States Language, self-hosted¶

Stegflow runs Step Functions-style state machines on your own infrastructure. It speaks the AWS States Language and is wire-compatible with the Step Functions SDK, so existing definitions and activity workers run unchanged.

Self-hosted AGPL-3.0 AWS-compatible

Get started

What is it¶

A state machine describes a workflow as a set of steps (tasks, choices, parallel branches, waits) and the transitions between them. Stegflow runs these workflows: you define them in JSON, and the engine executes them end to end, handling retries, timeouts, branching and parallelism, with a console to author and watch every run.

Workflows are written in the Amazon States Language, the same portable JSON format used by AWS Step Functions. If you already have Step Functions definitions, they run on Stegflow unchanged. If you have never used Step Functions, it is simply a well-specified way to describe workflows, running on your own infrastructure.

Features¶

Workflow engine

All eight state types with full semantics, including Map with concurrency & failure tolerance, and Parallel branches.
- JSONPath + JSONata, full parity across both query languages
- AWS intrinsic functions
- Retry, Catch, timeouts & heartbeats, exponential backoff + jitter
Security & multi-tenancy

Hard tenant isolation by design, with fine-grained, policy-based authorization.
- OpenID Connect (OIDC) single sign-on
- Users, groups, roles & policies, per tenant and per resource
- Per-tenant credentials & branding
- Invitations & admin impersonation
Lifecycle & deployment

Ship state machines safely, roll out gradually, and schedule them like cron jobs.
- Immutable versioning of state machines
- Aliases + canary deploys, weighted routing between versions
- Scheduled executions, cron + timezone
- Tags to organize at scale
UI & observability

Author, run and inspect workflows from a built-in web console.
- Interactive graph visualizer
- Real-time execution monitoring
- Built-in code editor
- Complete execution history
- Execution RSS feeds
- Prometheus metrics
Integrations

Connect tasks to the outside world, and compose workflows from workflows.
- HTTP Service, call any external API
- Activities, long-running workers, task tokens, heartbeat
- Nested workflows, compose workflows from workflows
- Custom service resource handlers
Self-hosted & sovereign

Run entirely on your own infrastructure, on-prem, private cloud, or air-gapped. Your workflow data never leaves your network.
- No vendor lock-in, open standards, portable definitions
- Predictable cost, no per-transition billing
- Air-gap friendly - no external dependencies required

See it in action¶

Author, run and monitor workflows from the built-in web console. Click any screenshot to open it full size.

Open source¶

Built in the open, AGPL-3.0. The engine is solid and the docs are growing. Follow progress on Codeberg, open issues, and help shape the roadmap.

View on Codeberg