web-content-fetcher

Pass

Audited by Gen Agent Trust Hub on Mar 29, 2026

Risk Level: SAFE
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill utilizes third-party proxy services r.jina.ai and defuddle.md to fetch and process web content. These are established tools for converting HTML to Markdown and rendering dynamic web pages.
  • [PROMPT_INJECTION]: As a tool for fetching arbitrary web content, the skill is subject to indirect prompt injection risks if the retrieved data contains malicious instructions. This is an inherent characteristic of web-scraping utilities.
  • Ingestion points: Data is ingested from arbitrary URLs provided by the user or discovered via search.
  • Boundary markers: None provided; the skill does not specify delimiters to isolate fetched content from agent instructions.
  • Capability inventory: The skill reads external content which is then processed within the agent's context.
  • Sanitization: The skill relies on the proxy services (jina.ai, defuddle.md) to clean the HTML into Markdown, which provides a layer of structural sanitization but does not filter natural language instructions.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 29, 2026, 08:53 AM