email-campaign

Pass

Audited by Gen Agent Trust Hub on May 4, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill demonstrates a surface for indirect prompt injection by interpolating user-controlled input ($ARGUMENTS) directly into the agent's instructions in SKILL.md and meta.json without using boundary markers or sanitization.
  • Ingestion points: The $ARGUMENTS variable in SKILL.md (Step 5) and meta.json (Step 3).
  • Boundary markers: Absent; user input is not delimited from system instructions.
  • Capability inventory: The skill is limited to text generation (emails) and Midjourney prompt generation; it has no file-write, network, or shell execution capabilities.
  • Sanitization: Absent; the skill does not validate or filter the content of $ARGUMENTS.
  • [SAFE]: The skill's file access is limited to its own configuration files (meta.json, brand-voice.md) and project-specific logic files (core/global-constraints.json, core/marketing-triggers.json). No access to sensitive system directories or credentials was detected.
Audit Metadata
Risk Level
SAFE
Analyzed
May 4, 2026, 02:11 PM
Security Audit — agent-trust-hub — email-campaign