README.md
EU AI Act compliance scanning for Claude Desktop and Cursor
AIR Blackbox MCP Server
EU AI Act compliance scanning for Claude Desktop, 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, and full compliance reports.
10 Tools
| Tier | Tool | What it does |
|---|---|---|
| Scanning | scan_code |
Scan Python code string for all 6 EU AI Act articles |
| Scanning | scan_file |
Read and scan a single Python file |
| Scanning | scan_project |
Recursively scan all .py files in a directory |
| Analysis | analyze_with_model |
Deep analysis via local fine-tuned model (Ollama) |
| Analysis | check_injection |
Detect prompt injection attacks (15 patterns) |
| Analysis | classify_risk |
Classify tools by EU AI Act risk level |
| Remediation | add_trust_layer |
Generate trust layer integration code |
| Remediation | suggest_fix |
Get article-specific fix recommendations |
| Documentation | explain_article |
Technical explanation of EU AI Act articles |
| Documentation | generate_compliance_report |
Full markdown compliance report |
Supported Frameworks
LangChain, CrewAI, AutoGen, OpenAI, Haystack, LlamaIndex, Semantic Kernel, and generic RAG pipelines.
Install
pip install air-blackbox-mcp
Claude Desktop Setup
Edit ~/Library/Application Support/Claude/claude_desktop_config.json:
{
"mcpServers": {
"air-blackbox": {
"command": "python3",
"args": ["-m", "air_blackbox_mcp"]
}
}
}
Restart Claude Desktop. The 10 tools will appear automatically.
Cursor Setup
Add to .cursor/mcp.json in your project:
{
"mcpServers": {
"air-blackbox": {
"command": "python3",
"args": ["-m", "air_blackbox_mcp"]
}
}
}
Usage Examples
In Claude Desktop or Cursor, just ask:
- "Scan this code for EU AI Act compliance"
- "Add a trust layer to this LangChain agent"
- "Check this text for prompt injection"
- "What does Article 12 require?"
- "Generate a compliance report for my project at ~/myproject"
- "Classify the risk level of
send_email"
Optional: Deep Analysis with Ollama
For AI-powered analysis beyond regex patterns:
# Install Ollama
brew install ollama
# Pull the fine-tuned compliance model
ollama pull air-compliance-v2
# The analyze_with_model tool will automatically use it
What Makes This Different
Other MCP compliance tools only scan. AIR Blackbox:
- Scans — 18+ regex patterns across 6 EU AI Act articles
- Analyzes — local fine-tuned model catches what regex misses
- Remediates — generates working code to fix findings
- Verifies — checks for prompt injection with 15 weighted patterns
- Classifies — maps tools to EU AI Act risk levels
- Documents — produces full compliance reports
Links
- AIR Blackbox Scanner — the CLI scanner
- airblackbox.ai — project homepage
- EU AI Act — the regulation
Tools (10)
scan_codeScan Python code string for all 6 EU AI Act articlesscan_fileRead and scan a single Python filescan_projectRecursively scan all .py files in a directoryanalyze_with_modelDeep analysis via local fine-tuned model (Ollama)check_injectionDetect prompt injection attacks (15 patterns)classify_riskClassify tools by EU AI Act risk leveladd_trust_layerGenerate trust layer integration codesuggest_fixGet article-specific fix recommendationsexplain_articleTechnical explanation of EU AI Act articlesgenerate_compliance_reportFull markdown compliance reportConfiguration
claude_desktop_config.json
{"mcpServers": {"air-blackbox": {"command": "python3", "args": ["-m", "air_blackbox_mcp"]}}}Try it
→Scan this code for EU AI Act compliance
→Add a trust layer to this LangChain agent
→Check this text for prompt injection
→What does Article 12 require?
→Generate a compliance report for my project at ~/myproject