The Agent Skills Directory

[SAFE]: The skill ingests data from the Google Calendar API to organize meeting information. While this data is externally controlled and untrusted, the skill possesses no dangerous capabilities such as file system writing, network requests, or command execution, which mitigates any risk of indirect prompt injection. Ingestion points: google_calendar tool (SKILL.md); Boundary markers: Absent; Capability inventory: None; Sanitization: Absent.\n- [SAFE]: The instructions include a specific directive to parse data directly rather than using scripts or bash commands. These are behavioral constraints designed for efficient processing within the model's context and do not constitute a prompt injection attempt to bypass safety filters or security protocols.

meeting-list