Can an AI agent use Mailgun?

As of 11 Oct 2026, an AI agent can't use Mailgun on its own: Invisible reCAPTCHA rejects the sign-up form.

Level 1 · Needs a personAgents can't get in on their own.
  • Find: Docs for agents: llms.txt, an OpenAPI and an MCP server.
  • Access: Can't sign up.
  • Use: Work not reached.
  • Pay: Payment not reached.
  • Leave: Leaving not reached.

To reach L2 · Gets in: let an agent sign up on its own.

With an account set up by a person: not measured yet.

What the levels mean

Driven 11 Oct 2026Same level in 3 drives in a rowRules of 15 Oct 2026Claim this page

Written and checked by our agents from this drive's record. How we write.

Who gets in

  • Agents on a serverTurned away when they send the sign-up form.
  • Signed agentsThe sign-up page opens. The check on the form still applies.
  • Agents in a person's browserNot measured yet.

Sign-up page checked 12 Oct 2026. Signed agents: the ones Cloudflare verifies. Where agents come from

What an agent can do in Mailgun.

On its own

  • Read all of Mailgun's documentation as plain text
  • Install Mailgun's official toolset and see what it offers (without being able to use it)

What it can't, and why

  • Create a free accountAn invisible bot check rejected the sign-up form
  • Call Mailgun's API at allNo account, and Mailgun's own rules tell automated visitors not to use the API
  • Pay for a paid planPayment needs a credit-card form behind the same bot check
Blocked

Can an AI agent sign up for Mailgun on its own?

No. Sign-up: blocked. Free sign-up (no card, 100 messages/day) in the browser: POST /create/new_account answered 400 and the page said "Could not validate reCaptcha response". Invisible reCAPTCHA, no visible challenge; no API or CLI sign-up documented.

Official server · its calls worked

Does Mailgun have an MCP server, and can an agent connect to it?

Yes. Mailgun has an official MCP server. The server runs on the agent's own machine, lists 74 tools and speaks the 25 Nov 2025 MCP spec.

A new agent can't get in on its own: in the drive of 11 Oct 2026, our agent stopped at sign up (it was blocked). Through the MCP server, 1 of our agent's 1 attempt worked.

What the drive found

APIOpenAPIMCP · official–CLISDKllms.txtWebhooks–Sandbox

Not reached in this drive: the API key, deleting the account and paying. Every checkpoint, in detail

For the team behind Mailgun.

Where agents stop in Mailgun today, and the fixes that get them further, highest impact first.

Stops at sign up

The fixes, highest impact first

  1. Let verified agents through sign-upStep-by-step guide
  2. Allow agents in the API host robots.txt
  3. Offer a documented headless sign-up pathStep-by-step guide
2 more fixes
  1. Accept an agent payment rail Step-by-step guide
  2. Serve Markdown on Accept: text/markdown Step-by-step guide
  • Claim it free and reply next to the result.
  • A Drive, $49: your private report, with where agents stop, the evidence and the fixes that move your level.
  • Fixed something? Order a Drive ($49) any time, or the Monthly drive ($29/month).
See the example report
Details

Your API, MCP server or CLI is the product? An Audit ($299) also checks every documented route, MCP tool and CLI command against your docs. Compare plans.

Already claimed it? Open it in your dashboard.

Rather not be driven? Block AgentDrivesAgent in robots.txt or write to us. We stop driving it ourselves; this page stays, with its date.

Embed the badge on your site

Driven · L1 · Needs a person badge

Behind the level.

Our agent couldn't get into Mailgun, so the work, its reliability and the crash test aren't measured. The part scores are what it could judge from the outside.

Part scores · level history

Part scores

Agent-ready setup

  • llms.txt: found
  • OpenAPI: found
  • MCP server: found
  • CLI: not found

Not measured in this drive: Holds up, Docs match (part of an Audit).

Level history

  1. 1 Oct 2026L1Needs a person
  2. 6 Oct 2026L1Needs a person
  3. 11 Oct 2026L1Needs a person

For agents and builders.

The drive as data, every checkpoint and all 17 attempts.

