web-ui-audit
Pass
Audited by Gen Agent Trust Hub on Jun 27, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill may fetch updated guidelines from Vercel Labs' official GitHub repository. This is a secure operation targeting a well-known, trusted source.
- [COMMAND_EXECUTION]: Uses the
nlcommand-line utility to index file lines for precise audit reporting. - [PROMPT_INJECTION]: Indirect prompt injection surface (Category 8) detected due to the nature of auditing user-provided code files. 1. Ingestion points: Untrusted source code files (HTML, CSS, React, etc.) provided as audit targets. 2. Boundary markers: Absent; target content is read directly into the agent's context. 3. Capability inventory: Includes local file reads, shell command execution (nl), and network access to fetch guidelines. 4. Sanitization: No specific validation or filtering of the content within the audited files is performed.
Audit Metadata