customer-research
Pass
Audited by Gen Agent Trust Hub on Aug 25, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill instructions and reference guides focus on legitimate customer research methodologies. The use of external platforms for research is documented as part of the skill's primary function and follows standard industry practices for data collection and analysis.
- [INDIRECT_PROMPT_INJECTION]: The skill defines a surface for processing untrusted data from various external sources.
- Ingestion points: Raw research material including transcripts, surveys, support tickets, and online content from platforms like Reddit, G2, and LinkedIn as described in SKILL.md and references/source-guides.md.
- Boundary markers: Present. The skill mandates the use of specific extraction frameworks (e.g., Jobs to Be Done, Pain Points) and synthesis templates (e.g., Persona Structure) which guide the agent to treat input as data to be categorized rather than instructions to be executed.
- Capability inventory: The skill focuses on analytical tasks such as clustering, scoring, and persona generation based on the provided data.
- Sanitization: Absent. The instructions do not explicitly direct the agent to filter or escape potentially malicious instructions embedded in the source material, though the structured synthesis process provides an inherent layer of protection.
Audit Metadata