The identity stack
Independent protocol layers and the configuration that joins them.
OpenAgentID is a family of independently useful libraries. A small facade connects them, but the providers, trust roots, persistence and policy remain explicit application choices.
| Need | Start here |
|---|---|
| Identify an entity | OAS |
| Verify a capability | ACT quickstart |
| Broker credentials | Arsenal |
| Verify authority and policy | AEGIS |
| Authenticate a peer | Transport comparison |
| Govern a skill/tool | Skills policy |
| Integrate with a framework | MCP and lifecycle hooks |
| Choose a runtime | Six language SDKs |
Compose deliberately
First identify the action and trust boundary. Choose token verification, identity authentication, or privileged authority verification accordingly. Connect the relevant provider, select its storage and configuration, then exercise success and denial paths. Add credential brokering only where the action needs a provider secret.
The source includes six SDK languages with different versions and API coverage. The catalog is the version source of truth for this documentation snapshot. No single facade call proves every underlying subsystem has been provisioned or is live.