Give your AI agents access to your notification infrastructure with the Courier CLI
Mike MillerMarch 12, 2026
Give your AI agents access to your notification infrastructure with the Courier CLI
We rebuilt the Courier CLI (command-line interface) from scratch, designed for both developers and AI agents. The previous version covered a subset of the API. The new one covers the full Courier API with a consistent courier [resource] <command> [flags] pattern across messages, profiles, lists, tenants, automations, preferences, bulk jobs, and more. Whether you're working in Cursor, Claude Code, or a CI pipeline, every Courier operation is now a terminal command away.
What's new in the Courier CLI
It's a standalone binary for macOS, Linux, and Windows with no runtime dependencies. Install via npm (npm install -g @trycourier/cli), which downloads a platform-specific binary through a postinstall step, or grab it from GitHub Releases.
Every command supports --format json for machine-readable output and --transform for filtering responses with GJSON syntax. Run --help on any command to see its flags and usage, which makes the CLI self-documenting for agents that discover tools at runtime. See the CLI docs for the full reference.
How AI agents send and manage messages
The real power isn't any single command. It's that agents can chain commands together, using structured output from one step to decide what to run next.
Send messages from the terminal
You tell your agent to send a message, and it translates that into a command:
courier send message \--message.to.email "alex@example.com" \--message.content.title "New login detected" \--message.content.body "We noticed a new login to your account from San Francisco. If this wasn't you, reset your password immediately."
That's inline content, but agents can also reference templates, add multi-channel routing with fallbacks, include dynamic data, or send to entire lists. The same command works across email, SMS, push notifications, in-app notifications, Slack, Microsoft Teams, WhatsApp, Discord, and webhooks. One command, any channel.
Debug delivery issues
This is where agents and the CLI really click. Your teammate says a user didn't get a message. Instead of opening the dashboard, you ask your agent to look into it:
- The agent searches recent messages for the user
- Finds the message ID and retrieves its status:
UNROUTABLE - Checks the user's profile. No email on file.
- Tells you the root cause and offers to fix the profile
Each step is a CLI command with --format json output that the agent parses to determine the next move. The entire investigation happens in your editor.
Manage users, lists, and tenants
Agents can create user profiles, subscribe users to lists, set up tenants with custom branding, and update notification preferences. All the operational work that normally means switching to the dashboard or writing one-off scripts.
See the CLI docs for the full command reference across all these resources.
Setting up the CLI for your AI agent
Install the CLI globally and set your API key:
npm install -g @trycourier/cliexport COURIER_API_KEY="your-api-key"
That's the entire setup. No config files, no auth flows, no SDK initialization. If your shell has the API key, your agent has access.
In Cursor
Cursor's agent mode can run shell commands directly, so the CLI works out of the box once installed. To get more out of it, add a project rule that tells the agent the CLI is available and points it at the CLI docs for command reference.
In Claude Code
Claude Code has direct shell access, so the same applies. Set COURIER_API_KEY in your shell profile (.zshrc, .bashrc) so it persists across sessions, and add a note to your CLAUDE.md project instructions letting the agent know the CLI is available.
MCP as an alternative
If you'd rather give your agent structured tool access instead of shell commands, we also have an MCP server that exposes the same Courier operations through the Model Context Protocol. Both Cursor and Claude Code support MCP. It's a different integration path to the same capabilities.
Get started with the Courier CLI
Beyond agents, the CLI is also the fastest way to interact with Courier from your terminal for scripting, CI/CD smoke tests, and debugging on the fly. The CLI docs have recipes and examples for all of it.
Install it, set your API key, and start using it.
Similar resources
Courier Skills: teach your AI agent to build with Courier
Courier Skills is a free, open-source knowledge base that teaches your AI coding agent how to build and debug notifications with Courier: the right primitive for each use case, the exact payload shapes, and the mistakes that produce unhelpful errors. The rebuilt version is smaller, leads every reference with the failure modes, and installs on any agent that reads skills with one command: `npx skills add trycourier/courier-skills`.
How Apple's on-device AI works, and what it changes for your users
Apple's 2026 on-device models (AFM 3) are good enough to read, rank, and summarize everything that lands on your phone, locally and for free. Here's how they actually work, in plain terms, and how a model that reads every message before your users do changes what you should send.
You can build anything now. That's exactly why you shouldn't build this.
AI made building software cheap, so the temptation is to build everything. But owning infrastructure is a permanent draw on your scarcest resource, attention. The value lives at two ends, the systems everything runs on and the product only you can make; the move is to build less, buy the opinionated platform, and let an agent operate it.
© 2026 Courier. All rights reserved.