The Agent Skills Directory

[SAFE]: No malicious behavior or technical vulnerabilities were identified in the skill instructions.- [EXTERNAL_DOWNLOADS]: Fetches machine-readable specifications and documentation from specification.website and mcp.mdn.mozilla.net. These sources are vendor-controlled or well-known technical documentation providers.- [PROMPT_INJECTION]: No instructions were found that attempt to bypass safety guidelines, jailbreak the agent, or extract internal system prompts.- [REMOTE_CODE_EXECUTION]: The skill operates via structured MCP tools and standard HTTP fetches for Markdown content, with no evidence of dynamic code execution or untrusted script piping.

specification-website