What is the OpenClaw Heartbeat scheduler?

The OpenClaw Heartbeat scheduler is a built-in system that wakes your AI agent at configurable intervals for proactive, unprompted execution. Instead of waiting for user messages, the agent periodically activates to check for new messages, run scheduled tasks, monitor systems, and perform maintenance. It is configured via a plain markdown file called HEARTBEAT.md in your workspace directory.

How do I configure HEARTBEAT.md in OpenClaw?

Create a file called HEARTBEAT.md in your OpenClaw workspace directory. Inside, define the wake interval (e.g., every 5 minutes, every hour), specify which tasks the agent should perform on each wake cycle, and list any conditions or priorities. The file uses plain markdown syntax — no database or special tooling required. OpenClaw reads this file automatically.

Why is the Heartbeat important for persistent AI agents?

The Heartbeat scheduler solves the persistence problem that killed earlier AI agents like AutoGPT. Without a heartbeat mechanism, AI agents only respond when prompted — they cannot proactively check inboxes, monitor systems, or run scheduled workflows. OpenClaw's Heartbeat turns the agent from a reactive chatbot into a persistent, always-on assistant that works in the background even when you are not actively using it.

What tasks can the OpenClaw Heartbeat trigger?

The Heartbeat can trigger virtually any task your agent is configured to perform: checking for new messages across communication channels (email, WhatsApp, Slack), running scheduled reports, monitoring system health, processing queued work items, syncing data between applications, sending follow-up messages, and performing routine maintenance. You define the exact tasks in your HEARTBEAT.md file.

How often should I set the OpenClaw Heartbeat interval?

The optimal interval depends on your use case. For message monitoring (email, chat), 1-5 minutes is common. For system monitoring and reporting, 15-60 minutes works well. For daily maintenance tasks, a 24-hour interval is appropriate. Start with a longer interval and reduce it as needed. Shorter intervals consume more resources but provide faster response times.

Does the Heartbeat run when my computer is off?

The Heartbeat scheduler requires the OpenClaw application to be running. If you shut down your computer or close OpenClaw, the heartbeat pauses and resumes when the application restarts. For truly 24/7 operation, many users run OpenClaw on a dedicated server, a Raspberry Pi, or a cloud VM using Docker. See the OpenClaw Docker setup guide for always-on deployment options.

Is HEARTBEAT.md the same as a cron job?

HEARTBEAT.md is similar to a cron job in that it triggers periodic execution, but it is far more powerful. Unlike cron, the Heartbeat wakes an intelligent AI agent that can reason about context, make decisions, and adapt its actions based on what it finds. A cron job runs a fixed script; a Heartbeat cycle wakes an agent that can think, prioritize, and act autonomously across multiple applications.

Can the Heartbeat be paused or conditionally disabled?

Yes. You can include quiet hours in your HEARTBEAT.md to pause execution during off-hours, instruct the agent to skip certain tasks on weekends, or set priority conditions that gate whether tasks run. The agent reads these conditions on every cycle and adjusts its behavior accordingly. This prevents unnecessary API usage and respects business-hours boundaries without requiring code changes.

How does HEARTBEAT.md interact with MEMORY.md and SOUL.md?

HEARTBEAT.md defines when the agent acts. SOUL.md defines how it behaves while acting. MEMORY.md stores what the agent learned from previous cycles. Together, these three files create a fully autonomous loop: the Heartbeat wakes the agent, the Soul guides its decisions, and Memory preserves context between cycles so the agent does not repeat work or lose track of ongoing tasks.

OpenClaw Heartbeat Scheduler: Always-On AI Agent

Written by Ty-Shane Howell · Last Updated: March 2026

TL;DR

HEARTBEAT.md defines periodic tasks your OpenClaw agent runs on a schedule — from morning briefings to system health checks to end-of-day summaries. It turns your agent from a reactive assistant into a proactive one that monitors, reports, and acts without being asked. Think of it as your agent's cron job system.

The OpenClaw Heartbeat scheduler wakes your AI agent at configurable intervals for proactive, unprompted execution. Configured via a plain markdown file called HEARTBEAT.md in your workspace directory, it transforms OpenClaw from a reactive chatbot into a persistent, always-on agent that checks messages, runs scheduled tasks, monitors systems, and performs maintenance — even when you are not actively using it. This is the key innovation that solved the "persistence problem" that killed predecessor AI agents.

The Heartbeat is what makes OpenClaw truly autonomous. Combined with SOUL.md, MEMORY.md, and AGENTS.md, it creates an AI agent that works for you around the clock.

Trusted by 194+ business owners who configured OpenClaw in our live workshop — endorsed by Josh Nelson's 7 Figure Agency community (20,000+ agency owners)

What Is the Heartbeat Scheduler?

The Heartbeat is OpenClaw's built-in scheduling system that gives your AI agent a pulse — regular wake cycles where it proactively checks for work and takes action.

Configurable Wake Intervals

Set your agent to wake every minute, every hour, or on any custom schedule. Fine-tune timing per task for optimal resource usage.

