zhy-markdown2wechat
Pass
Audited by Gen Agent Trust Hub on Mar 14, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The script
scripts/convert.jsuseschild_process.execSyncto runnpm installfor its dependencies. This is part of its documented zero-deployment feature.\n- [EXTERNAL_DOWNLOADS]: The skill downloads themarkedandjuicepackages from the NPM registry at runtime. As NPM is a well-known service, this is a standard operation for this skill's use case.\n- [REMOTE_CODE_EXECUTION]: The script usesrequirewith a computed path to load the dynamically installed modulesmarkedandjuicefrom a temporary directory.\n- [PROMPT_INJECTION]: The skill processes untrusted Markdown content and lacks sanitization or boundary markers. 1. Ingestion points:mdPathinscripts/convert.js. 2. Boundary markers: Absent inSKILL.md. 3. Capability inventory:child_process.execSyncandfs.writeFileSyncinscripts/convert.js. 4. Sanitization: Absent.
Audit Metadata