tinyfish-web
Pass
Audited by Gen Agent Trust Hub on Nov 1, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill orchestrates tools that process untrusted web content as documented in
SKILL.md. - Ingestion points: External web content retrieved via
search,fetch_content, andrun_web_automationtools. - Boundary markers: The skill contains explicit instructions to ignore embedded instructions found in web pages, search snippets, or form labels.
- Capability inventory: The tools orchestrated include browser automation (
run_web_automation), direct browser session control (create_browser_session), and persistent monitoring (create_monitor). - Sanitization: Instructions emphasize extracting only relevant data and strictly avoiding the use of local secrets or hardcoded credentials in tool parameters.
Audit Metadata