← Docs · Payloads · Markdown · Agent register · Home

Fetch

Remote web fetch for agents: given an http(s) URL, downloads the page, extracts the main article with Mozilla Readability, and returns clean markdown. XHTML pages with an XML encoding declaration (e.g. NCBI Bookshelf) are normalized before parsing so lxml does not reject already-decoded Unicode. Cloudflare bot challenges that block plain httpx are retried with curl_cffi Chrome TLS impersonation. If Readability yields almost no text (typical JS homepage with copy in noscript), the fetcher falls back to noscript / meta description. Supports max_length / start_index pagination and optional raw HTML. Respects robots.txt (a 403 on robots.txt itself is not a disallow — Cloudflare often wraps it). Hard JS/Turnstile walls after curl_cffi are returned as kind=bot_blocked, not a generic fetch_failed.

Service id: fetch
Version: 0.1.4
Status: available

Canonical payloads: https://mcp.glc-rag.hu/guide/fetch/payload · markdown · docs://fetch/payload

Authentication

MCP endpoint: https://mcp.glc-rag.hu/mcp (streamable HTTP)

Agents (recommended): self-register with account_type=agent to get an auto-approved token — see https://mcp.glc-rag.hu/guide/agent.

Or register as a human on the public site (all listed services are auto-approved), wait for system-admin approval, then create a token.

Authorization: Bearer mcp_...

Cursor mcp.json example:

{
  "mcpServers": {
    "fetch": {
      "url": "https://mcp.glc-rag.hu/mcp",
      "headers": {
        "Authorization": "Bearer mcp_YOUR_TOKEN"
      }
    }
  }
}

Tools

fetch

Fetches a URL from the internet and extracts its contents as markdown. Although originally you did not have internet access, this tool grants you access to public web pages. Returns title, markdown content, size, and truncation metadata (compatible with the official MCP fetch server).

Input schema:

{
  "type": "object",
  "properties": {
    "url": {
      "type": "string",
      "description": "URL to fetch (http or https)"
    },
    "max_length": {
      "type": "integer",
      "description": "Maximum number of characters to return (default 5000).",
      "minimum": 1,
      "maximum": 1000000,
      "default": 5000
    },
    "start_index": {
      "type": "integer",
      "description": "Start returning output at this character index (use after a truncated fetch).",
      "minimum": 0,
      "default": 0
    },
    "raw": {
      "type": "boolean",
      "description": "Return raw page body without Readability simplification.",
      "default": false
    }
  },
  "required": [
    "url"
  ],
  "additionalProperties": false
}

Examples:

{
  "url": "https://example.com"
}

{
  "url": "https://example.com",
  "max_length": 2000
}

{
  "url": "https://example.com",
  "start_index": 5000,
  "max_length": 5000
}

fetch_status

Fetch service health and default limits.

Input schema:

{
  "type": "object",
  "properties": {},
  "additionalProperties": false
}

Examples:

{}

Usage notes

Call fetch with url. If truncated is true, call again with start_index = next_start_index to continue. Set raw: true for unsimplified HTML. Same extraction stack as @modelcontextprotocol/server-fetch (httpx + readabilipy + markdownify), plus curl_cffi for Cloudflare-class bot walls.

Errors / limits

Non-http(s), private IPs, robots.txt disallow, HTTP ≥400, oversized body, DNS failure, or an uncleared bot challenge → error object with error string. Invalid API key → platform 401.

Agent discovery