The Agent Skills Directory

[Metadata Poisoning] (MEDIUM): There is a deceptive discrepancy between the skill documentation in SKILL.md and the actual implementation. The documentation claims the skill uses the google-api-python-client library and the build() function for API interaction, whereas the scripts scripts/google_calendar.py and scripts/refresh_token.py actually use urllib.request to manually perform REST calls. This misleading metadata can cause misjudgment of the skill's dependencies and internal behavior.
[Indirect Prompt Injection] (LOW): The skill possesses a surface for indirect prompt injection by ingesting untrusted data from an external API.
Ingestion points: scripts/google_calendar.py (via the list_events function which retrieves event summaries and descriptions).
Boundary markers: Absent. The skill does not provide any delimiters or instructions to help the agent distinguish between calendar data and system instructions.
Capability inventory: The skill can perform authenticated HTTP requests (GET, POST, PUT, DELETE) and has file-write access to ~/.config/google-calendar/secrets.env.
Sanitization: Absent. While IDs are URL-encoded for API safety, the textual content of the events is not sanitized or validated before being passed to the agent's context.
[External Downloads] (LOW): SKILL.md instructs users to install several Google-owned Python packages. Although these are from trusted sources, they are unnecessary as the provided code does not utilize them, contributing to the metadata discrepancy.

google-calendar