gemini-image-coder
Pass
Audited by Gen Agent Trust Hub on Oct 9, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted user input in the form of text prompts and editing instructions which are passed to the Gemini API.
- Ingestion points: Input is accepted through the
promptargument ingenerate_image.pyand theinstructionargument inedit_image.py. - Boundary markers: The scripts do not utilize delimiters or specific system instructions to isolate user input from the agent's task instructions, making them susceptible to embedded commands.
- Capability inventory: The skill has the capability to write files to the local file system (
image.save) and communicate with external Google APIs. - Sanitization: There is no validation or escaping of the user-provided strings before they are incorporated into the prompt sent to the AI model.
Audit Metadata