Tanod Web

What Tanod Web MCP server does

Tanod Web MCP server belongs in the 🔎 Search & Data Extraction category and exposes a broad set of read-only tools for public web research and structured lookup. Web search uses the Mojeek index, while news search uses GDELT and social search covers public Bluesky and Hacker News posts. Search results can be returned as ranked links, or paired with Markdown from the top pages.

The server also works directly with public HTTP(S) pages. It can render static pages as Markdown, execute JavaScript in a sandboxed browser when requested, capture screenshots, read metadata, extract links, parse sitemaps, and evaluate robots.txt rules. Additional tools inspect security headers, convert submitted HTML to text, extract text from public PDFs, parse RSS or Atom feeds, and convert web pages to PDF.

Research-oriented lookups cover company profiles and public jobs from Greenhouse, Lever, and Ashby boards; academic papers and paper digests; podcast discovery; LEI records; RDAP registration data; DNS, email authentication, and TLS details; email preflight checks; IP-to-ASN information; and phishing-list screening.

How it works

Tanod Web MCP server is accessed through a streamable HTTP MCP endpoint and does not require authentication. The tools make requests to their underlying indexes, registries, or public URLs on the server side. Private, internal, and IP-literal destinations are refused by tools that fetch URLs, and redirects are limited for several site-inspection operations.

Most returned web, page, feed, PDF, social, and on-chain content is marked or described as untrusted data. Agents should treat that material as source content rather than instructions. Static fetching does not execute JavaScript; use the JavaScript option on render_page when a client-rendered page is necessary.

Setup and configuration

Configure an MCP client with the streamable HTTP endpoint published at https://tanod.dev/mcp/web. No API key or environment variable is specified for access. The service has per-IP free allowances, with paid pricing after applicable limits. Allowances differ by tool: for example, web search and search-with-contents provide three free daily calls per IP, while several lookup and utility tools provide five or ten. Batch page fetching and some rendering, screenshot, PDF, and digest operations have no free tier.

Tools and capabilities

• Search the web, recent news, social posts, academic papers, podcasts, companies, jobs, and LEI records.
• Fetch one or several public pages as Markdown, including optional JavaScript rendering, and capture screenshots.
• Read page metadata, extract links, parse sitemaps, and check crawler access through robots.txt.
• Inspect security headers, DNS records, SPF, DMARC, DKIM, MTA-STS, TLS certificates, RDAP records, IP networks, and email-domain signals.
• Screen URLs against public phishing lists without fetching the submitted URL.
• Convert supplied HTML to text, extract public PDF text, parse feeds, and print public pages to PDF.

The service reports per-page or per-source failures for many batch and multi-source operations. Several tools impose size, URL-count, redirect, timeout, or output limits, so a large crawl or complete site audit is outside the supported scope.

Limitations and notes

Results depend on the freshness and coverage of third-party indexes and registries. search_with_contents, page rendering, PDF extraction, feeds, and similar outputs may contain incomplete or truncated text. Static fetches cannot read content that requires JavaScript unless render_page is used.

Security-header scoring evaluates one final response rather than an entire site. DNS and email findings are heuristic or structural checks, and verify_email does not contact an SMTP server or confirm that a mailbox exists. Phishing checks compare hosts with public lists, so an unlisted domain is not proof that it is safe. IP lookup reports registration and network information, not physical geolocation. Company profiles use public signals and do not infer missing facts.