Workflow Engine

Orchestrate with accountability.

The workflow engine coordinates jobs, approvals, retries, and the audit trail for every run. It ensures every action is tracked and governed.

Execution

Robust primitives.

Built for reliability. Approvals, retries, and schema validation come standard.

DAG execution with depends_on

Approval, delay, notify, and condition steps

Retries with exponential backoff

for_each fan-out and aggregation

Schema validation for inputs/outputs

Rerun from step and dry-run mode

Run Audit Trail

Append-only audit trail per run
Run events and step states
Decision records and approvals
Correlation with job trace IDs

Every run emits events and metrics for operators and auditors, ensuring complete observability.

workflow.yaml

name: incident-triage
steps:
  triage:
    type: worker
    topic: job.incident.enrich
  
  approval:
    type: approval
    depends_on: [triage]
    reason: "Review before remediation"
    
  remediate:
    type: worker
    topic: job.incident.remediate
    depends_on: [approval]
    constraints:
      max_lines_changed: 500
      max_runtime_sec: 900

Definition

Explicit, reviewable workflows.

Runs stay predictable because every dependency, approval, and constraint is declared in code.

Approval GatesApproval steps pause the run until cleared by an authorized human or system.

Explicit DependenciesExecution flow is defined via depends_on, making the DAG clear and deterministic.