create-html-artifact
Pass
Audited by Gen Agent Trust Hub on Oct 4, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill interprets user requests to generate interactive HTML artifacts containing JavaScript, creating an indirect prompt injection surface.
- Ingestion points: The skill processes natural language requests from the user and reads an intermediate HTML fragment file during the build process.
- Boundary markers: The
scripts/build.pyscript enforces a document skeleton, explicitly rejecting inputs that contain document-level tags such as<html>or<body>to maintain structure. - Capability inventory: The skill uses a local script to perform filesystem write operations and includes capabilities for generating client-side JavaScript.
- Sanitization: A structural validation step in the build script ensures the fragment adheres to the expected format, although it does not sanitize the internal content of the fragment.
- [EXTERNAL_DOWNLOADS]: The skill generates HTML documents that reference external assets from well-known services.
- The build script injects the Mermaid diagramming library from
cdn.jsdelivr.netinto the final artifact. - The reference documentation suggests including the Motion library from
cdn.jsdelivr.netfor interaction design. - [COMMAND_EXECUTION]: The skill's workflow involves executing a local Python script to assemble the final deliverable.
- The agent is instructed to run
python3 scripts/build.py, which performs file input/output operations restricted to the local filesystem.
Audit Metadata