Build against the NatorOS operating layer.
Use the public API, OpenAPI description, OAuth metadata, MCP resources, and sandbox endpoints to integrate with governed AI workers and workflows.
Quickstart
The shortest path for a new integration or agent crawler.
- 1Open
Read the product context
Start with
/llms-full.txtso your agent understands what NatorOS is and is not. - 2Open
Inspect the schema
Use
/openapi.jsonfor endpoints, response shapes, OAuth scopes, and error models. - 3Open
Check auth metadata
Use
/.well-known/oauth-protected-resourcefor OAuth protected-resource discovery. - 4Open
Test safely
Use the sandbox workflow-run endpoint before touching production workspaces.
Core surfaces
All production writes require scoped credentials. Public sandbox endpoints never read or write customer data.
GET /api/v1/agentsDiscover available AI workers and their capabilities.
GET /api/v1/workflowsList governed workflows available to a workspace.
GET /api/v1/runsRead workflow-run status, history, and async job progress.
POST /api/v1/sandbox/workflow-runsSafely test a workflow run without customer data.
POST /api/v1/webhooksRegister signed event delivery endpoints.
Scopes
Request only what your integration needs.
agents:readagents:writeworkflows:readworkflows:writeruns:readrecords:readrecords:writewebhooks:write