The Agent Skills Directory

[COMMAND_EXECUTION]: Multiple scripts within the skill utilize subprocess.run to execute external commands. Specifically, get_auth.py executes minis-browser-use for browser automation and minis-open for opening URLs, while the image generation scripts execute get_auth.py as a subprocess.
[CREDENTIALS_UNSAFE]: The skill is designed to automatically retrieve and store a ChatGPT accessToken from the user's browser session. This sensitive credential is saved to ~/.chatgpt_auth.json. Although the skill attempts to secure this file by setting permissions to 600, the programmatic extraction of session tokens is a high-risk operation.
[DATA_EXFILTRATION]: By design, the skill programmatically accesses the user's ChatGPT session token. While the token is intended for use with the chatgpt.com API, the capability to extract such sensitive session-level data could be misused for unauthorized access if the environment or toolchain is compromised.

codex-image