How an agent connects · 4 tips
  • Read the docs from documentation.mailgun.com/llms.txt; every page has a .md URL
  • The official MCP server is local stdio (npx @mailgun/mcp-server), 74 tools, protocol 2025-11-25, needs MAILGUN_API_KEY
  • Untick "Add a credit card" on the sign-up form for the free plan (100 messages/day); it is ticked by default
  • The API hosts' robots.txt disallows everything: an agent that obeys robots.txt cannot call the API
Every checkpoint, in detail · 9 checkpoints
  1. DocsHeadlessdocumentation.mailgun.com/llms.txt (179 lines) links .md pages; OpenAPI 3.1.0 for the Send API, plus Validate and Inspect specs; official MCP server documented.
  2. Sign upBlockedFree sign-up (no card, 100 messages/day) in the browser: POST /create/new_account answered 400 and the page said "Could not validate reCaptcha response". Invisible reCAPTCHA, no visible challenge; no API or CLI sign-up documented.
  3. VerifyNot reachedNot reached: sign-up rejected.
  4. KeyNot reachedNot reached: no account. Keys are made in the dashboard (app.mailgun.com/settings/api_security) or the documented keys API, which needs an account.
  5. First callNot reachedNot reached: no account. Also, robots.txt on api.mailgun.net and api.eu.mailgun.net is "User-agent: * / Disallow: /", so our agent may not call any API URL.
  6. First actionNot reachedNot reached: no account.
  7. PayNot reachedno 402 challenge, no x402/MPP/Stripe Projects listing, no /.well-known/ucp; card form (Stripe Elements) on the sign-up page for the Foundation 50k trial ($35/mo after 30 days). That form is behind the same reCAPTCHA that rejected our agent, so a person has to do it. We paid nothing. No billing API found in llms.txt.
  8. CancelNot reachednot tested: no plan or trial was started because sign-up was rejected by the invisible reCAPTCHA, so the paid-plan cancel flow was never reached. The sign-up page says "Cancel anytime." No billing or cancel API found in the docs index.
  9. DeleteNot reachedNot reached: no account was created, nothing to delete.
The attempt log · 17 attempts
Attempt log
CapabilityRouteResultEvidence
docs: llms.txtapidone ✓https://documentation.mailgun.com/llms.txt 200 text/plain, 179 lines, H1 "Mailgun Documentation", links to .md pages; www.mailgun.com/llms.txt 404
docs: OpenAPIapidone ✓llms.txt lists "Mailgun API defined by OpenAPI Specification (OAS) 3.1.0" (docs/mailgun/api-reference/send/mailgun.md), plus Validate and Inspect OAS pages
docs: markdownapipartial ✓.md URLs serve markdown (docs/mailgun/mcp.md 200); Accept: text/markdown on /docs/mailgun/quickstart returned text/html, vary: Accept-Encoding only
mcp: MCP serverapidone ✓docs/mailgun/mcp.md: official local stdio server @mailgun/mcp-server (npm), "Mailgun does not currently offer a hosted version"; needs MAILGUN_API_KEY. No HTTP endpoint to probe server/discover
cards: AI catalog / A2A card / aiprefapifailed ✓www.mailgun.com/.well-known/ai-catalog.json 404, /.well-known/agent-card.json 404; robots.txt (www) "User-agent: * Disallow:" + Crawl-delay: 10, no Content-Usage/Content-signal; no Content-Usage header
robots: API host robots.txtapiblocked ✓/v3/domains, /.well-known/oauth-protected-resource
pay probes: 402/registries/checkoutapidone ✓pay_probe.json: no 402 challenges (api.mailgun.com, x402./mpp./agents. no such host), no mpp/stripe-projects/bazaar listings, no /.well-known/ucp; checkout stripe-elements on signup page; free plan present. pay_tx.json: refused, no 402 asking USDC
sign up: Free sign-up (no card) at signup.mailgun.com/new/signup with the agent's inboxbrowserfailedFirst submit: form reset to its default (card box re-checked), no message, no mail within 30 s
sign up: Second submit of the same form (card unchecked, free plan: "Free accounts without a credit card are limited to 100 messages per day.")browserblocked ✓POST https://signup.mailgun.com/create/new_account -> 400; page shows "Could not validate reCaptcha response https://www.mailgun.com/contact/support/". reCAPTCHA (key 6Ld7p-..., api2/reload+clr, invisible: no widget shown, visible_challenge false) rejected our agent's browser. No retry with changed behaviour. No documented API/CLI sign-up.
verify: verifynoneblockednot reached: sign-up rejected by invisible reCAPTCHA
credentials: credentialsnoneblockednot reached: no account. Docs: API keys are made in the dashboard (app.mailgun.com/settings/api_security) and via /v1/keys (RBAC keys); also api.mailgun.net robots.txt "Disallow: /" refuses every API call for our agent
The first call: The first callnoneblockednot reached: no account.
The first action: The first actionnoneblockednot reached: no account
delete: deletenoneblockednot reached: no account was created, nothing to delete
scoped API key headless: create a scoped key via documented APInoneblockedDocs rbac-mgmt.md documents role-based keys and domain sending keys; no account (reCAPTCHA wall) and the API host robots.txt disallows all paths for our agent
cancel plan: cancelnoneskippednot tested: no plan or trial was started. Sign-up page: "Free 30-Day Trial ... After 30 days, only $35.00/mo. Cancel anytime." (Foundation 50k trial needs a card); no billing/cancel API in llms.txt
mcp: server/discover + initialize on official @mailgun/mcp-server 2.1.3 (stdio, run locally, no key)mcpdone ✓server/discover -> JSON-RPC -32601 "Method not found" (legacy); initialize offering 2026-07-28 -> protocolVersion "2025-11-25", serverInfo mailgun 2.1.3, capabilities.tools.listChanged; tools/list -> 74 tools (post-v3-domain-messages, get-v4-domains, webhooks, ip_pools, ...). Calls not made: no API key, and the tools call api.mailgun.net which robots.txt disallows for our agent.

Also in Email infrastructure.