Have an AI agent? Let it set up Breeze.
Breeze is the first RMM platform with a native Model Context Protocol endpoint. Point Claude, ChatGPT, or any MCP-capable agent at our server and describe what you want. The agent handles signup, payment, baseline configuration, and staff invitations — typically in under 5 minutes. No human required in the web UI.
Try it in Claude today
Connect the Breeze MCP server in Claude.ai, then paste this prompt — you'll have a working Breeze tenant before your coffee cools.
Set up Breeze RMM for my company. My admin email is [email protected]. Prefer to skip the agent? Start Breeze Cloud Beta the regular way →
MCP endpoints
Pick the region closest to your admin. Each endpoint provisions tenants in its own region only.
| Region | Endpoint URL | Transport |
|---|---|---|
| US | https://us.2breeze.app/api/v1/mcp/message | HTTP (JSON-RPC) |
| EU | https://eu.2breeze.app/api/v1/mcp/message | HTTP (JSON-RPC) |
Example prompt
Paste this into Claude.ai (with the Breeze MCP server connected) or any MCP-capable agent. Substitute real email addresses you control.
Set up Breeze RMM for Acme Corp. The admin email is [email protected] — that's a business address. Once it's running, send install invites to [email protected] and [email protected], then poll for enrollments and tell me when both devices are online with a health summary. What happens end-to-end
- 1
Agent calls create_tenant
Business email validated, tenant row created, activation email dispatched. No credentials required.
- 2
Admin clicks the activation link
Lands on a hosted page, confirms the org, and proceeds to Stripe Checkout for a payment method (no charge — setup mode).
- 3
Agent polls verify_tenant
Once status flips to active, the agent gets an API key with full scope to continue.
- 4
Agent provisions baseline
configure_defaults applies a sensible alert policy, device group, and notification channel. send_deployment_invites emails install links to the staff addresses the user supplied.
- 5
Devices enroll, agent reports
As devices come online, the agent polls get_fleet_status and get_fleet_health to synthesize a first health report — typically under 5 minutes wall-clock from prompt to finding.
The three bootstrap tools
These tools are callable without any API key. Everything else requires the tenant's API key (returned by verify_tenant) and, for mutations, a payment method on file.
create_tenant Creates a new Breeze tenant. Takes org_name, admin_email (business domain), admin_name, region. Sends an activation email to the admin and returns a tenant_id.
verify_tenant Polls activation status. Returns pending_email → pending_payment → active. Once active, returns an API key scoped to the new tenant.
attach_payment_method Returns a Stripe Checkout URL for the admin to add a payment method. Required before any mutating tool (invites, configuration) can be called.
FAQ
Is this safe? What can the agent actually do?
Before payment is attached, the tenant API key is readonly — the agent can only verify status. Mutating tools (invites, configuration changes, delete_tenant) require a payment method on file. Every agent-initiated action is audited with actorType=api_key and initiated_by=integration so you can see exactly what happened.
What data does the agent see?
Only data for the tenant it provisioned. Breeze is multi-tenant with row-level security isolation at the database layer, so even a compromised API key cannot read other tenants' data.
Can I delete a tenant an agent created?
Yes — the agent itself can call delete_tenant with a typed confirmation phrase (soft delete with a 30-day restore window), or you can delete via the web UI or support.
Why business-email-only for admin?
Anti-abuse. Free-provider (gmail, outlook, yahoo) and disposable-email domains are rejected at signup. If you legitimately need a free-provider admin, contact support.
What about self-hosted Community edition?
Self-hosted Breeze does not enable the MCP bootstrap flag by default — it's a SaaS-only feature. Self-hosters who want agent-driven setup can flip the MCP_BOOTSTRAP_ENABLED env var and provide their own Stripe/email credentials.
For agent authors
These machine-readable files describe Breeze to agents that crawl the site:
- /.well-known/mcp.json — MCP discovery manifest (servers, auth, tools, rate limits)
- /llms.txt — Product context for LLM search engines
- /pricing.md — Structured pricing for agents evaluating Breeze on behalf of a user