Skip to main content
Version: v0.2 (Next)

Automations & OpenClaw Heartbeat

ActonOS features a robust automation framework combining Standard Cron/Interval Triggers with an Autonomous Heartbeat Daemon built in strict alignment with the OpenClaw Heartbeat Contract.

1. OpenClaw-Aligned Heartbeat Contract​

The ActonOS Heartbeat daemon runs periodic system checkups to maintain proactive agent vigilance without burning unnecessary token budgets:

Idle Heartbeat Token Optimization​

  • When there are no active backlog tasks and no actionable directives in HEARTBEAT.md, the heartbeat does not invoke the LLM at all.
  • It returns an immediate internal HEARTBEAT_OK acknowledgement, ensuring zero token waste during quiescent periods.

Headless Autonomous Execution Mode​

  • When an actionable directive is present, the routine cycle injects an explicit rule block informing the model that execution is unattended and no human is present to greet.
  • A built-in looksLikeIdleChatter safety filter automatically discards accidental conversational greetings (e.g., "Hello, how can I help you today?") and only forwards genuine alerts.

Active Hours & Trigger Cooldowns​

  • Active Hours Window: Configure daily operating hours (e.g., 08:00 to 22:00 in timezone Asia/Ho_Chi_Minh) so non-critical background routines sleep overnight.
  • Trigger Cooldown (15s): Coalesces rapid successive triggers (e.g., multiple file changes or task updates) into a single batch cycle.
  • Manual Pulse: Operators can trigger an immediate checkup anytime via the Pulse Now button.

2. Setting Up Automated Routines​

  1. Navigate to Automations in the left sidebar.
  2. Click + New Automation.
  3. Choose the Trigger Type:
    • Cron Expression: Standard 5-field cron (e.g., 0 9 * * 1-5 for 9:00 AM every weekday).
    • Interval: Run every $N$ minutes or hours.
    • Event Trigger: Triggered when specific bus events occur (e.g., file.created, channel.message).
    • Webhook Trigger: Expose an authenticated HTTP webhook endpoint.
  4. Define the Action:
    • Assign an agent and provide the standing instruction (e.g., Check open GitHub pull requests, summarize reviews, and post to Slack).
    • Select the target output channel for completed reports.
  5. Click Save Automation.