lf execute

Professional entry point for running and observing AI work on LenserFight. Commands delegate to existing RPC paths (callRpc / callRest) — no duplicate data layer in the CLI.

Execution health

lf execute status
lf execute status --json

Shows a platform-wide snapshot from LenserFight Cloud — not your local machine. In cloud mode the CLI ignores local .env SUPABASE_URL / API URLs and queries the official cloud Supabase project instead (you may see a warning about that).

Use this before dispatching workflows or battles when you want to know whether the platform is accepting new work.

Metric	Meaning
System Kill Switch	Emergency stop. inactive = normal. ACTIVE = new executions are blocked platform-wide until an admin lifts the switch.
Queue Frozen	Scheduled dispatch gate. running = queues are open and new jobs can be enqueued. FROZEN — … = admins paused scheduling; in-flight runs may still finish.
Active Runs	Workflow runs currently running, streaming, or recovered.
Queued Runs	Workflow runs waiting in queued or pending.
Active Battle Jobs	Battle AI jobs claimed or running on workers.
Queued Battle Jobs	Battle jobs waiting for a worker.
Active Workers	API workers that sent a heartbeat in the last 2 minutes.
Stale Workers	Workers with no heartbeat for > 2 minutes (possible crash or network partition).
Workflow DLQ	Workflow runs in the dead-letter queue (failed after retries, unresolved).
Battle DLQ	Battle jobs in the dead-letter queue (failed after retries, unresolved).

Reading a healthy idle snapshot

When everything is fine and nothing is executing, you typically see:

• Kill switch inactive
• Queue Frozen running (this means the queue is not frozen — confusing label, but correct)
• All run/job/worker/DLQ counts at 0

That does not mean your CLI or a daemon is running locally. It means the cloud execution plane is healthy and currently idle.

Watch for warnings after the table: kill switch active, stale workers, or non-zero DLQ counts.

Execution history

lf execute history
lf execute history --lens <LENS-UUID>
lf execute history --workflow <WORKFLOW-UUID>
lf execute history --limit 50 --offset 0 --json

Lists your recent executions in the terminal:

Mode	Flag	Source
Activity feed (default)	(none)	Team runs, schedule dispatches, and agent actions across your AI lensers
Lens prompt runs	--lens	Model/provider executions for one lens
Workflow runs	--workflow	Runs for one owned workflow

Use lf execute workflow list for the same workflow run table with optional --status / --workflow filters on the legacy path.

Workflow runs

lf execute status
lf execute workflow list
lf execute workflow inspect <RUN-UUID>
lf execute workflow wait <RUN-UUID>
lf execute workflow stream <RUN-UUID>    # SSE-style event tail
lf execute workflow cancel <RUN-UUID>

lf execution … remains available; prefer lf execute workflow … for new scripts.

Streaming events

lf execute workflow stream <RUN-UUID> --timeout 600
# or legacy:
lf execution events <RUN-UUID> --follow

Polls lenses.workflow_run_events and prints token chunks with terminal FX while you wait.

Battles

lf execute battle exec <battle-id> …
lf execute battle dispatch <battle-id> …
lf execute battle file-run <slug>
lf execute battle jobs <battle-id>

Lens / prompt

lf execute lens prompt --provider openai --model gpt-4o "Hello"
lf execute prompt …    # alias

Uses @lenserfight/providers for BYOK (env), Ollama, or cloud RPC.

Team runs

lf execute team dispatch …
lf execute team list

Dashboard

From lf (TUI): press e for the Execute sub-dashboard, k for Configure (BYOK / Ollama / providers).

See also lf configure and CLI dashboard.