The Agent Skills Directory

[SAFE]: The skill implements legitimate Google Calendar management functionality using official Google API libraries. No malicious patterns or data exfiltration attempts were detected.
[PROMPT_INJECTION]: The skill identifies a potential surface for indirect prompt injection, which is inherent to tools that process external user-controlled content like calendar events.
Ingestion points: The skill retrieves calendar event summaries and descriptions via the list_events and get_event functions in scripts/google-calendar.py.
Boundary markers: None identified. Instructions do not explicitly tell the agent to ignore instructions embedded within event data.
Capability inventory: The agent can execute scripts/google-calendar.py via the Bash tool and has file system access for configuration storage.
Sanitization: Event content is presented to the agent without additional filtering, relying on the underlying model's guardrails to handle potentially malicious instructions in calendar data.
[CREDENTIALS_UNSAFE]: The skill handles authentication securely. It uses OAuth 2.0 and the system keyring (keyring package) to store sensitive tokens, avoiding hardcoded secrets or unsafe storage practices.

google-calendar