pr-to-video
Pass
Audited by Gen Agent Trust Hub on Oct 19, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
- [COMMAND_EXECUTION]: The skill relies on shell commands to interact with development tools, specifically the GitHub CLI (
gh) for PR ingestion and thehyperframesCLI for video project initialization, linting, and rendering. This is standard for developer-oriented AI agent skills. - [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted content from GitHub Pull Requests. It demonstrates security awareness by implementing data sanitization (stripping HTML comments) and strictly bounding the context provided to sub-agents. The potential for injection exists but is mitigated by the architecture and primary use case.
- [DYNAMIC_EXECUTION]: The skill generates specialized HTML, CSS, and JavaScript for video frames and captions. This dynamic code generation is the core functionality of the skill. The resulting files are self-contained and used within a controlled rendering environment.
- [EXTERNAL_DOWNLOADS]: Targeted network requests are made to GitHub's official avatar services to retrieve contributor profile pictures. These requests are protected by an SSRF guard that restricts hostnames to trusted GitHub domains.
Audit Metadata