Mastra agents can now send and receive iMessages via Photon Spectrum by installing the @photon-ai/chat-adapter-imessage, adding it to the agent’s configuration, and providing Photon credentials; Photon provisions dedicated iMessage numbers in the cloud, handling delivery, retries, and presence, while Mastra offers declarative channel setup, deduplication, and full traceability, enabling seamless, scalable iMessage integration without needing a Mac or personal Apple ID.

"Is there an iMessage API?" is one of the most-asked questions in our Discord, and the honest answer: *no official one exists, but here's how it actually works* — deserves more than a one-line reply. At Photon, we operate iMessage infrastructure that handles millions of API calls a day, which means we know exactly where the real protocol ends and the spoofed blue bubbles begin. This guide is the full answer: why Apple ships no public API, how managed infrastructure closes the gap, what native features you can genuinely send, and how to vet a provider without getting burned by an SMS gateway in a blue costume.

At Photon, we spend all day inside the plumbing of consumer messaging. Spectrum, our open-source framework, routes millions of messages daily across iMessage, RCS, SMS, WhatsApp, and Telegram, which means every routing decision, encoding quirk, carrier fallback, and Apple platform change lands on our infrastructure first, before it ever becomes a blog post. So we wrote the guide we wish we could just link people to. It covers how each channel actually works under the hood, where the support boundaries really are as of mid-2026, and when each one is the right tool. Yes, we build a product that makes the answer "all three, automatically". We'll show you how at the end. But the first 90% of this guide is channel-agnostic and useful whether you ever touch Spectrum or not.

A Mac mini in our fleet became unresponsive after 37 days of uptime. The kernel had wired 15.1 GB of the machine’s 16 GB of memory. Only 67 MB remained free, the load average was 250, and the system could no longer fork a process. Photon runs Mac hardware continuously for our iMessage API relays. We found that every Mac we checked was accumulating wired memory at a steady rate. At the observed rates, the fastest machine would exhaust its memory in about five weeks. The cause was a kernel memory leak triggered by nested `#!` scripts. The leak can be reproduced with six lines of shell. It does not require administrative privileges, and the allocated memory is released only after a reboot.

NanoClaw now offers hosted iMessage support through Photon, eliminating the need for a local Mac, personal Apple ID, or hardware management; agents can send and receive texts, attachments, reactions, typing indicators, read receipts, and effects on any OS via Photon’s cloud infrastructure, with simple setup using a Photon account and NanoClaw’s setup commands.

Photon processes over 10 million iMessage API calls per day across production deployments for companies like Rho, Vercel, Hermes Agent, Ditto, and FlipText. The platform achieves the same uptime, delivery speed, and feature coverage as Apple's own iMessage service — not through incremental improvements to the standard Mac-bridge architecture, but through a fundamentally different technical approach. This post breaks down the three engineering layers that make that possible: protocol-native Apple communication, Astrolabe (our open-source macOS fleet manager), and enterprise cloud routing on AWS EKS.

Your iMessage lines now scale themselves. Route users to the healthiest line with a single API call.