The Agent Skills Directory

[CREDENTIALS_UNSAFE]: The skill programmatically extracts the CODEMAGIC_API_TOKEN from the user's shell profile (~/.zshrc) using grep and cut.
[COMMAND_EXECUTION]: Employs powerful CLI tools including kubectl, gcloud, and security to modify sensitive infrastructure state, cloud secrets, and the system-level macOS Keychain.
[DATA_EXFILTRATION]: Performs broad, recursive searches (grep -r) across the entire project directory to locate and display the names and values of secrets stored in .env, .yaml, and source code files.
[COMMAND_EXECUTION]: Uses sed -i to modify local environment files. This pattern is vulnerable to command injection if the new key value contains characters that break the sed delimiter or syntax.
[COMMAND_EXECUTION]: Executes dynamic Python snippets via python3 -c to parse and process JSON data retrieved from cloud services and external APIs.
[CREDENTIALS_UNSAFE]: Accesses and modifies multiple local .env files, including those in build artifacts and backend directories, which often contain production secrets.

rotate-key