Enables any MCP-compatible AI coding client such as Claude Code, Codex, or Cursor to run web searches and fetch pages as Markdown through a self-built search engine, with per-day and per-month spend caps, prompt-injection fencing on retrieved content, and local usage/cost reporting. Search results can be filtered by site, language, and date, and repeated page fetches within a short window are not billed twice.
Self-hosted MCP server that provides web scraping and crawling tools, integrating seamlessly with AI frameworks like OpenAI Agents SDK, Cursor, and Claude Code.
An MCP server that gives AI agents web access through discoverable tools for keyless web search, reading public pages as text or Markdown, and querying platform APIs such as GitHub, RSS, V2EX, YouTube, Reddit, X/Twitter, and LinkedIn. It supports keyless defaults with optional per-platform credentials, plus timeouts, cancellation, retries, and backend diagnostics for controlled requests.
A self-hosted MCP server for web search with multi-engine support, combining search, URL fetching, and deep research along with proxy routing and deduplication.