report-evidence-citation

Pass

Audited by Gen Agent Trust Hub on Jul 14, 2026

Risk Level: SAFE
Full Analysis
  • [DATA_EXFILTRATION]: The skill defines a standard reporting workflow that involves sending data to a specific API endpoint (POST /api/issues/{id}/comments). This appears to be an internal platform capability for report submission rather than an exfiltration attempt to an external or malicious domain.
  • [CREDENTIALS_UNSAFE]: The skill includes a proactive security measure by instructing the agent to ensure no plain-text API keys are exposed, requiring masking (first 8 characters followed by ***) if they appear in outputs.
  • [INDIRECT_PROMPT_INJECTION]: The skill defines a surface where untrusted data (external research, URLs) is processed and subsequently posted to an API. The instructions mitigate this risk by enforcing strict structural boundaries, such as mandatory fact/opinion labeling and a verification checklist that must be passed before results are reported.
  • [PROMPT_INJECTION]: No attempts to override safety filters or bypass system instructions were detected. The instructions are focused on adherence to company reporting standards.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 14, 2026, 04:31 PM
Security Audit — agent-trust-hub — report-evidence-citation