Agent Browser Operator
Pass
Audited by Gen Agent Trust Hub on Mar 28, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [SAFE]: The skill instructions and metadata are transparent and do not contain any malicious patterns, behavioral overrides, or safety bypass attempts.
- [EXTERNAL_DOWNLOADS]: The skill documentation references the Playwright library, an open-source tool maintained by Microsoft, and the OpenClaw documentation, both of which are trusted sources for browser automation.
- [COMMAND_EXECUTION]: The skill provides standard installation commands for various agent environments (Claude Code, Cursor, Codex) using 'npx', which is the expected distribution method for this vendor's tools.
- [PROMPT_INJECTION]: As an interactive browser tool, the agent may process untrusted data from web pages. While this represents a theoretical surface for indirect prompt injection, no exploitable patterns or unsafe interpolations were identified in the provided skill definition.
Audit Metadata