The Agent Skills Directory

Manual Command Generation: The skill generates paste-ready shell commands (e.g., for Subversion or Git) instead of executing them directly. This design serves as a security control by ensuring a human reviews and executes the operations under their own credentials.
External Data Ingestion: The skill fetches data from project-specific URLs and public PGP keyservers to validate release manager keys. It includes explicit instructions to treat this external content as untrusted data, which is a best practice for mitigating potential indirect prompt injection attempts.
Local Execution for Configuration Checks: A pre-flight check executes a local Python module to verify project setup. This is a standard operational check within the skill's framework to ensure required configurations are present before processing.
Privilege and Credential Management: The skill explicitly states it does not access private keys or passphrases and does not perform automated commits, adhering to the principle of least privilege.

keys-sync