xy-mp-layout
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill directs the agent to utilize command-line utilities like
xmllintandrg(ripgrep) to perform structure validation and security auditing on the generated HTML files before delivery to the user.\n- [INDIRECT_PROMPT_INJECTION]: The skill serves as a processing engine for user-provided Markdown content. It includes explicit safeguards to ensure that the resulting HTML is sanitized and free of potentially malicious executable or remote content.\n - Ingestion points: User-provided Markdown files and text blocks as defined in the skill instructions and evaluation cases.\n
- Boundary markers: Instructions establish a strict scope, limiting the agent to layout formatting and explicitly excluding content modification or analysis.\n
- Capability inventory: Writing files to the local directory
公众号HTML输出/and executing shell commands for verification.\n - Sanitization: A mandatory 9-point static check ensures the removal of
<script>,<style>,class,id, and external resource references (URLs, imports) from the final output.
Audit Metadata