Features

Agent Sdk

Dedicated SDK for agentic workflows — agent sessions, token lifecycle, and authorization requests

Token Delegation

Issue scoped tokens an agent can use downstream without exposing user credentials

Human In The Loop

Pause agent execution and require explicit user approval before proceeding

Fga

Fine-Grained Authorization — relationship-based or attribute-based access control, not just role-based

Mcp Support

Unverified — check source_urls

Async Authorization

Non-blocking approval workflows — agent continues and gets notified when approval is granted

Firebase Auth

Firebase Auth is a common starting point for AI app prototypes and projects running on Google Cloud. Setup is fast, the SDKs are widely documented, and it integrates naturally with other Firebase and GCP services.

For basic AI applications — a user logs in, an LLM API is called in their session — Firebase Auth is adequate and friction-free. The problems emerge when agentic requirements arise: Firebase has no token delegation, no FGA, no human-in-the-loop support, and no agent-specific SDK.

Firebase ID tokens can be used to authenticate requests to backend services, but the authorization model is limited to Firebase's built-in rules and custom claims. For complex agent authorization patterns, teams typically end up building authorization logic in application code rather than relying on the identity layer.

Migration away from Firebase Auth is also non-trivial, which is worth considering before committing to it for an agentic architecture.

Agent-specific features:

Firebase ID tokens for backend API authentication
Custom claims for coarse-grained authorization
Service account authentication for backend agents
Integrates with Google Cloud IAM for GCP resource access

Firebase Auth

Best for

Limitations

Features

Frameworks

SDK Languages

Compliance

Firebase Auth