web-to-markdown
Pass
Audited by Gen Agent Trust Hub on Sep 23, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes content from untrusted external URLs provided by the user. Malicious instructions within those pages could potentially influence agent behavior during the conversion process. Ingestion points: Web content is fetched from user-supplied URLs via WebFetch or Playwright as detailed in SKILL.md. Boundary markers: None; instructions do not include specific markers or warnings to the AI to ignore embedded commands in the source text. Capability inventory: The skill utilizes tools to write files and execute shell scripts. Sanitization: The skill relies on AI-based markdown transformation rather than programmatic filtering of the input data.
- [DYNAMIC_EXECUTION]: To handle JavaScript-heavy websites, the skill provides a bash heredoc template in SKILL.md to generate and execute a Node.js script using Playwright at runtime.
- [COMMAND_EXECUTION]: The skill uses shell commands to execute local Node.js scraping scripts and perform file write operations via the Write tool.
- [EXTERNAL_DOWNLOADS]: The skill instructions recommend installing well-known libraries such as playwright and @modelcontextprotocol/server-playwright via npm to support browser-based content retrieval for dynamic sites.
Audit Metadata