AgentAuth

AgentAuth gives an AI assistant identity and authorization tools for agents: credentials, tokens, delegations and permission checks against an AgentAuth service

What you can do

Authenticate an agent against the AgentAuth service, using a default API key or a key passed per call, to obtain credentials or tokens. Manage agent credentials and tokens for identity and authorization. Create and handle delegations between agents. Run permission checks to see whether an action is allowed. All operations go to the AgentAuth service at the configured base URL.

When to use it

Use it when an assistant needs to act as an identified agent against an AgentAuth deployment, for example to obtain tokens, delegate access or verify permissions before performing a task. It is only useful if you already run or subscribe to an AgentAuth service.

Requirements

Runs as a local Python package (agentauth-mcp-server) over stdio on a desktop machine. Requires the AGENTAUTH_URL environment variable pointing at the AgentAuth service base URL, such as a hosted instance or a self-hosted localhost address. Optionally set AGENTAUTH_API_KEY as the default key for the authenticate tool; a key can also be supplied per call. Network access to that service is needed.