htmlify
Pass
Audited by Gen Agent Trust Hub on Sep 2, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTIONPERSISTENCE
Full Analysis
- [COMMAND_EXECUTION]: The skill uses
execFileSyncinsrc/output.tsto executeosascripton macOS. This is used to control the Safari browser to open generated HTML files. The command is designed to close all existing Safari windows to provide a distraction-free view, a behavior that is explicitly documented in theSKILL.md. The input paths are properly URL-encoded usingpathToFileURLto prevent shell injection. - [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted Markdown content to generate HTML companions.
- Ingestion points: Markdown artifacts are read from the file system in
src/readArtifact.ts. - Boundary markers: The skill does not use specific delimiters to wrap the content, but it treats section headings as structural boundaries.
- Capability inventory: The skill can write files to the
.superpowers-html/directory and execute browser-opening commands viaosascript. - Sanitization: The skill effectively mitigates risk by using
isomorphic-dompurifyand themarkedlibrary's parser to sanitize all generated HTML content, stripping potentially malicious<script>tags andon*event handlers before rendering (implemented insrc/helpers/markdown.tsand individual component files). - [PERSISTENCE]: The skill includes a setup script (
scripts/setup-htmlify-hook.sh) intended to install aPostToolUsehook. This ensures the skill automatically runs whenever specific Markdown files are modified within the Claude Code environment.
Audit Metadata