ai-visibility-prompt-research
Pass
Audited by Gen Agent Trust Hub on Aug 31, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill operates as a high-level research assistant and does not execute any code, scripts, or external commands.
- [DATA_EXPOSURE]: The skill requires the user to provide first-party sales call and pre-sales transcripts. This is a common requirement for marketing research, and the skill includes instructions to the agent to withhold the brand and maintain separate categories, providing a layer of human review before results are finalized.
- [EXTERNAL_DOWNLOADS]: The skill references one external URL (episteme.airfleet.co) as an example of where a user might upload the resulting research file. This is a legitimate marketing technology service and is not used for automated data exfiltration within the skill's logic.
- [PROMPT_INJECTION]: No prompt injection patterns were detected. The instructions are clearly defined and focus on maintaining research integrity (e.g., 'withhold the brand until the final pass').
- [INDIRECT_PROMPT_INJECTION]: The skill has a surface for indirect prompt injection as it processes external content from the 'open web' (forums, subreddits, etc.). However, it includes mandatory 'Noise filters' to strip unrelated or malicious content and requires a 'Validation before handoff' step where the human user reviews the output before use.
Audit Metadata