skills/mukul975/anthropic-cybersecurity-skills/implementing-email-sandboxing-with-proofpoint/Gen Agent Trust Hub
implementing-email-sandboxing-with-proofpoint
Pass
Audited by Gen Agent Trust Hub on Sep 20, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: No security vulnerabilities or malicious patterns were detected. The skill focuses on legitimate cybersecurity operations for Proofpoint TAP integration.
- [EXTERNAL_DOWNLOADS]: The Python scripts perform network operations to fetch threat intelligence data from tap-api-v2.proofpoint.com. These are official endpoints for the well-known security service, and usage is appropriate for the skill domain.
- [DATA_EXFILTRATION]: No sensitive system or user data is improperly accessed. API credentials are handled securely through environment variables or command-line arguments, preventing hardcoded secret exposure.
- [INDIRECT_PROMPT_INJECTION]: The skill includes an attack surface for indirect prompt injection via ingestion of external API data and log files. Ingestion points: scripts/agent.py and scripts/process.py (JSON data from API/logs). Boundary markers: The data is processed as structured JSON. Capability inventory: Network requests (requests.get) and local file writing. Sanitization: The scripts use explicit field extraction for metrics calculation rather than executing or directly outputting raw untrusted content.
Audit Metadata