The Agent Skills Directory

[SAFE]: No security issues detected. The skill's instructions involve standard configuration tasks for the Pi Agent, such as updating settings.json and models.json. While it mentions accessing ~/.pi/agent/auth.json to verify credentials, this is a local operation intended for configuration purposes and does not involve any external data transmission. The presence of disable-model-invocation: true in the frontmatter further limits the agent's ability to perform unauthorized actions while following these instructions.

pi-custom-model