html-interactive-playground
Pass
Audited by Gen Agent Trust Hub on Oct 5, 2026
Risk Level: SAFE
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill establishes a data ingestion path where user-provided values from the generated HTML playground are sent back to the agent via a submission function. It proactively mitigates this risk by explicitly instructing the agent to treat this data as raw input only and never as new instructions, commands, or tool calls.
- [DYNAMIC_EXECUTION]: The skill generates HTML and JavaScript artifacts. It enforces security best practices for dynamic content creation by requiring the use of textContent and createElement while strictly forbidding innerHTML to prevent Cross-Site Scripting (XSS) vulnerabilities.
- [DATA_EXFILTRATION]: Communication between the generated artifact and the AI agent is routed through a local loopback address (127.0.0.1) using a single-session nonce. This ensures that parameter tuning data remains within the local environment and is not exfiltrated to external servers.
- [EXTERNAL_DOWNLOADS]: The skill permits the use of Google Fonts via standard link tags, which is a well-known and trusted service for web assets.
Audit Metadata