Orchestrate a Task

On hosted NFLTR at nfltr.xyz, one task is a bounded unit of work. This guide covers start task → task watch → task result, plus a flaky-test example.

What every dispatch carries

Objective, explicit task_mode or execution_profile.mode, optional descriptive actor metadata, worker binding (--worker / --require-label), and explicit JSON coordination statuses.

Step 1 — Dispatch

$ nfltr orch start task --worker MacBook-pro.dev-impl --task-mode implement --objective "Document orch task result in cli.html" --watch

$ nfltr orch start task --worker MacBook-pro.dev-impl --task-mode inspect --objective "..." --dry-run

Step 2 — Watch status

$ nfltr orch task watch --task task-1779278291162547000

Step 3 — Read the result

$ nfltr orch task result --task task-1779278291162547000

Example: fix a flaky test

$ nfltr orch start task --worker MacBook-pro.platform-impl --task-mode implement --objective "Stabilize TestWidgetCache; run go test ./pkg/widget" --timeout-ms 900000

$ nfltr orch task status --task task-impl-flaky-cache --events 5

$ nfltr orch task result --task task-impl-flaky-cache

Blocking questions

Answer status=question from the planner or dashboard; avoid duplicate dispatches on the same repo path until the first task is terminal.

Dashboard

Runs on the Causal Operations Atlas at nfltr.xyz/dashboard mirror CLI state — attention rail, causal canvas, task drawer, and Trace Lab. See Observe orchestration for inner-LLM detail.

Next: Multi-agent orchestration.

Practical tips

NFLTR persists task history on the hosted control plane at nfltr.xyz, so reconnecting planners and dashboard viewers see the same timeline. Share runs with teammates using task ids and dashboard deep links instead of pasting full JSON envelopes. For production-impacting objectives, configure approval gates so a human confirms in the drawer before work continues. Combine the CLI examples above with the dashboard when you need mobile monitoring or inner-LLM turn detail documented in the observe orchestration guide.