video-report
Fail
Audited by Gen Agent Trust Hub on Feb 17, 2026
Risk Level: HIGHEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONCOMMAND_EXECUTION
Full Analysis
- EXTERNAL_DOWNLOADS (HIGH): The skill instructions mandate downloading a user-provided URL and using it as a source for code generation. There is no validation or sanitization of the URL source.
- REMOTE_CODE_EXECUTION (HIGH): The skill implements a 'write-then-execute' pattern. It writes user-controlled data (the URL) into
packages/example/src/NewVideo.tsxand immediately executes that code viabunx remotion render. An attacker could provide a URL such as"}><script>alert(1)</script>or TSX-specific breakouts to execute arbitrary logic during the render process. - COMMAND_EXECUTION (MEDIUM): The skill executes shell commands (
bunx) in a directory (packages/example) that is modified at runtime based on untrusted input. - INDIRECT PROMPT INJECTION (LOW): The skill has a significant attack surface for indirect injection.
- Ingestion points: Video URL provided by the user in the prompt.
- Boundary markers: Absent; the URL is directly interpolated into the
.tsxfile template. - Capability inventory: File system write access and shell command execution (
bunx). - Sanitization: Absent; no escaping or validation of the URL string is performed before it is written to the source file.
Recommendations
- AI detected serious security threats
Audit Metadata