agent-browser

Pass

Audited by Gen Agent Trust Hub on Feb 21, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • Indirect Prompt Injection (LOW): The skill creates a vulnerability surface for indirect prompt injection as it processes and acts upon untrusted data from external websites.
  • Ingestion points: templates/capture-workflow.sh and templates/form-automation.sh use agent-browser to ingest web content (text, titles, and DOM structure) into the agent's context.
  • Boundary markers: Absent; the ingested content is not wrapped in delimiters or accompanied by instructions for the agent to ignore embedded commands.
  • Capability inventory: The skill leverages the agent-browser tool to perform actions such as navigation, form interaction, and writing local files (PDFs, screenshots, and session state files).
  • Sanitization: No sanitization or validation of the ingested web content is performed before it is processed.
  • SAFE (SAFE): Credential management practices are secure. The templates utilize environment variables (e.g., APP_USERNAME), and the documentation (e.g., references/authentication.md) explicitly warns against committing session state files to version control.
Audit Metadata
Risk Level
SAFE
Analyzed
Feb 21, 2026, 04:23 PM