UTM MCP Server: Build and Validate Campaign Links from Your AI Agent

What the MissingLinkz MCP exposes

When you run mlz mcp, the server exposes the following tools over the MCP stdio transport. Any MCP-compatible client (Cursor, Claude Code, Claude Desktop, or a custom agent) can call them by name:

mlz_build_link

Generates a UTM-tagged campaign link from your destination URL, source, medium, and campaign name. Automatically normalises values to lowercase-hyphenated format — Google Ads becomes google-ads before the URL is assembled. Returns a tracked_url and the assembled params object.

mlz_preflight

The full pre-publish check in one call. Builds the UTM link, validates the destination (SSL, HTTP resolution, redirect chain, UTM parameter preservation), and inspects the landing page for social sharing readiness (OG tags, Twitter Card, viewport). Returns a ready: true/false verdict that an agent can gate on before publishing any link.

mlz_inspect_destination

Inspects a URL without building a UTM link. Returns Open Graph tags, Twitter Card tags, viewport presence, canonical URL, favicon status, and page load time. Useful when you want metadata about a landing page independently of UTM generation.

mlz_validate_url

URL-level validation: SSL, HTTP resolution, redirect chain, and response time. No OG or UTM logic — just the liveness check.

mlz_list_campaigns

Lists campaign names from your workspace so an agent can pick from existing campaigns rather than invent new names that fragment attribution data.

The complete tool list also includes mlz_suggest_naming, mlz_list_links, and mlz_check_usage. The three tools above cover the most common agent workflows: build, validate, and inspect. The rest are covered in the MCP marketing tools pillar.

Connect to Cursor in 2 minutes

Cursor reads MCP server configuration from .cursor/mcp.json in your project root (or the global ~/.cursor/mcp.json for all projects). Add the following block:

{
  "mcpServers": {
    "missinglinkz": {
      "command": "mlz",
      "args": ["mcp"]
    }
  }
}

If you haven’t installed the CLI yet:

After saving the config and reloading Cursor, the MissingLinkz tools appear in the MCP tool list. You can prompt Cursor’s AI with natural language and it will call the right tool automatically. For example:

“Build a UTM link for this LinkedIn campaign”

Cursor calls mlz_build_link with the source, medium, and campaign parameters extracted from your conversation context.

“Validate this landing page before I share it”

Cursor calls mlz_preflight and returns the full go/no-go report including OG tag status, SSL, and redirect chain results.

“Check if the og:image is configured on this URL”

Cursor calls mlz_inspect_destination and returns the full Open Graph and Twitter Card metadata.

For a complete walkthrough of connecting MissingLinkz to Cursor, including authentication and example prompts, see the Cursor MCP setup guide.

Connect to Claude Code in 2 minutes

Claude Code reads MCP configuration from .claude/settings.json in your project, or from the global settings file. Add the mcpServers block:

{
  "mcpServers": {
    "missinglinkz": {
      "command": "mlz",
      "args": ["mcp"]
    }
  }
}

The MCP server uses stdio transport only — no ports opened, no network server started. The agent communicates with the server over stdin/stdout, which keeps the setup minimal and secure. Running mlz mcp directly in a terminal will appear to hang; that is expected behaviour (it is waiting for JSON-RPC messages on stdin). The MCP client handles that communication.

Once connected, a Claude Code agent can call mlz_preflight within a task and gate further steps on the ready field in the response. For example:

{
  "ready": true,
  "tracked_url": "https://acme.com/lp?utm_source=linkedin&utm_medium=social&utm_campaign=q3-launch",
  "checks": [
    { "check": "og_tags", "status": "pass", "message": "All essential Open Graph tags present." },
    { "check": "twitter_card", "status": "pass", "message": "Twitter Card tags configured." },
    { "check": "ssl", "status": "pass", "message": "URL uses HTTPS." },
    { "check": "resolution", "status": "pass", "message": "Destination responded with 200." }
  ],
  "summary": { "total": 12, "passed": 12, "warnings": 0, "failed": 0 },
  "recommendation": "All checks passed. Campaign link is ready to publish."
}

The equivalent CLI command produces the same output:

$ mlz preflight \
    --url "https://acme.com/lp" \
    --source "linkedin" \
    --medium "social" \
    --campaign "q3-launch"

Why the ready flag is the gate your agent workflow is missing

Most UTM MCP tools return a URL string. Your agent receives the tracked URL and proceeds to post it, schedule it, or store it — without knowing whether the destination is reachable.

