Head to head · Sending email · October 2026 research run

Loops API + MCP vs Mailgun API + MCP

Mailgun API + MCP scores 71.8 (BB) on agent readiness against Loops API + MCP's 66.4 (B), and leads in 5 of 7 scored categories. Loops API + MCP leads on reliability and agent ergonomics. Both do sending email.

Best email delivery APIs for AI agents · All 174 email comparisons

Which one, for what

Loops API + MCP B

Best for SaaS teams that send lifecycle and product email from one contact list and want an OAuth MCP.

Ahead on

  • Reliability, 90 against 72
  • Agent ergonomics, 70 against 60

Also in its favour

  • Free to start without a card
  • Rated higher by the reviewer agents, 3.5 against 3.0 out of 5

Watch for

No scoped or read-only keys, and no read-only MCP mode

Mailgun API + MCP BB

Best for Teams that want inbound routing, validation and per-key roles in one account, in US or EU regions.

Ahead on

  • Security & auth, 74 against 46
  • Payments & pricing, 60 against 45

Also in its favour

  • Agent-ready, a grade of BB or better
  • Runs on your own machine
  • No incidents deducted, where Loops API + MCP loses 3 points for them

Watch for

74 MCP tools load by default, each a one-line summary, with no readOnlyHint or destructiveHint

Why the edge

Each category where one leads by five points or more, with the reason behind each side's score from its assessment. The numbers are in the table below; this is what they stand for.

Security & auth Mailgun API + MCP +28

Loops API + MCP 46

Bearer API keys created and revoked in settings. The hosted MCP uses browser OAuth with Client ID Metadata Documents and no documented scopes (22). No read-only or send-only keys and no read-only MCP mode found (0). No inbound mail, so tools return only account data (10). No audit log found (0). A security reporting page (updated 19 October 2026) takes reports at security@loops.so, acknowledges within a few days and pays for reports scored CVSS v4 4 or higher, and a compliance page links the SOC 2 trust centre at trust.oneleet.com/loops and a DPA. No security.txt (404 on 2 November) and no published advisories. Below customer-io (16, which adds HIPAA and penetration tests) and sendgrid (15, which adds ISO 27001) (14 of 20). 22 + 0 + 10 + 0 + 14 = 46.

Mailgun API + MCP 74

HTTP Basic auth with a key. Four key roles (Admin, read-only Analyst, Developer, Support), Domain Sending Keys limited to POST /messages for one domain, and IP allowlisting. Rotation is advised but keys don't expire (28). The Analyst role is read-only, sending keys cover the least-privilege case, and the MCP server exposes no delete operations. Confirmation is left to the client (15). The MCP can fetch stored messages, so inbound mail can reach the model. The README's prompt-injection section asks users to review tool calls and goes no further (6). Logs API with 1 to 30 days of retention by plan, and Mailgun says security logs are kept 365 days (10). SOC 2 Type II, ISO/IEC 27001 and a third-party bug bounty per the security page, re-read on 1 November 2026, which names no platform or disclosure address. Sinch's trust portal, linked from the site, lists ISO 27001:2022 and SOC 2 Type 2 and takes vulnerability reports at security@sinch.com. No security.txt (404). Level with sendgrid's 15 on certifications and a bounty, and above ahasend and brevo (12, ISO 27001 and a disclosure route, no bounty) (15).

Reliability Loops API + MCP +18

Loops API + MCP 90

Statuspage at status.loops.so with components for sending, transactional, API, SMTP relay and webhooks (20). One incident in the 90 days to 2 November, "Delay in email metric processing" on 11 September (about 90 minutes, metrics slow to appear in the app, none lost), which doesn't touch the API or sending, so the record reads as clean (30). 10 requests a second per team, and 60 a minute for the content endpoints (15). 429 with x-ratelimit headers and advice to retry with exponential backoff, and an Idempotency-Key header on event and transactional sends (15). No SLA found on the pricing page (0). GA (10). 20 + 30 + 15 + 15 + 0 + 10 = 90.

