Introduction
Actera (actera.dev) is an authorization and runtime control layer for AI agents. It evaluates a proposed action against business policy before execution and returns one of three effects:
- Allow: the integration may continue.
- Require approval: the action must pause until an authorized person approves or rejects it.
- Block: the integration must not perform the action.
Actera sits at the point where your trusted application can still stop an action. Your application supplies the agent identity, action, and relevant context. Actera applies the policies assigned to that agent and records an authoritative decision for real enforcement requests.
Who Actera serves
Section titled “Who Actera serves”Platform and security teams use Actera to define reusable controls. Application teams integrate authorization at action boundaries. Operations teams inspect decisions, approvals, and reported follow-up activity without reconstructing an outcome from logs spread across systems.
A practical starting point
Section titled “A practical starting point”Begin with one agent and one sensitive action. Complete the TypeScript Quickstart, then configure and assign a policy from behavior you can test.
Continue with How Actera works for the full enforcement path. Integration and policy details are covered in MCP authorization, decisions and human approval, stateful usage limits, and policy management and assignment.