The MissingLinkz MCP returns a structured ready boolean alongside every preflight report. An agent can check this field in a single conditional:

• If ready: true — publish the link.
• If ready: false — stop, surface the failed checks, and wait for a human review.

This pattern is infrastructure-grade: the agent doesn’t need to parse a prose summary or infer failure from the shape of the response. The boolean is explicit. The check categories cover everything that can silently break a campaign at launch:

SSL / HTTPS

Ad platforms and social networks penalise HTTP destinations. An agent building a link for a LinkedIn ad shouldn’t publish it if the destination isn’t on HTTPS.

Destination resolution

Does the URL return a 200? A 404 or redirect to a parked domain will still produce a valid UTM string — but the campaign will be dead on arrival.

Redirect chain and UTM preservation

Redirect chains (URL shorteners, CDN rewrites, landing page builders) frequently strip query strings. mlz_preflight follows the full chain and reports whether UTM parameters survive every hop. See also: how redirect chains strip UTM parameters.

Open Graph tags

Every link shared on LinkedIn, X, or Facebook needs og:title, og:description, and og:image. Missing OG tags produce broken social previews that suppress click-through. Agents building links for social distribution should gate on OG tag presence.

Twitter Card

twitter:card controls whether X renders the full image preview or falls back to a link-only display. A failing Twitter Card check is a silent engagement killer.

An agent that only builds UTM links is automating the easy part. The hard part — the part that actually breaks campaigns — is what happens to the link after it’s generated. That is what mlz_preflight covers. Read more about what preflight checks in the all-in-one SSL, OG, and UTM preflight guide.

UTMMind MCP vs MissingLinkz MCP

UTMMind’s MCP server (listed on mcpservers.org) lets agents build UTM links and enforce naming governance rules. It’s a strong tool for UTM taxonomy. But it does not validate the destination: there is no SSL check, no OG tag inspection, no redirect chain analysis in UTMMind’s MCP responses. The two tools are complementary — not identical.

For a deeper look at this comparison and when each tool is the right choice, see UTMMind MCP vs MissingLinkz: governance vs. validation. For a full roundup of MCP servers available for campaign link validation, see best MCP servers for campaign link validation in 2026.

Related reading

Frequently asked questions

Do I need an API key to use the MissingLinkz MCP server?

UTM link generation (mlz_build_link) works offline without an API key. Destination validation tools (mlz_preflight, mlz_inspect_destination, mlz_validate_url) require an active API key because they make outbound HTTP requests from the MissingLinkz API. You can register for a free key with mlz auth register --email [email protected].

Is the MCP server available on Windows?

Yes. The MCP server uses stdio transport and runs wherever Node.js is available. The configuration format and commands are identical on macOS, Linux, and Windows.

Can I run the MCP server as a background service?

The MCP server is designed to be launched on-demand by your MCP client (Cursor, Claude Code, etc.), not as a persistent daemon. Each MCP client manages the lifecycle: it starts mlz mcp when it needs a tool call and keeps it running for the session. You do not need to start it manually.

What is the difference between mlz_preflight and mlz_validate_url?

mlz_validate_url checks the URL itself: SSL, HTTP resolution, redirect chain, and response time. mlz_preflight does all of that plus builds the UTM link, checks Open Graph tags, Twitter Card tags, viewport, canonical URL, and favicon, and returns a single ready boolean. Use mlz_validate_url for a quick liveness check; use mlz_preflight for the full pre-publish gate.

Does the MissingLinkz MCP work with n8n or LangChain?

Any agent framework that supports the Model Context Protocol’s stdio transport can connect to mlz mcp. For frameworks that don’t support MCP natively, use the REST API directly via POST /v1/preflight — the response schema is identical to the MCP tool output.