agent-browser
Pass
Audited by Gen Agent Trust Hub on Sep 18, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill instructs the user to install the
agent-browserpackage from NPM and subsequently runagent-browser installto fetch the Chromium binary. Both the package and its source repository are maintained by Vercel Labs, which is recognized as a trusted organization. - [INDIRECT_PROMPT_INJECTION]: The skill enables the agent to browse arbitrary websites and process their content via accessibility tree snapshots, creating a vulnerability surface where malicious web content could influence the agent's behavior.
- Ingestion points: Untrusted data from external websites is brought into the agent's context using the
agent-browser snapshotcommand (SKILL.md). - Boundary markers: The instructions do not define specific delimiters or warnings to prevent the agent from following instructions embedded within the processed web snapshots.
- Capability inventory: The agent is granted extensive capabilities including file system access (screenshot/PDF generation), network routing control, and state management (saving/loading authentication cookies).
- Sanitization: There are no provided mechanisms for sanitizing or filtering the accessibility tree data before the agent interprets it.
Audit Metadata