image-ads
Pass
Audited by Gen Agent Trust Hub on Jul 30, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill uses shell commands (
curl) and local Python scripts (e.g.,kie_generate.py) to interface with external image generation APIs and perform image processing tasks. - [EXTERNAL_DOWNLOADS]: The skill communicates with external APIs at
kieai.redpandaai.coandapi.kie.aito manage image generation jobs and fetch results. - [DATA_EXFILTRATION]: Automated upload functionality transfers generated images to Google Drive using the
googleapis.comAPI. This process utilizes OAuth credentials (Client ID, Secret, and Refresh Token) retrieved from environment variables, which is a safe practice. - [PROMPT_INJECTION]: The skill presents an indirect prompt injection surface by interpolating user-provided hooks, offers, and CTA text directly into prompts for the image generation model.
- Ingestion points: User input collected in Step 1 (SKILL.md).
- Boundary markers: Absent; inputs are interpolated directly into prompt templates.
- Capability inventory: Network operations (Google Drive, KIE AI) and local file writing.
- Sanitization: No explicit escaping or validation of user-provided strings is performed before they are used in model prompts.
- [SAFE]: The skill correctly implements security best practices by managing sensitive credentials (API keys and OAuth tokens) through environment variables rather than hardcoding them in the scripts or instructions.
Audit Metadata