Heartbeat

What This Skill Does

The Heartbeat skill is designed to bring architectural rigor to OpenClaw agent monitoring. Rather than relying on simple, noisy polling, this skill enables agents to implement adaptive cadence monitoring. It focuses on reliability by preventing runaway costs, mitigating alert fatigue, and ensuring that agent activities remain strictly aligned with user-defined active hours and risk tolerance levels. By utilizing a structured memory-based system stored in ~/heartbeat/, the skill ensures that configuration, historical snapshots, and tuning decisions are transparent and version-controlled.

Installation

To integrate the Heartbeat skill into your OpenClaw environment, execute the following command in your terminal: clawhub install openclaw/skills/skills/ivangdavila/heartbeat

Once installed, initiate the setup by running the setup.md script. This will capture critical environmental parameters such as your timezone, preferred active hours, and specific risk tolerance for automated actions.

Use Cases

1. Proactive System Health Checks: Configure an agent to monitor log files or resource usage, only triggering alerts if pre-defined thresholds are exceeded, thereby avoiding constant, empty status reports.
2. Adaptive Workflow Automation: Implement a hybrid heartbeat/cron strategy where routine reporting occurs on a fixed schedule (cron), while critical, variable-state task monitoring adapts based on real-time agent context (heartbeat).
3. Cost-Aware Monitoring: Guard expensive external API calls by implementing a two-stage heartbeat pattern, where a lightweight local pre-check determines if the more costly remote check is truly necessary.
4. Context-Sensitive Alerting: Ensure that alerts are suppressed during non-active hours by codifying your sleep schedule directly into the agent’s heartbeat memory, preventing midnight noise.

Example Prompts

1. "Analyze my current workflow and set up a heartbeat that checks for new email leads every 30 minutes, but only alerts me if the lead score is above 80."
2. "I'm getting too many notifications. Audit my current heartbeat configuration and implement a cooldown period for all non-critical alerts."
3. "Transition my server uptime monitoring from a fixed cron job to an adaptive heartbeat that triggers more frequently if latency spikes above 200ms."

Tips & Limitations

• Strict Output Contract: Always adhere to the HEARTBEAT_OK convention. If your monitor finds nothing, emit nothing. This is the single most important rule to prevent log bloat.
• Snapshotting: Always maintain the snapshots/ folder. If a new heartbeat configuration introduces unexpected behavior, reverting to the previous state is a simple file-swap operation.
• Escalation Rules: Never define an alert without an escalation path and a corresponding cooldown timer. Without these, you invite alert spam.
• Avoid Over-Automation: Use heartbeat for state-aware monitoring; do not attempt to replace system-level cron jobs for rigid, time-sensitive execution requirements. The heartbeat is for intelligence, not clock-watching.