Module executor 

Workflow executor — runs compiled DAG with level-based parallel execution.

The executor runs a WorkflowDag by:

1. Processing levels sequentially (level 0 → level 1 → … → level N)
2. Within each level, running all steps concurrently (since they have no mutual dependencies)
3. Tracking step results and passing context between levels
4. Handling timeouts, retries, and approval gates
5. Supporting {{variable}} template interpolation in tool_args and prompt fields
6. Executing LLM inference for prompt-based steps when a provider is available

Step results are accumulated in a WorkflowContext that flows between levels. Template interpolation replaces {{step_id_output}} and {{step_id}} with previous step outputs, and {{variable_name}} with context variables.

Structs

StepResult

The result of a single workflow step execution.

WorkflowContext

Workflow execution context — shared state that flows between levels.

WorkflowExecutor

Workflow executor — runs a compiled DAG.

WorkflowResult

The result of the entire workflow execution.

Enums

StepStatus

The status of a workflow step execution.

Functions

interpolate_template

Interpolate {{variable}} template patterns in a string.