geo-report
Pass
Audited by Gen Agent Trust Hub on Aug 12, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill operates as a document templating system, processing local files generated by other SEO/GEO tools into a unified report. No dangerous behaviors such as command injection, credential theft, or unauthorized network activity were found.
- [DATA_EXFILTRATION]: Although the skill is configured with network access (WebFetch), the instructions focus entirely on internal report generation and data synthesis. There is no evidence of instructions aimed at exfiltrating sensitive local data to external endpoints.
- [PROMPT_INJECTION]: The skill contains specific instructions regarding the tone and structure of the output (e.g., "Write exactly ONE paragraph", "No jargon. No hedging."). These are legitimate creative constraints for report generation and do not attempt to bypass AI safety filters or override core system instructions.
- [COMMAND_EXECUTION]: While the skill configuration allows for Bash access, the body of the skill does not contain any shell commands or scripts. It relies on the agent to use tools for reading and writing markdown files as part of its intended workflow.
Audit Metadata