azure-diagrams

Fail

Audited by Gen Agent Trust Hub on Feb 17, 2026

Risk Level: CRITICALDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
  • DATA_EXFILTRATION (LOW): The skill is designed to transmit user-provided Azure infrastructure details (such as IP addresses, resource names, and network topology) to an external API (https://app.eraser.io). While this is the primary intended purpose of the skill, users should be aware that sensitive configuration data is being sent to a third-party service.
  • PROMPT_INJECTION (LOW): The skill is vulnerable to indirect prompt injection (Category 8). It parses untrusted data sources including ARM templates, CLI output, and natural language descriptions. An attacker could embed malicious instructions within these data sources to attempt to hijack the agent's behavior or exfiltrate unrelated local files by disguising them as 'resources' to be diagrammed.
  • COMMAND_EXECUTION (SAFE): The skill requests permission for Bash(curl:*) to perform its API request. While the wildcard permission is broad, the usage is constrained within the instructions to a specific endpoint (app.eraser.io) for rendering diagrams.
Recommendations
  • Contains 1 malicious URL(s) - DO NOT USE
Audit Metadata
Risk Level
CRITICAL
Analyzed
Feb 17, 2026, 04:52 PM