Proactive Execution

Your agent does not wait for prompts. It autonomously checks inboxes, processes queues, runs reports, and handles tasks on its own schedule.

Plain Markdown Config

All configuration lives in a single HEARTBEAT.md file. No database, no YAML, no JSON schemas. Just readable markdown in your workspace.

Why the Heartbeat Changes Everything

Before OpenClaw, AI agents like AutoGPT suffered from the "persistence problem" — they could only respond when a user sent a message. Once the session ended, the agent went silent. OpenClaw solved this with the Heartbeat scheduler combined with messaging platform accessibility (WhatsApp, Telegram, email). The agent stays alive, keeps working, and reaches you wherever you are. This is why OpenClaw is the first AI agent to achieve true persistence and accessibility, and a major reason it has earned over 200,000 GitHub stars. OpenClaw is free, MIT-licensed, and created by Peter Steinberger.

How the Heartbeat Works

The Heartbeat follows a simple cycle: sleep, wake, execute tasks, sleep again. Here is how each cycle unfolds.

Step 1: Timer Triggers Wake

Automatic

The configured interval elapses (e.g., 5 minutes since last cycle). OpenClaw activates the agent for a new heartbeat cycle.

Step 2: Agent Reads HEARTBEAT.md

Instant

The agent loads your HEARTBEAT.md file to understand what tasks to perform, in what order, and under what conditions.

Step 3: Agent Scans for Work

Varies

Based on the instructions, the agent checks inboxes, monitors systems, reviews queues, and evaluates whether any action is needed.

Step 4: Agent Executes Tasks

Varies

If work is found, the agent takes action: drafts replies, sends alerts, updates records, generates reports, or performs any configured task.

Step 5: Agent Sleeps Until Next Cycle

Automatic

Once all tasks are complete (or if nothing needed attention), the agent goes back to sleep until the next heartbeat interval.

Copy-Paste HEARTBEAT.md Templates

These ready-to-use templates cover the three most common Heartbeat use cases. Copy, customize the placeholders, and drop into your workspace directory.

Template 1: Morning Briefing

Check email, weather, calendar, and news — generate a daily summary by 8 AM.

HEARTBEAT.md — Morning Briefingmarkdown

# Heartbeat Configuration — Morning Briefing Agent

## Schedule
- Wake every **5 minutes** from 7:30 AM to 9:00 AM
- Outside of briefing window: wake every **60 minutes** for urgent messages only

## Morning Briefing Tasks (Run at 7:50 AM daily)

### 1. Check Email
- Scan inbox for unread messages since last briefing
- Flag urgent items (keywords: urgent, ASAP, deadline, client)
- Draft replies for time-sensitive messages

### 2. Check Weather
- Fetch today's weather forecast for the user's location
- Note any travel-disrupting conditions

### 3. Check Calendar
- Review today's appointments and meetings
- Confirm all meeting links are present
- Flag back-to-back conflicts

### 4. Check News
- Scan configured news sources for industry headlines
- Filter by relevance to user's business topics

### 5. Generate Summary
- Compile all findings into a concise morning briefing
- Send summary to user via WhatsApp by 8:00 AM
- Format: bullet points, priority order, action items first

## Quiet Hours
- Do not run briefing tasks between 9 PM and 7:30 AM
- Urgent message check only during quiet hours

Template 2: Business Monitor

Check CRM pipeline, support tickets, and sales metrics — send a daily report to your team.

HEARTBEAT.md — Business Monitormarkdown

# Heartbeat Configuration — Business Monitor Agent

## Schedule
- Wake every **15 minutes** during business hours (8 AM - 7 PM)
- Wake every **60 minutes** outside business hours

## Priority 1: CRM Pipeline (Every Cycle)
- Check for new leads in GoHighLevel/HubSpot
- Review deals that have been inactive for more than 48 hours
- Flag any hot leads that need immediate follow-up
- Update pipeline stage based on latest activity

## Priority 2: Support Tickets (Every 2nd Cycle)
- Check for new support tickets or unresolved issues
- Escalate tickets open longer than 4 hours without response
- Summarize ticket volume and category trends

## Priority 3: Sales Metrics (Every 3rd Cycle)
- Pull daily revenue figures from payment processor
- Compare against daily target
- Flag if revenue is more than 20% below target

## Daily Report (Sent at 6:00 PM)
- Compile full day summary: leads, pipeline, revenue, tickets
- Compare to same day last week
- Send report to owner via WhatsApp and email
- Save report to ./reports/YYYY-MM-DD.md

## Error Handling
- If CRM API fails, retry after 5 minutes
- If 3 consecutive failures, alert owner immediately
- Log all failures to ./logs/heartbeat-errors.md

Template 3: Developer Watch

Check CI/CD status, error logs, and PR reviews — notify the team of failures immediately.

