Implementation guides

AI Agent Authorization Integrations

Choose the framework or service boundary where you need to enforce an agent's delegated authority. Each guide shows the relevant package, verification point, scopes, and current limitations.

Grantex is an open-source delegated authorization and trust layer for AI agents. It gives each agent a verifiable identity and scoped, time-limited, revocable authority from a human or organization, with multi-agent delegation, service-side verification, and audit records.

Grantex complements OAuth 2.0 and MCP: OAuth handles application and user authorization, MCP connects models to tools, and Grantex proves which agent may perform which action for which principal.

Framework and API guides

Use the primary TypeScript or Python SDK for core grants and verification, then add a framework adapter when it reduces integration work. Review the release-status page before production use.

LangChain

Scoped tool authorization for LangChain agents and tool calls.

Integration guide →

CrewAI

Role-specific grants for agents working together in a CrewAI crew.

Integration guide →

OpenAI Agents SDK

Scope enforcement and optional audit wrappers for Python agent tools.

Integration guide →

Anthropic SDK

Verify grants before Claude-powered applications execute tools.

Integration guide →

Google ADK

Python authorization helpers for Google Agent Development Kit tools.

Integration guide →

AutoGen

Per-agent permissions for AutoGen multi-agent workflows.

Integration guide →

Vercel AI SDK

TypeScript scope enforcement around AI SDK tool execution.

Integration guide →

MCP

Grantex management tools for MCP clients and documented auth choices for MCP servers.

Integration guide →

Express.js

JWKS-backed grant verification and per-route scopes for Express APIs.

Integration guide →

FastAPI

Typed FastAPI dependencies for token verification and route scopes.

Integration guide →

MPP

Verifiable agent-passport identity for machine-payment requests.

Integration guide →

Additional ecosystem integrations

These packages have documentation but do not yet have separate landing pages.

Strands Agents SDK

TypeScript and Python helpers for verifying grants before Strands tools run.

Documentation →

A2A agent-to-agent authorization

Propagate and verify scoped grant tokens across Agent2Agent task requests.

Documentation →

Where should authorization run?

At the tool or API boundary

  • Verify the JWT signature and claims using JWKS.
  • Require the exact scope before execution.
  • Use online or synchronized state when current revocation matters.

Inside the agent application

  • Register the agent and request a scoped grant.
  • Carry the grant token to the receiving service.
  • Add an audit wrapper when you need tool-execution records.