Mailgun API + MCP 72

Statuspage at status.mailgun.com with dated history (20). The history pages, re-read on 1 November 2026, list ten incidents from 3 August to 1 November. EU messaging outages of 38 minutes on 27 August ("an outage in our EU datacenter") and 17 minutes on 4 September, labelled critical, touched one region, so they read as partial. The rest were US click and open tracking under a denial-of-service attack for 95 minutes on 10 and 11 August, GCP us-west1 log and webhook delays on 20 August, a US event-log backlog of about 7 hours on 31 August, EU route delays on 14 September, 30 September and 28 October, and two Microsoft 365 delivery problems outside Mailgun. Minor only (20). The 13 July US validation outage counted as a major on 1 October has left the 90-day window. The OpenAPI spec gives 500 requests per 10 seconds for the Metrics API and documents 429 responses. No general send limit is published (7). No Retry-After or backoff guidance and no idempotency key on POST /messages, per the 1 October check, not re-read on 1 November (5). The SLA page commits to 99.99 per cent monthly availability of the API, SMTP and outbound delivery, excluding maintenance, with a 5 per cent credit for each 30 minutes below it, and the pricing page ticks it for Foundation and Scale (10). GA (10). 20 + 20 + 7 + 5 + 10 + 10 = 72.

Payments & pricing Mailgun API + MCP +15

Loops API + MCP 45

No x402, MPP or L402 (0). Paid plans are priced by subscribed contacts on a slider, Growth from $49 a month for up to 5,000 contacts with unlimited sends, re-read on 2 November (10). Free plan of 4,000 sends in any rolling 30 days to 1,000 contacts, with no card (20). Browser signup, no autonomous route (0). Re-scored on 13 November 2026 for the new payments rubric (protocol 10, the other three lines 30 each): 30 is 45.

Mailgun API + MCP 60

No x402, MPP or L402 (0). Plan prices and per-1,000 overage ($1.80, $1.30, $1.10) published without login (20). Free plan of 100 emails a day with no card (20). A person signs up in a browser and verifies a domain, so no autonomous route (0). Re-scored on 13 November 2026 for the new payments rubric (protocol 10, the other three lines 30 each): 40 is 60.

Agent ergonomics Loops API + MCP +10

Loops API + MCP 70

The MCP covers contacts, lists, events, transactional sends, campaigns, workflows and content. We couldn't count its tools, and API responses are small (15). perPage and cursor pagination on lists (17). Errors come back with a message, and the SDK raises RateLimitExceededError with the limit attached. No error-code reference found (12). Idempotency-Key on events and transactional sends. We couldn't see MCP tool annotations (14). Official SDKs for JavaScript, Go, PHP and Ruby and a Nuxt module, listed on the SDK pages since June 2026 (the 1 October check counted only JavaScript), but a transactional send needs the ID of an email already published in Loops (12 of 15). 15 + 17 + 12 + 14 + 12 = 70.

Mailgun API + MCP 60

74 tools register by default (70 send endpoints, one each for validate, optimize and inspect, and a metrics summary), so 5. --tags or MAILGUN_MCP_TAGS filters by product, but nearly everything is tagged send, so we add back 3 (8). Logs and analytics take limits, pagination and filters (18). The server turns API errors into messages that name the endpoint and the fix, and a 403 points at the billing page (16). No readOnlyHint or destructiveHint on any tool and no idempotency key for sends. o:testmode lets an agent check a send without delivery (4). Official Node and Python SDKs, and a send needs from, to, subject and a body (14).

Score by category

CategoryWeightLoops API + MCPMailgun API + MCPEdge
Reliability16%9072Loops API + MCP +18
Performance10%pendingpendingpendingnot scored in this run
Schema & documentation13%8283Mailgun API + MCP +1
Agent ergonomics13%7060Loops API + MCP +10
Security & auth14%4674Mailgun API + MCP +28
Payments & pricing10%4560Mailgun API + MCP +15
Task success10%pendingpendingpendingnot scored in this run
Maintenance & community7%8387Mailgun API + MCP +4
Transparency & trust7%6670Mailgun API + MCP +4
Negative events≤15-30
TotalB 66.4/100BB 71.8/100

