html-to-interaction-prompts
Pass
Audited by Gen Agent Trust Hub on Aug 25, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The instructions direct the agent to use shell tools for project maintenance and validation, specifically
git(for diffing and adding files) andffprobe(to verify the readability of generated MP4 video files). - [COMMAND_EXECUTION]: The workflow suggests serving HTML content on
localhostif local file access is restricted, which involves initiating a local web server to bypass browser security policies. - [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection (Category 8) due to its processing of untrusted external content. Ingestion points: The skill reads HTML, CSS, and JavaScript from local files or live URLs provided by the user. Boundary markers: The skill does not implement delimiters or safety warnings to separate untrusted source data from the instructions in the final prompt output. Capability inventory: The agent can write to the file system (dated article folders) and execute shell commands (git, ffprobe). Sanitization: There are no specified procedures for sanitizing or escaping the content extracted from the analyzed HTML before it is used to construct prompts for subsequent AI models.
Audit Metadata