agile-proto
Pass
Audited by Gen Agent Trust Hub on Sep 21, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSDYNAMIC_EXECUTIONCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The
templates/scripts/bootstrap.shscript downloads theme configuration from the author's official GitHub repository athttps://cdn.jsdelivr.net/gh/djalmajr/htm-ui@main/THEMING.md. This data is used to dynamically style the resulting prototype. - [DYNAMIC_EXECUTION]: A Python script embedded within
templates/scripts/bootstrap.shis executed to parse the downloaded markdown file and inject the relevant CSS style block into the project'sindex.htmlfile. - [COMMAND_EXECUTION]: The skill instructs the agent to run shell commands for local development, including
bunx serveto host the prototype andbunx skillsto add the complementaryhtm-uiskill. - [INDIRECT_PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection as it generates functional UI code based on potentially untrusted user requirements.
- Ingestion points: User-defined scene requirements and the external markdown file (
THEMING.md) processed during the bootstrap phase. - Boundary markers: The provided templates do not implement specific delimiters or instructions to ensure that the agent ignores directives that might be present in the user-provided content.
- Capability inventory: The skill facilitates file writing (modifying
index.htmland creating route files) and includes several external scripts and modules from third-party CDNs. - Sanitization: The skill relies on the standard rendering and parsing logic of the
htm/preactlibrary to handle the generated prototype code.
Audit Metadata