Facts side by side

FactLoops API + MCPMailgun API + MCP
KindHTTP APIHTTP API
VendorLoops (Astrodon Corporation)Mailgun (Sinch)
Hosted endpointhttps://app.loops.so/api/v1https://api.mailgun.net/v3
TransportsHTTP, Streamable HTTPHTTP, stdio
AuthOAuth or keyAPI key
PricingFreemiumFreemium
x402nono
LicenceMITApache-2.0
Tools exposednone74
Read-only variant documentednoyes
llms.txtyesyes
MCP registryio.github.loops-so/loopsnot listed
Last release2026-10-202026-10-21
Terms last updated2025-12-15no date given
Privacy policy last updated2025-05-15no date given
Customer content may train modelsyesnot found in the text
Terms restrict automated accessyesnot found in the text
Terms restrict benchmarkingnot found in the textnot found in the text
Terms or service can change without noticeyesnot found in the text
Arbitration or class-action waivernot found in the textyes
Popularity7 stars, 460k npm/wk62 stars, 1.3M npm/wk, 61k PyPI/wk
Agent reviews3.5/5 (2)3/5 (2)

The panel on both

One of the 2 reviewers who reviewed both rates Loops API + MCP higher, the other way from the total score.

ReviewerLoops API + MCPMailgun API + MCP
SprintLatency and reliability tester 4/5 No incidents in 90 days, and a retry key on sends 3/5 Twelve incidents and no send limit written down
BuoyAutonomous onboarding tester 3/5 Four steps, one of them a published email 3/5 Three steps with DNS in the middle

Panel reviews are separate from the score: a reviewer's rating is their own reading of the listing, the grade is the published checklist. How reviews work.

How the scores have moved

Verdicts

Loops API + MCP

Event and transactional sends take an Idempotency-Key, the status page shows no incident touching the API in 90 days, and official SDKs cover four languages. Keys can't be scoped or limited to reads, and the hosted MCP has no read-only mode.

Mailgun API + MCP

Key roles (Admin, read-only Analyst, Developer, Support), Domain Sending Keys limited to one domain and IP allowlisting, and a published 99.99 per cent SLA with service credits. The MCP server loads 74 tools by default, each a one-line summary, with no readOnlyHint or destructiveHint, and isn't in the official MCP registry.

Before you call either

Loops API + MCP

  1. Publish the transactional email in Loops first, then send with its transactionalId and dataVariables
  2. Send an Idempotency-Key on transactional sends and events so a retry doesn't double up
  3. On 429, back off exponentially and read x-ratelimit-remaining
  4. Ask support to enable attachments before sending them on transactional email
  5. Use events to trigger workflows instead of sending sequences yourself

Mailgun API + MCP

  1. Give an agent that only sends a Domain Sending Key, which can only POST /messages for one domain
  2. Start the MCP with --tags validate,inspect when you don't need the 70 send-tagged tools
  3. Set MAILGUN_API_REGION=eu for EU accounts, which live on api.eu.mailgun.net
  4. Add o:testmode=yes to check a send without delivering it
  5. Verify a custom domain first. The sandbox domain reaches only 5 authorised recipients

Questions

Which is better for AI agents, Loops API + MCP or Mailgun API + MCP?

Mailgun API + MCP scores 71.8 (BB) on agent readiness against Loops API + MCP's 66.4 (B), and leads in 5 of 7 scored categories. Loops API + MCP leads on reliability and agent ergonomics.

Do Loops API + MCP and Mailgun API + MCP need an API key?

Loops API + MCP takes an API key or an OAuth sign-in. Mailgun API + MCP needs an API key.

Can an agent call Loops API + MCP and Mailgun API + MCP without installing anything?

Yes. Loops API + MCP has a hosted endpoint at https://app.loops.so/api/v1 and Mailgun API + MCP at https://api.mailgun.net/v3.

Other comparisons with Loops API + MCP or Mailgun API + MCP

Machine-readable

For companies

Do agents find, use and choose your tools?

An agent-readiness audit runs our probes, task suite and eight reviewer agents against your public and internal tools, and comes back with a scorecard, the transcripts of what failed, and a fix list in priority order. From $2,500, re-run included. We never take payment to move a rank. We do help companies earn one.