Good questions.
Clear answers.

What is Aphura?

Aphura is a security control platform for AI agents. It brings together agent identity, permissions, signed policy, human approvals, tool-call inspection, and audit evidence so organizations can govern actions routed through it.

Does Aphura replace my AI model?

No. Aphura governs the agent’s access to tools and systems. It works at the action boundary, alongside the models and agent frameworks you choose.

How does the MCP gateway work?

The customer-local gateway sits between an MCP client and an MCP server. It checks reviewed tool definitions, inspects traffic, and authorizes calls. Execution mode adds exact-operation human approvals, action budgets, and audited handoffs.

What happens when an agent is suspended?

Aphura can suspend future platform-mediated access and revoke recorded credentials and delegations. It does not automatically revoke every external cloud credential, stop a bypassing agent, or undo an upstream action already in progress.

Where does Aphura run?

The hosting foundation is designed for Azure using AKS, Azure Database for PostgreSQL, Key Vault and Blob Storage. Enforcement runs in a customer-local environment. The agent and gateway can run in separate pods, with mutual TLS and reviewed network restrictions.

What stage is the platform at?

Aphura is in active development. The core platform and MCP workflow have local integration and browser coverage. Real Azure deployment, enterprise identity-provider qualification, recovery exercises, and customer acceptance remain ahead.