baoyu-markdown-to-html
Pass
Audited by Gen Agent Trust Hub on Aug 14, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONREMOTE_CODE_EXECUTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The
scripts/main.tsfile contains adownloadFilefunction that uses thehttpandhttpsmodules to fetch images from arbitrary URLs found within the input Markdown content. These images are saved locally to a temporary directory during conversion. - [EXTERNAL_DOWNLOADS]: The syntax highlighting utility in
scripts/md/utils/languages.tsis configured to fetch additional Highlight.js language definitions from a remote CDN athttps://cdn-doocs.oss-cn-shenzhen.aliyuncs.com. - [COMMAND_EXECUTION]: The main entry point
scripts/main.tsusesspawnSyncto execute the rendering engine usingnpx -y bun. This facilitates the internal workflow of the skill but involves executing shell commands to manage its lifecycle. - [REMOTE_CODE_EXECUTION]: The file
scripts/md/utils/languages.tsemploys dynamicimport()to load JavaScript modules from a remote URL constructed at runtime. While this is used for extending syntax highlighting capabilities, loading and executing code from a remote server presents a potential supply chain risk if the CDN is compromised. - [INDIRECT_PROMPT_INJECTION]: The skill acts as a processing engine for user-provided Markdown files. This represents an indirect prompt injection surface where maliciously crafted Markdown (e.g., exploiting specific rendering extensions or diagram parsers) could attempt to influence the agent's behavior or compromise the output integrity. The skill lacks explicit sanitization or boundary markers for the ingested content.
Audit Metadata