The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes local Python scripts to facilitate authentication and launch system browsers.
Evidence: scripts/open_login_page.py utilizes subprocess.run to call standard system utilities such as open, xdg-open, and cmd /c start. These calls are restricted to hardcoded vendor URLs and do not incorporate untrusted input.
[SAFE]: The skill implements robust security practices for handling sensitive credentials locally.
Evidence: The skill enforces restricted filesystem permissions (0600) for its local configuration file ~/.chanjing/credentials.json to prevent unauthorized access.
Evidence: Instructions in SKILL.md explicitly forbid the AI from requesting secrets via chat, mandating the use of local CLI commands instead.
Evidence: Network activity is confined to the official vendor API (open-api.chanjing.cc) for the sole purpose of token management.

chanjing-credentials-guard