The Agent Skills Directory

[PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection because its primary function is to ingest HTML data from arbitrary external URLs.
Ingestion points: Untrusted data enters the agent's context through the getHTML function and the npx html-get command as described in SKILL.md.
Boundary markers: The skill does not implement or recommend the use of boundary markers or instructions to treat the fetched HTML as data rather than instructions.
Capability inventory: The skill utilizes network operations to fetch data and supports command-line execution via npx.
Sanitization: No sanitization or security-focused filtering of the retrieved HTML content is present to prevent embedded malicious prompts from influencing the agent.
[EXTERNAL_DOWNLOADS]: The skill requires the installation of several Node.js packages including html-get, browserless, and puppeteer. These are standard dependencies for the headless browser and HTML retrieval functionality described.
[COMMAND_EXECUTION]: The skill demonstrates the use of npx to execute the html-get package directly. This allows the agent or user to run the tool as a CLI utility for debugging and metadata extraction tasks.

html-get