement of Applicability (SoA) addresses all relevant controls.

For AI systems, ISO 27001 is foundational — but it needs extension. This server audits your ISMS against all 93 controls, performs ISO 27005 risk assessments with AI-specific threat scenarios, generates gap analyses with prioritized remediation roadmaps, produces SoAs, classifies security incidents, and bridges to ISO 42001 for AI-specific governance.

Quick Start

npx meok-setup --pack governance

Tools

Tool	Description	Parameters
audit_isms	Audits your ISMS against all 93 Annex A controls. Returns per-theme compliance status (PASS/PARTIAL/FAIL), gap identification, critical gap flagging for high-priority controls (A.5.1, A.5.15, A.8.5, A.8.24, etc.), and certification readiness assessment.	organization_context, scope_description, controls_implemented
risk_assessment	Performs ISO 27005 information security risk assessment. Evaluates 10 AI-relevant threat categories (adversarial attacks, model theft, training data breach, supply chain compromise, insider threat, etc.), calculates likelihood × impact risk scores, and produces a treatment plan with specific Annex A control recommendations.	system_description, assets, threat_scenarios, existing_controls
gap_analysis	Compares your current controls to ISO 27001 requirements. Supports three targets: "full" (all 93 controls), "core" (critical subset), or "ai-focused" (AI-relevant controls only). Returns prioritized remediation roadmap in 3 phases: Critical (0–30 days), Standard (30–90 days), Remaining (90–180 days).	current_controls, target_certification, focus_themes
crosswalk_to_ai	Maps ISO 27001 controls to ISO 42001 AI-specific requirements. Shows how existing ISMS controls extend to AI governance (model security, training data protection, AI incident management) and identifies where AI-specific controls are needed.	controls, focus_area
generate_soa	Generates a Statement of Applicability per clause 6.1.3(d). Documents all 93 controls as Implemented, Excluded (with justification), or Not Yet Addressed. Required artifact for ISO 27001 certification audits.	organization_name, controls_implemented, controls_excluded, exclusion_justifications
incident_classification	Classifies security incidents per controls A.5.24–A.5.28. Determines severity (LOW→CRITICAL), priority (P1–P3), notification requirements, and response procedures. Includes AI-specific incident categories: adversarial attacks, data poisoning, model theft, prompt injection, bias incidents.	incident_description, affected_assets, detection_method, data_breach, ai_system_involved

Usage Examples

Audit your AI company's ISMS

Use the audit_isms tool with:
  organization_context: "AI startup with 150 employees building ML models for healthcare diagnostics. Uses AWS for infrastructure, processes patient data, has a small security team."
  scope_description: "All AI systems, ML pipelines, patient data processing, cloud infrastructure, and development environments"
  controls_implemented: ["A.5.1", "A.5.9", "A.5.12", "A.5.15", "A.5.24", "A.6.3", "A.8.5", "A.8.7", "A.8.8", "A.8.15", "A.8.24"]

Expected output: Overall coverage ~12% (11/93 controls). Critical gaps flagged in A.5.34 (PII protection), A.8.12 (data leakage prevention), A.8.25 (secure SDLC). Certification NOT ready — 82 gaps to address.

Assess risk for your ML pipeline

Use the risk_assessment tool with:
  system_description: "Production ML pipeline processing financial data for fraud detection. Uses gradient boosting models trained on 10M+ transaction records. Served via REST API with 99.9% SLA."
  assets: ["training data", "ML model weights", "feature store", "API keys", "model serving infrastructure", "customer transaction data"]
  existing_controls: ["A.5.15", "A.8.5", "A.8.15", "A.8.24"]

Expected output: Risk register with 10 threat assessments. Highest risks: training data breach (likely × high = risk score 16), model theft (possible × critical = 20). Treatment plan recommends implementing A.8.12, A.8.16, A.5.12 for the highest-priority gaps.

Generate a gap analysis for AI-focused certification

Use the gap_analysis tool with:
  current_controls: ["A.5.1", "A.5.2", "A.5.15", "A.5.24", "A.6.3", "A.8.5", "A.8.7", "A.8.15", "A.8.16", "A.8.24"]
  target_certification: "ai-focused"

Expected output: 25 AI-critical controls evaluated. ~40% coverage. Phase 1 critical gaps: A.8.8 (vulnerability management), A.8.12 (data leakage), A.5.34 (PII protection). Estimated remediation: 3–6 months.

Classify a security incident involving AI

Use the incident_classification tool with:
  incident_description: "Adversarial evasion attack detected on production fraud detection model. Attackers crafted transactions that bypassed ML model scoring. Approximately 200 fraudulent transactions processed before detection."
  affected_assets: ["fraud detection model", "transaction processing system", "customer accounts"]
  detection_method: "automated"
  data_breach: false
  ai_system_involved: true

Expected output: Severity HIGH (P1), AI incident category: adversarial attack. Immediate response: activate incident plan (A.5.24), contain model, preserve inference logs, assess model integrity. AI-specific controls: A.5.7, A.8.8, A.8.16.

Installation

Claude Desktop

Add to claude_desktop_config.json:

{
  "mcpServers": {
    "iso-27001-ai": {
      "command": "npx",
      "args": ["-y", "meok-iso-27001-ai-mcp"]
    }
  }
}

Or install via Smithery:

npx smithery mcp add nicholastempleman/iso-27001-ai-mcp

Cursor

Add to .cursor/mcp.json:

{
  "mcpServers": {
    "iso-27001-ai": {
      "command": "npx",
      "args": ["-y", "meok-iso-27001-ai-mcp"]
    }
  }
}

VS Code

Add to .vscode/mcp.json:

{
  "servers": {
    "iso-27001-ai": {
      "command": "npx",
      "args": ["-y", "meok-iso-27001-ai-mcp"]
    }
  }
}

pip

pip install meok-iso-27001-ai-mcp

Related Servers

Server	Purpose
iso-42001-ai	AI management system — Annex A controls and Annex B risk assessment
gdpr-compliance-ai	GDPR DPIA, data subject rights, breach notification
eu-ai-act-compliance	EU AI Act risk classification and Annex IV documentation
soc2-compliance-ai	SOC 2 Trust Service Criteria and control matrix
csoai-governance-crosswalk	12 compliance frameworks mapped through 52 articles

Pricing

• Free tier: 10 calls/day per tool
• Pro: £79/mo — unlimited calls + cryptographically signed compliance attestations

License

MIT © MEOK AI Labs

💸 Try MEOK in 30 seconds — instant buy ladder

Tier	Price	What you get	Stripe
Smoke test	£1	Signed sample MCP-Hardening report + Article 50 PDF	https://buy.stripe.com/dRmcN75ScdQS7oh1Uc8k90U
Quick Kit	£9	EU AI Act Article 50 implementation guide (C2PA + EU-Icon)	https://buy.stripe.com/cNi00la8s1460ZT0Q88k90V
Founder Call	£29	30-min 1-on-1 with the founder	https://buy.stripe.com/8x228ta8s6oqbExaqI8k90W

Refundable. UK Stripe — VAT-clean. Builds on the 81-MCP MEOK fleet. Verify any signed report at https://meok.ai/verify.