gemini
Pass
Audited by Gen Agent Trust Hub on Oct 30, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from an external source, creating a potential surface for indirect prompt injection.\n
- Ingestion points: Data is fetched from the Gemini Predictions API (api.gemini.com) via the get_events, get_event, search_markets, and get_todays_events commands. This includes event titles, participant names, and market descriptions.\n
- Boundary markers: While the skill documentation explicitly warns that provider payloads (specifically the _raw field) contain untrusted text, it does not mandate the use of specific delimiters or boundary markers when the agent processes this information.\n
- Capability inventory: The skill's potential impact is significantly mitigated as it is designed for read-only access. It lacks capabilities for file system writes, subprocess execution, or state-changing network requests.\n
- Sanitization: No explicit sanitization or filtering of the retrieved text is mentioned in the instructions, relying instead on the agent's internal safety filters and the provided warnings.
Audit Metadata