gpt-pro
Pass
Audited by Gen Agent Trust Hub on Jul 15, 2026
Risk Level: SAFECOMMAND_EXECUTIONDATA_EXFILTRATION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructs the agent to use standard local system utilities such as
tar,grep,mkdir, andcatto organize files and create a compressed archive for the user. These commands are executed locally for legitimate administrative purposes. - [DATA_EXFILTRATION]: While the primary function of the skill is to facilitate the transfer of project context to an external model (ChatGPT Pro), several security controls are present:
- Sanitization: The instructions mandate the removal of sensitive information (private keys, mnemonics, API keys, passwords) from configuration files before they are added to the package.
- Manual Control: The agent does not send data automatically. Instead, it places a prompt in the system clipboard and saves an archive to the user's Documents folder, requiring the user to manually paste and upload the files to OpenAI (a well-known service).
- Transparency: The skill clearly communicates which files are being collected and where they are being stored.
- [EXTERNAL_DOWNLOADS]: The README provides standard instructions for users to clone the repository from GitHub. This is documented as a standard distribution method from a trusted-style repository structure.
Audit Metadata