openclaw agent runs a single agent turn from the command line without an inbound chat message. Use it for scripted workflows, testing, and programmatic delivery. Full flag and behavior reference: Agent CLI reference.

Quick start

Sends the message through the Gateway and prints the reply.

Reads a valid UTF-8 file as the agent message body.

# Target a phone number (derives session key)
openclaw agent --to +15555550123 --message "Status update"

# Reuse an existing session
openclaw agent --session-id abc123 --message "Continue the task"

# Target an exact session key
openclaw agent --session-key agent:ops:incident-42 --message "Summarize status"
```

# Deliver to Slack
openclaw agent --agent ops --message "Generate report" \
  --deliver --reply-channel slack --reply-to "#reports"
```

Flags

Behavior

• By default, the CLI goes through the Gateway. Add --local to force the embedded runtime on the current machine.
• Pass exactly one of --message or --message-file. File messages preserve multiline content after removing an optional UTF-8 BOM. Files larger than 4 MiB are rejected before dispatch.
• If the Gateway request fails, the CLI falls back to the local embedded run; a Gateway timeout falls back with a fresh session instead of racing the original transcript.
• Session selection: --to derives the session key (group/channel targets preserve isolation; direct chats collapse to main). With --agent, --channel, and --to together, routing follows the channel's canonical recipient and session.dmScope. Stable outbound-only identities use a provider-owned session isolated from the agent's main session.
• --session-key selects an explicit key. Agent-prefixed keys must use agent:<agent-id>:<session-key>, and --agent must match that agent id when both are supplied. Bare non-sentinel keys are scoped to --agent when supplied; for example, --agent ops --session-key incident-42 routes to agent:ops:incident-42. Without --agent, bare non-sentinel keys are scoped to the configured default agent. Literal global and unknown remain unscoped only when no --agent is supplied; the embedded fallback path resolves those sentinel sessions to the configured default agent.
• --reply-channel and --reply-account affect delivery only.
• Thinking and verbose flags persist into the session store.
• Output: plain text by default, or --json for structured payload + metadata.
• With --json --deliver, the JSON includes delivery status for sent, suppressed, partial, and failed sends. See JSON delivery status.

Examples

# Simple turn with JSON output
openclaw agent --to +15555550123 --message "Trace logs" --verbose on --json

# Turn with a model override
openclaw agent --agent ops --model openai/gpt-5.4 --message "Summarize logs"

# Turn with thinking level
openclaw agent --session-id 1234 --message "Summarize inbox" --thinking medium

# Multiline prompt from a file
openclaw agent --agent ops --message-file ./task.md

# Exact session key
openclaw agent --session-key agent:ops:incident-42 --message "Summarize status"

# Legacy key scoped to an agent
openclaw agent --agent ops --session-key incident-42 --message "Summarize status"

# Deliver to a different channel than the session
openclaw agent --agent ops --message "Alert" --deliver --reply-channel telegram --reply-to "@admin"

Related

Source: docs/tools/agent-send.md