google-image-creator
Pass
Audited by Gen Agent Trust Hub on Oct 9, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSCREDENTIALS_UNSAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes local TypeScript scripts (
generate-image.tsandlist-models.ts) usingnpx tsxto manage API requests and file operations.\n- [EXTERNAL_DOWNLOADS]: Fetches image data from Google's official API (generativelanguage.googleapis.com). This is a well-known service provided by a trusted organization.\n- [CREDENTIALS_UNSAFE]: The skill requires the use ofGOOGLE_API_KEYorGEMINI_API_KEY. It follows security best practices by recommending the use of environment variables and providing instructions for safe key management rather than hardcoding credentials.\n- [INDIRECT_PROMPT_INJECTION]: The skill provides an attack surface by ingesting external prompts and file paths for image generation and storage.\n - Ingestion points: The
promptandoutputPatharguments inscripts/generate-image.ts.\n - Boundary markers: No explicit delimiters are used in the interpolation of the prompt into the API request.\n
- Capability inventory: The script uses
fs.writeFileto persist generated images to the local filesystem.\n - Sanitization: No validation or sanitization is performed on the
outputPathparameter, which could theoretically allow file overwrites if the agent is provided with an unintended path.
Audit Metadata