Skills
skills/baptistearno/typebot.io/agent-browser/Gen Agent Trust Hub

agent-browser

Pass

Audited by Gen Agent Trust Hub on Feb 17, 2026

Risk Level: SAFEPROMPT_INJECTIONDATA_EXFILTRATION
Full Analysis
  • [Indirect Prompt Injection] (LOW): The skill contains templates for extracting text and structure from arbitrary web pages. \n
  • Ingestion points: agent-browser get text body in capture-workflow.sh and agent-browser snapshot -i in various templates.\n
  • Boundary markers: Absent in extraction templates.\n
  • Capability inventory: Full browser control via agent-browser (navigation, form submission, session management).\n
  • Sanitization: Absent; extracted text is saved directly to files without filtering.\n- [Data Exposure] (LOW): The templates manage sensitive authentication state (cookies and session tokens) via local files. \n
  • Evidence: templates/authenticated-session.sh saves session data to auth-state.json. While the references/authentication.md documentation correctly advises users to use .gitignore and environment variables, the default template behavior stores secrets on the local filesystem.
Audit Metadata
Risk Level
SAFE
Analyzed
Feb 17, 2026, 06:33 PM