The Agent Skills Directory

[SAFE]: The skill implements a robust 'redaction-by-default' architecture where personal data is masked in logs and agent context, only revealing raw values at the moment of form entry.
[SAFE]: Data is stored locally in the user's home directory or iCloud Drive with strict 0600 file permissions, ensuring no data touches external servers or the chat history unless explicitly required.
[EXTERNAL_DOWNLOADS]: The skill references a shell script hosted on the author's GitHub repository (https://raw.githubusercontent.com/leeguooooo/profile-use/main/install-app.sh).
[TRUST-SCOPE-RULE]: This download is used to install a native macOS GUI for the skill and is documented as a legitimate part of the installation process for this author's toolset.
[COMMAND_EXECUTION]: The Python script and Swift app execute shell commands to interact with rbw (a Bitwarden CLI) and perform file operations.
These operations are used strictly for local vault access and profile management, following the principle of least privilege.
[CREDENTIALS_SAFE]: Password management is handled through a Bitwarden/Vaultwarden integration (rbw). The skill never asks for or stores the master password, requiring the user to unlock the vault themselves in their own terminal environment.
[DATA_EXFILTRATION]: No network exfiltration patterns were detected. The scripts contain no networking code (curl/requests) except for the installation script mentioned above.

profile-use