MCP.so
登录
A

AIR Blackbox MCP Server

@airblackbox

关于 AIR Blackbox MCP Server

EU AI Act compliance scanning for Python AI agents. 10 tools that scan, analyze, and fix compliance gaps across LangChain, CrewAI, AutoGen, OpenAI, and RAG pipelines. Checks 6 EU AI Act articles with HMAC-SHA256 tamper-evident audit trails. pip install air-blackbox-mcp. August 20

基本信息

分类

AI 与智能体

传输方式

stdio

发布者

airblackbox

提交者

Jason Shotwell

配置

使用下面的配置,将此服务器添加到你的 MCP 客户端。

{
  "mcpServers": {
    "air-blackbox": {
      "command": "python",
      "args": [
        "-m",
        "air_blackbox_mcp"
      ]
    }
  }
}

工具

未检测到工具

工具是从 README 中自动提取的。维护者可以在 ## Tools 标题下列出工具,即可填充这部分内容。

概览

What is AIR Blackbox MCP Server?

EU AI Act compliance scanning for Claude Desktop, Claude Code, Cursor, and any MCP-compatible client. Unlike other compliance scanners that only report problems, AIR Blackbox also remediates — generating working code fixes, trust layer integrations, GDPR compliance checks, bias analysis, and full compliance reports.

How to use AIR Blackbox MCP Server?

Install via pip install air-blackbox-mcp (basic, 10 tools) or pip install air-blackbox-mcp[full] (all 14 tools). Configure in Claude Desktop by editing claude_desktop_config.json, or in Claude Code/Cursor via .cursor/mcp.json or .claude/mcp.json. Users invoke tools through natural language prompts such as "Scan this code for EU AI Act compliance".

Key features of AIR Blackbox MCP Server

  • 14 tools: 10 base + 4 SDK-powered scanning, analysis, remediation, documentation
  • Scans Python code for all 6 EU AI Act articles using 18+ regex patterns
  • Remediation generates working code fixes and trust layer integrations
  • Optional deep analysis via local fine-tuned Olama model (air-compliance-v2)
  • Smart fallback: full SDK engine first, lightweight built-in scanner otherwise
  • Detects prompt injection attacks with 15 weighted patterns
  • Optionally supports GDPR scanning, bias analysis, action validation, and compliance history

Use cases of AIR Blackbox MCP Server

  • Scan Python code or project for EU AI Act compliance in any MCP client
  • Add a trust layer to a LangChain, CrewAI, or AutoGen agent
  • Generate a full compliance report with article-specific fix recommendations
  • Validate agent actions before execution (Article 14) with consent gates
  • Track compliance trends and export audit trails over time

FAQ from AIR Blackbox MCP Server

What tools are included in the basic installation?

The basic installation provides 10 tools: scan_code, scan_file, scan_project, analyze_with_model, check_injection, classify_risk, add_trust_layer, suggest_fix, explain_article, and generate_compliance_report.

What does the full SDK (pip install air-blackbox-mcp[full]) add?

It unlocks 4 additional tools: scan_gdpr (GDPR compliance), scan_bias (bias and fairness analysis), validate_action (pre-execution approval gates), and compliance_history (past scans, trends, scores).

Does this server require Ollama or any external model?

No, Ollama is optional. The server works standalone with pip install air-blackbox-mcp using its built-in scanner. The fine-tuned compliance model (air-compliance-v2) is only used by the analyze_with_model tool when locally available.

What Python frameworks does it support?

It supports LangChain, CrewAI, AutoGen, OpenAI, Haystack, LlamaIndex, Semantic Kernel, Google ADK, Claude Agent SDK, and generic RAG pipelines.

How does the server handle the SDK not being installed?

It uses a smart fallback pattern: first tries the air-blackbox SDK (v1.6.0+), and if not found, gracefully falls back to the lightweight built-in scanner — no breaking changes or configuration required.

评论

AI 与智能体 分类下的更多 MCP 服务器