classify_annex3

Instructions

Implementation Reference

• src/index.ts:247-279 (registration)

  Tool registration of classify_annex3 as an MCP tool, defining its name, description, and inputSchema (company, industry, features, useCase, euExposure).

  {
    name: "classify_annex3",
    description:
      "Deep Annex III classification using Claude (Anthropic) with full EU AI Act regulatory context. Maps your product against all 8 Annex III high-risk categories + Article 50 transparency + GPAI Article 53 + GPAI provider/deployer reasoning. Returns 5 priority remediation actions tailored to your stack. ~60 seconds (LLM-powered).",
    inputSchema: {
      type: "object",
      properties: {
        company: {
          type: "string",
          description: "Company name being classified.",
        },
        industry: {
          type: "string",
          description: "Industry/vertical (e.g., 'HR-tech', 'fintech', 'healthtech', 'productivity', 'voice-AI').",
        },
        features: {
          type: "array",
          items: { type: "string" },
          description: "AI features in the product. Examples: 'content-generation', 'scoring', 'decision-making', 'api-consumer', 'own-model', 'recommendation', 'classification'.",
        },
        useCase: {
          type: "string",
          description: "Plain-English description of what the AI does and who it serves. Min 20 chars.",
        },
        euExposure: {
          type: "string",
          enum: ["eu-customers-output", "eu-employees-only", "no-eu", "considering-eu"],
          description: "EU exposure level. 'eu-customers-output' = serves EU customers with AI-influenced outputs. 'eu-employees-only' = internal tools used by EU employees only. 'no-eu' = no EU footprint. 'considering-eu' = planning EU expansion.",
        },
      },
      required: ["company", "industry", "features", "useCase", "euExposure"],
    },
  },

• src/index.ts:322-355 (handler)

  Handler for classify_annex3 tool call. Makes a POST request to https://eucomplyhub.com/api/annex3-classify with the provided arguments, returns the JSON response, or an error message if the API call fails.

  if (name === "classify_annex3") {
    try {
      const response = await fetch(`${API_BASE}/api/annex3-classify`, {
        method: "POST",
        headers: { "Content-Type": "application/json" },
        body: JSON.stringify(args),
      });
  
      if (!response.ok) {
        const errText = await response.text();
        throw new Error(`Classifier API returned ${response.status}: ${errText.slice(0, 200)}`);
      }
  
      const result = await response.json();
      return {
        content: [
          {
            type: "text",
            text: JSON.stringify(result, null, 2),
          },
        ],
      };
    } catch (err: any) {
      return {
        content: [
          {
            type: "text",
            text: `Error calling classifier: ${err.message}. Try the web version: https://eucomplyhub.com/annex3`,
          },
        ],
        isError: true,
      };
    }
  }

• src/index.ts:1-17 (helper)

  File header describing classify_annex3 as a 'Deep Annex III classification (Claude-powered)' tool that maps to https://eucomplyhub.com/api/annex3-classify.

  #!/usr/bin/env node
  
  /**
   * @eucomplyhub/mcp-eu-ai-act
   *
   * MCP (Model Context Protocol) server exposing eucomplyhub.com's free EU AI Act
   * compliance classifiers to AI assistants (Claude Desktop, Cursor, Windsurf, etc.)
   *
   * Two tools exposed:
   *   1. classify_annex3 — Deep Annex III classification (Claude-powered)
   *      Maps to: https://eucomplyhub.com/api/annex3-classify
   *   2. quick_risk_class — 30-second multi-jurisdiction risk classifier
   *      (deterministic, no LLM call — same logic as eucomplyhub.com/risk-class)
   *
   * @license MIT
   * @see https://eucomplyhub.com
   */