Authentication & Authorization

MSPs deal with three kinds of authenticated traffic — humans, AI agents, and API integrations — and most platforms treat them as three separate problems. Breeze treats them as one. OAuth 2.1 covers AI clients. Scoped API keys cover automation. MFA and email verification cover the humans. Every path produces the same audit trail and respects the same tenant boundaries.

OAuth 2.1 for AI Clients

External AI assistants — Claude.ai, ChatGPT, Cursor, Windsurf — connect to Breeze through a full OAuth 2.1 stack with Dynamic Client Registration and PKCE. Clients self-register on first connect, the user approves scopes through a standard authorization screen, and the client receives short-lived access tokens with rotating refresh tokens.

Grant revocation is immediate. When you de-authorize an AI client, every sibling token tied to that grant is invalidated in the same request — no waiting for expiry, no window where a stolen token still works. Per-grant rate limits run alongside per-key limits so a misbehaving client is throttled and visible, not silently dropped.

API Keys with Risk Tiers

For machine-to-machine traffic that doesn’t fit an OAuth flow, Breeze issues API keys with explicit scopes that map directly onto the 4-tier risk engine.

There is no admin-override key that bypasses risk classification. Keys are tenant-scoped — a key for Tenant A cannot reach Tenant B regardless of which tool is called.

MFA and Email Verification

Breeze supports TOTP-based MFA for every account, with enrollment through any standard authenticator app. MFA setup, enable, and disable flows all confirm the current password before mutating credentials, and TOTP secrets at rest are encrypted under a dedicated key separate from session signing.

New signups must verify their email before the tenant fully activates. The verification flow distinguishes “this link was replaced” from “this link was already used” so the message matches reality, and resend is one click. When verification clears after a payment method is attached, the partner activates automatically — no support ticket required.

Tenant Isolation, Down to the Database

Authentication establishes who you are; authorization decides what you can reach. Every tenant-scoped table runs with row-level security enforced at the database, so an application-layer bug cannot leak data across organizations even if a query accidentally omits a tenant filter.

That guarantee holds the same way regardless of auth path — OAuth grant, API key, or session.

Audit Parity Across Every Path

Whether an action comes from the web UI, an OAuth-authenticated AI client, or an API-key-driven script, the audit entry is identical: actor identity, tool or endpoint called, parameters, tenant context, risk tier evaluation, and outcome.

The audit pipeline reads the trusted client IP from the configurable reverse-proxy chain, so anomaly review and abuse investigation see the real source address instead of an internal proxy IP. Forged forwarding headers cannot fool the audit log or the rate limiter.

Why It Matters for MSPs

One identity story to explain to a compliance auditor. One place to revoke access when something looks wrong. One set of guarantees that hold regardless of who — or what — is calling. That’s the difference between a platform built for AI-era operations and one with auth bolted on after the fact.

Learn More

• MCP Server — How OAuth-authenticated AI clients reach Breeze tools.
• AI Assistant — Risk tiers, approvals, and the governance layer scopes plug into.
• Configuring Breeze AI: The Self-Hoster’s Guide — Full configuration reference for OAuth, MFA, and API keys.