sharp
Pass
Audited by Gen Agent Trust Hub on Mar 29, 2026
Risk Level: SAFEDATA_EXFILTRATIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill documents an image proxy pattern that ingests untrusted data from a URL parameter, creating an indirect injection and SSRF vulnerability surface. \n
- Ingestion points:
searchParams.get('url')in the Next.js API route example inSKILL.md. \n - Boundary markers: None present in the example code. \n
- Capability inventory:
fetch()network requests and complex image processing. \n - Sanitization: No URL validation, domain whitelisting, or protocol checking is implemented in the documentation example. \n- [DATA_EXFILTRATION]: The SSRF vulnerability surface identified in the
Next.js / API Routessection could allow an attacker to probe internal network resources, access internal services, or retrieve cloud instance metadata (e.g., AWS IMDS). \n- [EXTERNAL_DOWNLOADS]: The skill documents and provides examples for fetching content from remote URLs using thefetchAPI for server-side processing.
Audit Metadata