driving-cursor-bugbot
Pass
Audited by Gen Agent Trust Hub on May 20, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill automates PR maintenance by interacting with the GitHub API via the
ghCLI. It performs authorized operations such as listing, replying to, and resolving review comments, which aligns with its stated purpose of managing automated review findings. - [SAFE]: The skill includes a filtering mechanism that restricts its processing to comments authored by 'cursor' or 'bugbot' users. This reduces the attack surface for indirect prompt injection by ensuring the agent only acts upon data from recognized automated sources rather than arbitrary user-contributed comments.
- [SAFE]: The repository references (e.g.,
socket-wheelhouse) and file paths mentioned in the instructions are consistent with the vendor's own environment and legitimate maintenance workflows, showing no signs of malicious exfiltration or data exposure.
Audit Metadata