How to Use OpenClaw Agent Send

openclaw agent is the CLI path for starting an agent turn without waiting for a chat message. It is useful for scripts, cron jobs, smoke tests, and controlled delivery into chat channels. The command sends a message through the Gateway by default, prints the reply, and can optionally deliver that reply to WhatsApp, Slack, Telegram, Discord, or another configured channel.

30-second answer

Run openclaw agent --message with the text you want the agent to handle. Add --agent to target a configured agent, --to to derive a session from a destination, or --session-id to continue an existing session. Add --deliver plus reply-channel and reply-to when the final response should be sent somewhere instead of only printed.

Why it exists

Not every agent run starts with a human typing. A daily report, a deploy check, a triage pass, or a data refresh may need to trigger OpenClaw from a shell script. openclaw agent gives that script a first-class entrance instead of pretending to be a chat adapter.

Session control

Session selection is where many operators get sloppy. --to derives a session key from a target, preserving isolation for groups and channels while direct chats collapse to main. --session-id is for continuing a known existing session. --agent targets a configured agent’s main session. Pick deliberately, because session memory, directives, and context follow that choice.

Delivery flow

Without --deliver, the command prints the answer. With --deliver, OpenClaw sends the reply to a channel. The docs show Slack delivery using --deliver, --reply-channel slack, and --reply-to. Use reply-account when an account id matters. For automation, I prefer an explicit destination over relying on ambient defaults.

Debugging and control

The flags include --thinking, --verbose, --timeout, --local, and --json. --local forces the embedded runtime instead of Gateway. If the Gateway is unreachable, the CLI can fall back locally. Use --json when another process needs structured metadata. Use verbose carefully; it is helpful for diagnosis but noisy in channels.

Playbook angle

The Playbook pattern is to make scripted agent runs explicit: one command, one owner, one target, one timeout, and one result format. That makes cron-driven agents much easier to audit when they wake up at odd hours.

Runbook checklist

Before you automate this, run one small acceptance test with harmless input. Confirm the tool is available to the right agent, the credential is loaded from config or environment, the output shape matches the workflow, and the failure message is understandable to a tired operator. If the feature touches money, public channels, logged-in browsers, host commands, or customer data, put a review step before the side effect. If it only reads data, still record the source and timestamp so future sessions do not treat stale context as fresh truth. Keep the first version narrow, then expand once the logs show the agent is choosing the right tool for the right reason. When the docs are incomplete, prefer a conservative sentence over a clever invented shortcut that future agents cannot reliably verify. Add one monitoring habit as well: after the first real run, check the transcript or logs for missing prerequisites, broad prompts, stale assumptions, and accidental side effects. Tighten the instruction while the failure is fresh. The best OpenClaw workflows improve in small, documented passes instead of one giant rewrite after something breaks in public. For SEO pages, that same discipline matters: do not promise hidden capabilities, paid-provider limits, or setup shortcuts unless the current docs say so. Trust compounds when the guide is accurate even in the boring operational edge cases that matter during real maintenance windows.

Operator note

How to Use OpenClaw Agent Send works best when it is written into a small runbook instead of treated as a magic switch. Record who owns the workflow, which config keys are allowed, which credentials are required, what the agent may do without approval, and what counts as a failure. OpenClaw gives agents broad tools, but the reliable version is boring: one source of truth, one verification step, and one rollback path when a provider or channel behaves differently than expected.