extract-island
Pass
Audited by Gen Agent Trust Hub on Sep 2, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process data from external, untrusted web pages via browser navigation and snapshots.
- Ingestion points: Browser tools (
browser_navigate,browser_snapshot) are used to fetch content from user-provided URLs inSKILL.mdStep 1. - Boundary markers: There are no explicit instructions or delimiters mentioned to prevent the agent from following instructions embedded in the target webpage's HTML (e.g., hidden comments or text).
- Capability inventory: The skill uses MCP tools (
lexsis-ai) and standard browser interactions. It does not appear to have file-write or subprocess capabilities within the provided scope, though its output is intended for manual PR contribution. - Sanitization: No sanitization or filtering of the captured DOM content is described before it is analyzed and converted into JSON layouts.
Audit Metadata