API and MCP docs

/Endpoints

Endpoints and limits

Discovery also works via the MCP tool list_checks. Limits: 20 requests per 10 minutes per IP and 30 per 10 minutes per tested site (headers X-RateLimit-*, 429 with Retry-After). Each run is capped at 45 seconds; partial results come back with truncated: true.

/MCP

Connect an MCP client

Clients with remote MCP support (Claude, ChatGPT connectors, Cursor, VS Code) take the URL directly. Example for Cursor (.cursor/mcp.json):

{
  "mcpServers": {
    "tech-seo": {
      "url": "https://tech-seo.de/api/mcp/mcp"
    }
  }
}

Clients that only speak stdio can use mcp-remote:

{
  "mcpServers": {
    "tech-seo": {
      "command": "npx",
      "args": [
        "-y",
        "mcp-remote",
        "https://tech-seo.de/api/mcp/mcp"
      ]
    }
  }
}

/Result

Result format

MCP tools return this object as structuredContent plus a short Markdown summary as text. REST returns it as the response body.

{
  "check": "redirect_matrix",
  "version": "1.0.0",
  "input": { ... },
  "startedAt": "2026-10-02T12:00:00.000Z",
  "durationMs": 2140,
  "requestCount": 12,
  "status": "pass" | "warn" | "fail" | "error",
  "summary": "1-2 sentences with numbers",
  "findings": [
    { "id": "redirect.302_in_chain", "severity": "fail", "message": "...",
      "evidence": { ... }, "recommendation": "..." }
  ],
  "data": { ... check-specific facts ... },
  "truncated": false
}

status is fail if any finding is fail, otherwise warn if any is warn, otherwise pass. error means the check could not run (DNS failure, invalid target). Finding ids are stable and safe to match on.

Redirect matrixredirect_matrix

Tests all protocol/host/path variants of a domain (http/https, www/non-www, trailing slash, uppercase, index files, query string) and reports every redirect hop with its status code. Use when asked whether a site's redirects are set up correctly or why a URL resolves inconsistently. Returns pass/warn/fail with findings and the full hop list per variant. Takes 3 to 10 seconds.

~6 s · up to 120 requests · Try it in the browser →

curl -X POST https://tech-seo.de/api/v1/checks/redirect_matrix \
  -H 'content-type: application/json' \
  -d '{"domain":"example.com"}'

Header inspectorheader_inspector

Requests one URL with several user agents (browser, Googlebot, Bingbot, AI crawlers) and compares status codes, raw HTTP headers (X-Robots-Tag, Link canonical, caching, Vary, compression) and HTML head signals per UA. Use when asked whether bots see the same page as users, whether a site blocks AI crawlers, or which headers a URL really sends. Returns pass/warn/fail with findings, captured headers per UA and a diff matrix. Takes 2 to 6 seconds.

~4 s · up to 60 requests · Try it in the browser →

curl -X POST https://tech-seo.de/api/v1/checks/header_inspector \
  -H 'content-type: application/json' \
  -d '{"url":"https://example.com/"}'

robots.txt analyzerrobots_analyzer

Fetches and parses the robots.txt of a URL's origin exactly like Google (RFC 9309: most specific group, longest match, allow wins ties, HTTP status rules) and evaluates the URL for search engines and AI crawlers (Googlebot, Bingbot, GPTBot, ClaudeBot, PerplexityBot, Google-Extended and more). Also checks noindex vs. robots conflicts, whether CDN hosts block CSS/JS for Googlebot, and whether Sitemap directives resolve. Use when asked whether a page or bot is blocked, or why Google cannot see a noindex. Returns per-bot allow/deny with the matched rule. Takes 2 to 6 seconds.

~4 s · up to 40 requests · Try it in the browser →

curl -X POST https://tech-seo.de/api/v1/checks/robots_analyzer \
  -H 'content-type: application/json' \
  -d '{"url":"https://example.com/"}'

Sitemap auditsitemap_audit

Loads all XML sitemaps of a domain (from robots.txt, sitemap indexes recursively, gzip supported), takes inventory (URL count, lastmod quality, duplicates, foreign hosts, http URLs, size limits) and checks a stratified sample of URLs for status 200, redirects, noindex, robots.txt blocking and self-canonical. Only URLs on the own site of the sitemap (same host or subdomains) are fetched; URLs on other hosts are counted, not fetched. Use when asked whether a sitemap is clean or why sitemap URLs are not indexed. Returns findings with counts extrapolated to the whole sitemap and examples. Takes 10 to 40 seconds.

~25 s · up to 400 requests · Try it in the browser →

curl -X POST https://tech-seo.de/api/v1/checks/sitemap_audit \
  -H 'content-type: application/json' \
  -d '{"source":"example.com"}'

hreflang validatorhreflang_validator

Builds the hreflang graph of a language cluster starting from one URL (HTML link tags, HTTP Link headers and optionally XML sitemaps), fetches every alternate URL and checks return links, self-references, ISO language/region codes, x-default, status 200, self-canonical and noindex on targets. Use when asked whether hreflang is implemented correctly or why the wrong language version ranks. Returns findings, the URL × language matrix and all edges with their source. Takes 3 to 20 seconds.

~8 s · up to 150 requests · Try it in the browser →

curl -X POST https://tech-seo.de/api/v1/checks/hreflang_validator \
  -H 'content-type: application/json' \
  -d '{"url":"https://example.com/"}'

Bot verifierbot_verifier

Verifies whether IP addresses really belong to the crawler their user agent claims (Googlebot, Bingbot, GPTBot, OAI-SearchBot, ChatGPT-User, ClaudeBot, PerplexityBot, Applebot and others) using the operators' published IP ranges and forward-confirmed reverse DNS. Accepts IPs, IP plus user agent pairs, or raw access log lines. Use when asked whether bot traffic in logs is real, how much of it is spoofed, or before allowlisting or blocking a crawler. Returns pass/warn/fail with findings, a verdict per IP and totals per bot. Takes 1 to 15 seconds.

~8 s · up to 0 requests · Try it in the browser →

curl -X POST https://tech-seo.de/api/v1/checks/bot_verifier \
  -H 'content-type: application/json' \
  -d '{"ips":["66.249.66.1"]}'

Render diffrender_diff

Fetches one URL twice, as raw HTML and rendered in headless Chromium, and compares SEO signals: title, meta robots, canonical, hreflang, headings, main text, links, structured data and images. Use when asked whether content or links depend on JavaScript, whether search engines or AI crawlers that do not render see the page, or why JavaScript changes indexing signals. Returns pass/warn/fail with findings, both snapshots and a diff. Takes 8 to 30 seconds.

~18 s · up to 12 requests · Try it in the browser →

curl -X POST https://tech-seo.de/api/v1/checks/render_diff \
  -H 'content-type: application/json' \
  -d '{"url":"https://example.com/"}'

list_bots

Lists crawler and fetcher bots with purpose, robots.txt token, verification methods and the age of each published IP list. MCP tool list_bots and GET /api/v1/bots. It does not contact the site you are discussing.