llm-issue-discovery
Pass
Audited by Gen Agent Trust Hub on Apr 23, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: No security issues detected. The skill functions purely as a prompt for reasoning and analysis within the agent context.- [DATA_EXPOSURE]: The skill involves the user providing LLM logs (prompts and responses) for analysis. While logs may contain sensitive data, the skill does not automate data collection or exfiltration, relying on the user to provide the context manually.- [EXTERNAL_DOWNLOADS]: The instructions include informational links to the author's website (latitude.so). These are documented as references to the vendor's platform and do not involve automated background operations or unauthorized downloads.- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data (external LLM logs). The analysis confirms the skill has no available tools or capabilities (no network, file, or code execution access), meaning that potential instructions embedded in the logs can only affect the textual analysis in the final report without impacting system security.
Audit Metadata