Search public sources and return ranked, deduplicated, citation-ready results. Use this for discovery across the web, GitHub repositories, academic works, or Hacker News; use read_url instead when a specific page URL is already known. This read-only operation may contact the selected third-party providers, so availability and rate limits depend on them.

Fetch one known public HTTP or HTTPS page and return clean, LLM-ready Markdown. Use this after web_search or when the exact URL is already known; do not use it to discover pages or access authenticated/private content. This read-only operation sends the URL to the configured reader service and can fail when the site blocks retrieval.

Probe every configured search provider and the URL reader, returning availability, latency, and failure details. Use this to diagnose SearchForge setup or provider outages, not to search for content. The probe is read-only but makes small external test requests that may count against provider rate limits.