The Agent Skills Directory

[SAFE]: The skill connects to the vendor's MCP server at https://unbrowse.ai/mcp/milelion.com to proxy requests to milelion.com. This is a standard architectural pattern for the Unbrowse service.
[SAFE]: Authentication is handled via standard OAuth or Bearer tokens, with explicit instructions to use secret management for API keys (e.g., using $UNBROWSE_API_KEY environment variables).
[INDIRECT_PROMPT_INJECTION]: The skill processes data fetched from milelion.com through the milelion_com__get_search and milelion_com__read_page tools.
Ingestion points: External website content from milelion.com is ingested into the agent context via Unbrowse tools.
Boundary markers: Not explicitly defined in the tool documentation; the skill relies on the service to return structured "readable text".
Capability inventory: The skill provides search and page-reading capabilities, along with generic Unbrowse automation tools (unbrowse.run, unbrowse.browse.open).
Sanitization: The documentation states the service returns "verified results" and "readable text," which typically indicates server-side extraction of content to reduce noisy or malicious artifacts.

milelion