Agents
Agents are AI tools — like Claude Code, ChatGPT, or custom scripts — that can interact with your Broadcast instance programmatically. The Agents feature provides a discovery and onboarding layer so these tools can self-configure and operate Broadcast on your behalf.
Why Use Agents?
Instead of manually navigating the UI for every operation, you can instruct an AI agent to:
- Manage subscribers — add, tag, segment, and export contacts
- Send broadcasts — draft, review, and send email campaigns
- Monitor performance — check open rates, click rates, and delivery stats
- Manage sequences — create and configure automated email workflows
- Troubleshoot — diagnose delivery issues and check email server status
Three Ways to Connect
1. CLI (broadcast command)
A lightweight command-line tool that wraps the Broadcast API. Install it with a single curl command, authenticate with your API token, and start managing Broadcast from your terminal or AI agent.
Best for: Terminal-based workflows, shell scripts, and agents that can execute commands.
2. REST API (Agent Endpoints)
Broadcast provides dedicated API endpoints for agent discovery and onboarding:
GET /api/v1/whoami— Identify the current token and its permissionsGET /api/v1/status— Check channel health and readinessGET /api/v1/prime— Get a full capabilities manifest with available endpoints
These complement the existing 50+ API endpoints for subscribers, broadcasts, sequences, and more.
View the Agents API Reference →
3. MCP Server (Claude Code)
A Model Context Protocol server that exposes Broadcast operations as native Claude Code tools. Configure it once and Claude Code can directly manage your email marketing.
Best for: Claude Code users who want deep, native integration.
Quick Start
The fastest way to connect an agent is the copy-paste introduction:
- Go to Agents in your Broadcast sidebar
- Copy the Introduction Message
- Paste it into your AI agent (Claude Code, ChatGPT, etc.)
The introduction message includes your instance URL, API token, permissions, and quick-start commands — everything an agent needs to get started.
Setup Links
For a more secure onboarding flow, you can generate a Setup Link — a time-limited URL that serves agent configuration instructions. Share the link with your agent, and it will self-configure with the correct credentials and permissions.
Setup links expire after 1 hour and can only be accessed 3 times, so they’re safe to share in chat conversations.
Learn more about Setup Links →
Security
- API tokens control what agents can access — use the principle of least privilege
- Setup links are time-limited and access-limited
- The
whoamiendpoint never exposes the full token value - The
statusendpoint never exposes sensitive configuration (API keys, SMTP passwords) - All agent communication happens over HTTPS