Build on
Runtime Authority
You're integrating a runtime that sits between your AI agents and the systems they act on -- evaluating every signed action against your organization's delegated authority before it executes, and handing back a decision and its evidence.
Where Runtime Authority sits
Your agent never calls your enterprise system directly. It submits a signed intent to Runtime Authority, receives an Allow, Deny, or Human Review decision, and only proceeds if the decision allows it. See the full Architecture for how each component in between fits together.
Getting Started
Create an organization, register an agent, and get your first Authorization Receipt back.
Architecture
The full path from an agent's signed intent to a decision and its evidence, component by component.
Runtime API
The Intent API: POST /v1/intents, its three possible outcomes, and the request/response lifecycle.
Authentication
The Operator Key, per-user roles and permissions, per-developer API keys, and agent certificates.
Authority Graph
Delegated authority, roles, resources, and approval limits, and how governance becomes machine-readable.
Runtime Policies
How a policy is compiled, versioned, and deployed, and why that differs from the governance document it came from.
Authorization Receipts
Why logs aren't enough, the receipt lifecycle, and the planned architecture for independent verification.
Evidence Verification
Signature verification, evidence export, and audit workflows against the Evidence Portal today.
SDKs
The Python SDK, and the language roadmap for Node.js, Go, Java, .NET, and Rust.
Webhooks
Enterprise events -- IntentReceived, DecisionCompleted, ReceiptIssued, and more -- with real payloads.
Integration Examples
Rotating keys, sending a heartbeat, and retiring an agent, with real SDK code for each.
Integration Guides
Connecting Runtime Authority to agent frameworks, model providers, and enterprise systems.
Agent Registration
The full agent lifecycle: registered, active, suspended, retired, or revoked.
API Reference
Endpoints, schemas, headers, status codes, errors, and versioning.
Interactive API Reference
Generated directly from the running platform, not a static mirror that can drift out of date.