Connect Claude Desktop, LM Studio, Cursor or any MCP client to twelve real developer tools over a single free Streamable HTTP endpoint. Create an API key, paste the config, and start calling tools in seconds.
MCP (Model Context Protocol) is an open standard that lets AI assistants — Claude, LM Studio, Cursor and other LLM clients — call external tools through one uniform interface. Instead of pasting data into a chat window, the model asks an MCP server for it: DNS records, IP geolocation, QR codes and more.
Arttron runs a free hosted remote MCP server that exposes twelve developer tools over Streamable HTTP. There is nothing to install: add the endpoint and a free API key to your MCP client configuration and the tools appear in the model's tool list immediately.
Twelve real tools served by the MCP endpoint. Each one is also available as a plain REST endpoint and in the browser UI on this site.
dns_lookupResolve DNS records (A, AAAA, MX, TXT, NS, CNAME, SOA, SRV, PTR; SITE for several common types at once) for any domain.
dns_lookup — Example: “Resolve the MX records of example.com” → dns_lookup({ domain: "example.com", type: "MX" })
ip_lookupGeolocate an IP address — country, region and city, plus coordinates and timezone when the database provides them.
ip_lookup — Example: “Where is 8.8.8.8 located?” → ip_lookup({ ip: "8.8.8.8" })
qr_generateGenerate QR codes (PNG or SVG) with a customizable size and error-correction level.
qr_generate — Example: “Make a QR code for https://arttron.dev” → qr_generate({ data: "https://arttron.dev", format: "svg" })
list_versionsList available versions of Arttron Dev Server, including deprecation notices.
list_versions — Example: “Which dev-server versions are available?” → list_versions({})
download_dev_serverGet download links for Arttron Dev Server binary archives (tar.gz or zip).
download_dev_server — Example: “Give me the latest dev-server download” → download_dev_server({ version: "latest" })
proxy_imageFetch and proxy an external image after validating its content type and blocking private IP ranges.
proxy_image — Example: “Fetch this image URL for me” → proxy_image({ url: "https://example.com/logo.png" })
ssl_checkInspects the TLS certificate a host serves: verdict, days to expiry, validity window, subject CN, issuer, SAN names, serial number, SHA-256 fingerprint, negotiated protocol and cipher, and the certificate chain.
ssl_check — Example: “when does example.com’s certificate expire?” → ssl_check({ host: "example.com" })
hashCompute an MD5, SHA-1, SHA-256 or SHA-512 digest, a keyed HMAC-SHA1/256/512, or verify an existing bcrypt hash — with an optional constant-time comparison against an expected value.
hash — Example: “what is the SHA-256 of hello?” → hash({ algorithm: "sha256", input: "hello" })
http_checkInspect a URL over HTTP: final status, the full redirect chain, response headers with a security-header summary (HSTS, CSP, ...), the body title and timing (TTFB and total) — with a verdict and errorCode instead of an error when the URL is down, slow or looping.
http_check — Example: “why doesn’t https://example.com open?” → http_check({ url: "https://example.com/" })
simulate_http_errorReady-to-call URL that returns a chosen HTTP error (status, delay, provider body, or a dropped/hung connection) so you can test your client's error handling.
simulate_http_error — { "status": "503", "preset": "stripe" }
generate_agentGenerate ready-to-write Claude Code and Codex agent configuration files from a structured spec — deterministic output, no LLM.
generate_agent — Example: “create a read-only reviewer agent for Claude Code” → generate_agent({ platforms: ["claude"], name: "reviewer", role: "...", goal: "..." })
markdown_to_htmlConvert GitHub-flavored Markdown to sanitised HTML.
markdown_to_html — Example: “convert this markdown to HTML: ## Heading” → markdown_to_html({ markdown: "## Heading" })
Four steps from zero to your first tool call — the whole setup takes about two minutes.
Open the sign-up page and create an account with your email address or GitHub. No credit card and no installation: the MCP server is hosted for you.
In your dashboard, create a new API key and copy it. The plaintext key is shown only once, so store it safely — it is sent as a Bearer token in the Authorization header.
Paste the configuration below into your MCP client (Claude Desktop, LM Studio, Cursor or any Streamable HTTP-capable client) and replace {key} with the API key you just copied.
Restart the client and ask a question such as “resolve the DNS records of example.com”. The model discovers all twelve tools during the MCP handshake, so no extra prompting is needed.
MCP client configuration
{
"mcpServers": {
"arttron": {
"url": "https://arttron.dev/api/mcp",
"headers": {
"Authorization": "Bearer {key}"
}
}
}
}The endpoint is https://arttron.dev/api/mcp and it speaks MCP Streamable HTTP (JSON-RPC 2.0) with Bearer authentication; claude.ai can add the same URL as a custom connector over OAuth. /api/mcp/stream is the legacy SSE route for the REST API and is not an MCP endpoint. Want to try it without a client? The same twelve tools are available as REST endpoints and in the browser tools on this site.
The free tier gives you 200 tool calls per month — that monthly quota is the headline limit. No credit card, no local server, no Docker: everything runs on our infrastructure, you only paste a URL and a key.
DNS, IP geolocation, QR codes, version listing, downloads and image proxying live behind one Streamable HTTP endpoint, so a single MCP entry gives your assistant twelve capabilities.
On top of the monthly quota, each tool has a small per-second or per-minute limit (for example 20 DNS lookups per second or 30 IP lookups per minute) to keep the service responsive, and the image proxy blocks private IP ranges to prevent SSRF.
A free MCP server is a hosted implementation of the Model Context Protocol that you can connect to at no cost. This one is free for up to 200 tool calls per month, requires only an email account and needs no software installation.
Yes. Any MCP client that speaks Streamable HTTP works, including Claude Desktop, LM Studio and Cursor. Add the endpoint https://arttron.dev/api/mcp and your API key as a Bearer token — or, in claude.ai, add the same URL at Settings → Connectors → Add custom connector and approve the OAuth prompt.
Create an account, generate an API key in the dashboard, then add the endpoint and the key to your MCP client configuration (the JSON snippet on this page). Restart the client and the twelve tools appear in the model's tool list.
No. The server is fully hosted and speaks Streamable HTTP, so there is nothing to install locally — no Docker, no Node.js, no Python package. A browser and an MCP-capable client are enough.
Twelve: dns_lookup, ip_lookup, qr_generate, list_versions, download_dev_server, proxy_image, ssl_check, http_check, hash, simulate_http_error, markdown_to_html and generate_agent. Each tool publishes a JSON input schema that your client discovers automatically.
Yes. The free tier allows 200 tool calls per month per MCP key (100 for API keys) — that monthly quota is the headline limit. Individual tools additionally enforce small per-second or per-minute caps (for example 15 image proxies per second), all counted per API key. No credit card is required.
Sign in to your dashboard with your email, click “Create key” and copy the value that is displayed once. Use it as the Authorization: Bearer header value in your MCP client.
Yes. Every tool is also a plain REST endpoint and has a browser UI on this site (DNS lookup, IP information, QR generator, dev-server downloads), so you can try the same functionality without any AI client.
