image-fetcher

Pass

Audited by Gen Agent Trust Hub on Mar 15, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill exposes a surface for indirect prompt injection by processing untrusted data from the web.
  • Ingestion points: The skill retrieves content through the web_search tool and by navigating to external URLs using the Playwright browser interface (SKILL.md).
  • Boundary markers: There are no instructions to use delimiters or ignore embedded commands within the content retrieved from external sites.
  • Capability inventory: The skill can perform web searches, navigate a browser, capture screenshots, and write files to the reference/ and output/ directories (SKILL.md).
  • Sanitization: The skill implements proactive safety filters to block the file:// protocol and prevent access to localhost or private IP ranges (127.0.0.1, 10.x, 192.168.x), which mitigates common SSRF and file exfiltration vectors.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 15, 2026, 07:52 PM