The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill facilitates the ingestion of untrusted data from arbitrary external web pages, which poses a risk of indirect prompt injection attacks where malicious instructions are hidden in the fetched content.\n
Ingestion points: External content is retrieved via the <url> parameter and provided to the agent as Markdown (SKILL.md).\n
Boundary markers: There are no explicit delimiters or instructions provided to the agent to treat the fetched content as untrusted data or to ignore any commands found within it.\n
Capability inventory: Depending on the environment, the agent may have access to other tools (such as file writing or network access) that could be exploited by instructions found on a malicious webpage.\n
Sanitization: The tool converts HTML to Markdown for readability but does not include security-focused sanitization to filter out prompt injection attempts.\n- [EXTERNAL_DOWNLOADS]: The skill uses npx to fetch and execute the @arabold/docs-mcp-server package from the npm registry at runtime. This package belongs to the verified vendor for this skill.\n- [COMMAND_EXECUTION]: The skill performs shell command execution through npx to invoke its core functionality.

fetch-url