HEARTBEAT.md — Developer Watchmarkdown

# Heartbeat Configuration — Developer Watch Agent

## Schedule
- Wake every **5 minutes** during active development (9 AM - 6 PM)
- Wake every **30 minutes** off-hours

## Priority 1: CI/CD Status (Every Cycle)
- Check GitHub Actions / CircleCI for running pipelines
- Alert immediately if any build fails on main or staging branch
- Notify assigned developer via Slack DM on failure

## Priority 2: Error Log Monitoring (Every Cycle)
- Scan application error logs for new critical errors
- Group errors by type and frequency
- Alert if error rate exceeds threshold (>10 errors/minute)

## Priority 3: PR Reviews (Every 2nd Cycle)
- Check for pull requests awaiting review for more than 2 hours
- Identify reviewers who have been assigned
- Send Slack reminder to reviewers if PR is blocked

## Priority 4: Deployment Status (Every 3rd Cycle)
- Verify latest deployment is healthy
- Check uptime monitoring for anomalies
- Confirm all health check endpoints return 200

## Daily Dev Summary (Sent at 5:45 PM)
- Builds: passed / failed count
- PRs opened, merged, and awaiting review
- Error rate trends
- Send to #dev-updates Slack channel

## Quiet Hours
- Off-hours: only alert on production outages (P0 only)

Use Cases for Proactive Execution

The Heartbeat enables your agent to handle these tasks autonomously, without waiting for a prompt.

Message Monitoring

Check email, WhatsApp, Slack, and other channels for new messages. Draft replies, flag urgent items, and route conversations to the right people.

Recommended: Every 1-5 minutes

System Monitoring

Watch server health, uptime status, error logs, and performance metrics. Alert you immediately when something requires attention.

Recommended: Every 5-15 minutes

Data Synchronization

Keep your CRM, spreadsheets, project management tools, and databases in sync. Reconcile records and resolve conflicts automatically.

Recommended: Every 15-60 minutes

Scheduled Reports

Generate daily sales summaries, weekly project updates, or monthly analytics reports. Deliver them to the right channels at the right time.

Recommended: Every 1-24 hours

Follow-Up Automation

Check if leads need follow-ups, send reminders for overdue tasks, nudge stalled conversations, and keep pipelines moving.

Recommended: Every 30-60 minutes

Security & Compliance

Scan for unauthorized access attempts, verify backups completed, check certificate expiration dates, and audit permission changes.

Recommended: Every 1-6 hours

Heartbeat Best Practices

Follow these guidelines to get the most from your agent's always-on capabilities.

Start with longer intervals

Begin with 30-60 minute intervals and shorten them as you understand your agent's workload. This prevents unnecessary resource consumption while you tune the system.

Prioritize tasks within each cycle

List tasks in order of importance in your HEARTBEAT.md. If a wake cycle runs long, the agent handles critical items first and defers lower-priority work to the next cycle.

Use conditional logic

Instruct the agent to skip certain checks based on time of day or day of week. For example, skip Slack monitoring outside business hours or increase email check frequency during product launches.

Monitor resource usage

Each heartbeat cycle uses API calls and compute time. Track your usage and adjust intervals to balance responsiveness with cost efficiency.

Combine with SOUL.md and MEMORY.md

The Heartbeat works best when paired with a well-defined SOUL.md (personality and rules) and MEMORY.md (persistent context). Together, these three files create a fully autonomous agent.

Log heartbeat activity

Include a logging directive in your HEARTBEAT.md so the agent records what it did on each wake cycle. This makes debugging easier and helps you optimize the schedule over time.

The Heartbeat in Context

HEARTBEAT.md works alongside OpenClaw's other configuration files to create a fully autonomous agent.

SOUL.md

Defines who the agent is

Personality, rules, boundaries, and decision-making guidelines. The Heartbeat tells the agent when to act; SOUL.md tells it how to act.

Read the guide

MEMORY.md

Stores what the agent knows

Persistent context, conversation history, learned preferences, and accumulated knowledge. The Heartbeat can read and update MEMORY.md on each cycle.

Read the guide

AGENTS.md

Coordinates multiple agents

Multi-agent orchestration and delegation rules. The Heartbeat can wake specific sub-agents for specialized tasks within a cycle.

Read the guide

HEARTBEAT.md

Controls when the agent acts

The scheduling backbone that transforms a passive chatbot into a persistent, proactive assistant. Without it, the agent only responds when prompted.

Read the guide

OpenClaw Heartbeat FAQ

Configure Your Always-On Agent This Week

The 20-minute workshop includes done-for-you HEARTBEAT.md templates, step-by-step configuration, and best practices for persistent AI automation. One payment, no subscriptions, 30-day guarantee.

Save 10+ hours/week Cut AI costs by 97% Deploy in under 20 min

Get the Automation Playbook (Free)

One deploy-ready automation every week. Same strategies our clients pay thousands for. 400+ business owners already inside.

Need it done for you?

Book a Free Strategy Call See what we've built for real businesses →

Related OpenClaw Configuration Guides

OpenClaw .md Files Hub

Overview of all OpenClaw configuration files and how they work together.

MEMORY.md Guide

Configure persistent memory so your agent retains context across Heartbeat cycles.

BOOTSTRAP.md Guide

Set up one-time initialization tasks that run before your agent's first Heartbeat cycle.

AGENTS.md Guide

Orchestrate multiple agents, each with their own Heartbeat schedules and task queues.