Skip to main content
Smithers 0.25.5 makes a workflow run tell you what it produced. Previously bunx smithers-orchestrator workflow run hello finished with output: null even though the agent had produced a greeting, and the Get Started promise of “you see the output” did not hold. Now the run summary surfaces the workflow’s output, and every seeded workflow ends with a task that emits a useful result.

​
CLI

The workflow run / up summary now prints the run’s output cleanly. The run output is the last task’s result; the summary strips the internal run_id / node_id / iteration columns, unwraps a single row, and omits the output field entirely when there is nothing to show, so a successful run never prints a noisy output: null.

​
Workflows

Every seeded init-pack workflow now ends with a deterministic output task that aggregates its genuinely useful terminal result, so running one prints something meaningful instead of nothing:
  • hello surfaces the greeting it produced.
  • create-workflow / create-skill / extract-skill surface the files written and the skill path.
  • route-task and context-engineer surface the chosen route, the recommended workflow, and a summary.
  • triage-run, monitor, monitor-smithers, and context-doctor surface the run state, health, root cause, recommended action, and report path.
  • backpressure-plan, eval-author, report-slideshow, and smithering surface verdicts, counts, written paths, and (for smithering) the files changed and PR url.
The additions touch only the final output task; no existing task logic, ids, or agents changed, and all init-pack workflows still render. The legacy-inline workflows (implement, plan, review, and the rest) get the same treatment in a later release.

​
Migration notes

No breaking changes are identified in 0.25.5. Workflow authors who relied on RunResult.output already received the last task’s output; the CLI now simply presents it without internal columns. Add a final output task to your own workflows to control what their run prints.