Expand description
Workflow spec data model.
A workflow is a set of branches, each a DAG of nodes wired by
edges. A node is a typed expression: its node_type is a namespaced
verb (ingress.cron, transform.state_append, decision.consensus_voting,
guard.permission, execute.publish, sink.log, …) and
its config is the expression’s parameters. Composing these typed
expressions into a DAG is the whole programming model — the same shape as the
Python agent_core/workflow JSON specs, so existing builtin specs
deserialize unchanged.
This module is the data model only. The executor (scheduling branches,
resolving edge order, dispatching node types, threading run state) is the
next porting phase and will live in a sibling runtime module.
Structs§
- Branch
- One DAG within a spec. The root branch is conventionally
__root__. - Edge
- A directed wire from one node’s output to another’s input.
- Node
- A typed node expression.
- Spec
- A complete workflow specification.
Enums§
- Spec
Error - A spec is structurally invalid.