The Agent Skills Directory

[SAFE]: The skill provides security-conscious guidance for managing API keys and service accounts. It specifically warns against printing secrets and recommends hardening file permissions (e.g., chmod 0600 for auth.json) to prevent unauthorized access.
[COMMAND_EXECUTION]: Provides examples of using sudo to switch to a service account (joelclaw) and running Node.js scripts via heredocs to verify authentication status. These are documented as standard operational procedures for the project infrastructure and do not represent malicious privilege escalation.
[DATA_EXFILTRATION]: While the skill involves processing data via LLMs, it establishes clear boundaries for data handling. It encourages the use of service-account OAuth and provides status-check methods that confirm authentication status without exposing or exfiltrating raw API keys.

pi-inngest