brand-discovery
Pass
Audited by Gen Agent Trust Hub on Oct 5, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill instructs the agent to ingest and analyze untrusted external data sources, creating a surface for indirect prompt injection.
- Ingestion points: The skill explicitly directs the agent to analyze external content in
SKILL.md, including "Review and forum analysis (Reddit, Trustpilot, App Store, niche communities)", "Sales call recordings or transcripts", and "Support ticket themes". - Boundary markers: While the skill suggests providing "sanitized" interview notes in the output, it lacks explicit boundary markers or instructions to the agent to disregard embedded commands within the source data.
- Capability inventory: The skill consists entirely of markdown instructions and does not include any scripts, network tools, or file-writing capabilities beyond creating a discovery report, which significantly limits the potential impact of an injection.
- Sanitization: The instructions do not provide specific sanitization or escaping guidelines for the agent when processing external text strings into the final report.
Audit Metadata