gemini-image
Warn
Audited by Gen Agent Trust Hub on Aug 14, 2026
Risk Level: MEDIUMCREDENTIALS_UNSAFEDATA_EXFILTRATIONCOMMAND_EXECUTION
Full Analysis
- [CREDENTIALS_UNSAFE]: The skill instructions direct the agent to access
config/secrets.mdto retrieve an API Key, which is then used in a network request. - [DATA_EXFILTRATION]: The skill reads a local secret and transmits it to
https://api.apicore.ai/v1/images/generations. This endpoint is not a recognized well-known or trusted service provider, and sending local credentials to third-party domains represents a data exposure risk. - [COMMAND_EXECUTION]: The skill utilizes the
curlcommand to interact with an external API, transmitting local data (the API key) and user input to a remote server. - [INDIRECT_PROMPT_INJECTION]: The skill ingests user-supplied text and image URLs to construct a prompt that is sent to the image generation API without validation or sanitization.
- Ingestion points: User-provided description text and image URLs in Step 2 of
SKILL.md. - Boundary markers: None (the prompt is built by simple interpolation).
- Capability inventory:
curlnetwork requests to an external API. - Sanitization: None detected for the description or URL inputs.
Audit Metadata