- Isolated cron runs also treat run-level agent failures as job errors even when no reply payload is produced, so model/provider failures increment error counters and trigger failure notifications instead of clearing the job as successful.

- When an isolated agent-turn job reaches `timeoutSeconds`, cron aborts the underlying agent run and gives it a short cleanup window. If the run does not drain, Gateway-owned cleanup force-clears that run's session ownership before cron records the timeout, so queued chat work is not left behind a stale processing session.

- If an isolated agent-turn stalls before the runner starts or before the first model call, cron records a phase-specific timeout such as `setup timed out before runner start` or `stalled before first model call (last phase: context-engine)`. These watchdogs cover embedded providers and CLI-backed providers before their external CLI process is actually started, and are capped independently from long `timeoutSeconds` values so cold-start/auth/context failures surface quickly instead of waiting for the full job budget.

- If you use system cron or another external scheduler to run `openclaw agent`, wrap it with a hard-kill escalation even though the CLI handles `SIGTERM`/`SIGINT`. Gateway-backed runs ask the Gateway to abort accepted runs; local and embedded fallback runs receive the same abort signal. For GNU `timeout`, prefer `timeout -k 60 600 openclaw agent ...` over plain `timeout 600 ...`; the `-k` value is the supervisor backstop if the process cannot drain. For systemd units, keep the same shape by using a `SIGTERM` stop signal plus a grace window such as `TimeoutStopSec` before any final kill. If a retry reuses a `--run-id` while the original Gateway run is still active, the duplicate is reported as in-flight instead of starting a second run.

<a id="maintenance"></a>

- `--json` keeps stdout reserved for the JSON response. Gateway, plugin, and embedded-fallback diagnostics are routed to stderr so scripts can parse stdout directly.

- Embedded fallback JSON includes `meta.transport: "embedded"` and `meta.fallbackFrom: "gateway"` so scripts can distinguish fallback runs from Gateway runs.

- If the Gateway accepts an agent run but the CLI times out waiting for the final reply, embedded fallback uses a fresh explicit `gateway-fallback-*` session/run id and reports `meta.fallbackReason: "gateway_timeout"` plus the fallback session fields. This avoids racing the Gateway-owned transcript lock or silently replacing the original routed conversation session.

- For Gateway-backed runs, `SIGTERM` and `SIGINT` interrupt the waiting CLI request. If the Gateway has already accepted the run, the CLI also sends `chat.abort` for that accepted run id before exiting. Local `--local` runs and embedded fallback runs receive the same abort signal, but do not send `chat.abort`. If a duplicate `--run-id` reaches the Gateway while the original agent run is still active, the duplicate response reports `status: "in_flight"` and the non-JSON CLI prints a stderr diagnostic instead of an empty reply. For external cron/systemd wrappers, keep an outer hard-kill backstop such as `timeout -k 60 600 openclaw agent ...` so the supervisor can still reap the process if shutdown cannot drain.

- When this command triggers `models.json` regeneration, SecretRef-managed provider credentials are persisted as non-secret markers (for example env var names, `secretref-env:ENV_VAR_NAME`, or `secretref-managed`), not resolved secret plaintext.

- Marker writes are source-authoritative: OpenClaw persists markers from the active source config snapshot, not from resolved runtime secret values.