h5p-player-setup
Pass
Audited by Gen Agent Trust Hub on Oct 29, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill instructs users to download packages and assets from the vendor's npm registry entries and the jsDelivr CDN. Evidence: Mentions
npm install @missing-elements/h5p-offline-playerand URLs forcdn.jsdelivr.net/npm/@missing-elements/. - [REMOTE_CODE_EXECUTION]: Setup and verification instructions involve running vendor-provided tools via npx. Evidence: Commands such as
npx @missing-elements/h5p-embed@latestin references/embed.md andnpx @missing-elements/h5p-verify@latestin SKILL.md. - [COMMAND_EXECUTION]: Provides standard shell commands for package installation and file management on the host system. Evidence: Use of
npm installandcp -rfor moving package assets to public folders. - [INDIRECT_PROMPT_INJECTION]: The skill documents an interface for playing H5P packages, which act as untrusted data containing executable code. 1. Ingestion points: The
srcattribute of the<h5p-player>tag or the iframe source URL (SKILL.md, references/embed.md). 2. Boundary markers: The documentation explicitly provides anuntrustedattribute to sandbox content and instructions for cross-origin embedding to isolate the H5P runtime environment. 3. Capability inventory: The H5P runtime executes JavaScript from within the package, which can access the host page's context if not isolated. 4. Sanitization: Inuntrustedmode, the player validates the package content against library schemas and prevents the execution of package-provided scripts.
Audit Metadata