image-gen
Fail
Audited by Gen Agent Trust Hub on Mar 11, 2026
Risk Level: HIGHCREDENTIALS_UNSAFEEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
- [CREDENTIALS_UNSAFE]: A hardcoded API token was detected in the configuration file.
- File: scripts/config.json
- Evidence: "api_token": "ms-e6148603-7898-41aa-9509-f183687d9a18"
- [EXTERNAL_DOWNLOADS]: The skill fetches generated images from ModelScope's official API endpoint.
- File: scripts/generate.py
- Evidence: urllib.request.urlretrieve(url, output_path)
- Target: api-inference.modelscope.cn (ModelScope well-known service)
- [PROMPT_INJECTION]: The skill contains a surface for indirect prompt injection by interpolating user-controlled prompt text into API requests.
- Ingestion points: User-provided prompt argument in scripts/generate.py
- Boundary markers: None present to distinguish instructions from untrusted data
- Capability inventory: Network requests to ModelScope API and file system writes via image downloads
- Sanitization: No sanitization or validation performed on the prompt input
Recommendations
- AI detected serious security threats
Audit Metadata