browser-automation
Pass
Audited by Gen Agent Trust Hub on Apr 16, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADS
Full Analysis
- [SAFE]: The skill utilizes the 'browser-use' framework and Microsoft's 'Playwright' library, which are standard, reputable tools for browser automation and testing.
- [EXTERNAL_DOWNLOADS]: The installation section provides commands to fetch dependencies and browser binaries from official package registries (PyPI) and browser distributions. These are standard operations for setting up the execution environment.
- [SAFE]: All network activity is directed toward established public scientific databases (e.g., GEPIA2, GeneCards, Human Protein Atlas) for the explicit purpose of research data collection as described in the metadata.
- [PROMPT_INJECTION]: The skill is subject to indirect prompt injection risks inherent in processing untrusted web content.
- Ingestion points: Data is ingested from external scientific URLs listed in the 'Target Database Recipes'.
- Boundary markers: The skill uses structured task instructions for the Agent to define extraction scope.
- Capability inventory: The skill has capabilities for browser navigation and text extraction via the LLM agent.
- Sanitization: No explicit sanitization of web content is performed before processing.
Audit Metadata