Deployment responsibilities
Choose hosted or self-hosted Open Flow and configure the services your workflow requires.
The official hosted service operates an Open Flow deployment for you. A self-hosted installation makes you responsible for the Server and the external capabilities you configure. Both use the same product contracts.
Start a self-hosted Server
Follow the quickstart for a source-based Docker build, persistent storage, Workbench access, and agent configuration. The source references below cover container delivery and the complete stack with OpenConnector.
Keep Server and clients compatible with the selected model and Engine Contract. These documentation examples target package version 0.1.0-beta.77.
Configure capabilities
| Capability | Deployment responsibility |
|---|---|
| Persistence | Durable storage, retention, backup, and lifecycle cleanup. |
| Identity | Accounts, login, administrative roles, and Flow ownership checks. |
| Connector | Backend endpoint, authorized connections, account scope, and provider support. |
| Models | Available model service, credentials, budgets, and supported Task types. |
| Automatic Triggers | Scheduler, polling workers, callback origins, and provider readiness. |
| Waits and recovery | Durable decisions, supported checkpoints, expiration, and continuation. |
| Code runtime | Isolation, current Engine Contract, capability enforcement, and limits. |
Opening a Workbench or Preview does not provision these services. A workflow can be structurally valid while external execution remains unavailable.
Public endpoints
Webhook and provider event delivery need the public endpoint required by the deployment’s contract. Shared application events may also require provider-side setup and verification. Public Wait actions use scoped capabilities rather than a deployment login.
Verify your setup
Test a representative Run for each capability you intend to use. Publishing a Flow does not establish the behavior of every external account or provider event. Inspect actual Run results and provider readiness after deployment changes.