Automation & Cron

TL;DR: Sam has one cron job (broken) and a heartbeat system driven by a markdown file. There are no webhooks, no event-driven automation, and no Gmail PubSub — despite having all the credentials to set them up.

The Two Automation Systems

OpenClaw provides two distinct automation mechanisms, and Sam uses both in an overlapping, slightly conflicting way:

System	Purpose	Config Method	Current State
Heartbeat	Periodic check-ins from the main session	HEARTBEAT.md file instructions	🟡 Working but over-scheduled
Cron	Isolated, scheduled agent runs	cron/jobs.json	🔴 1 job, breaking (timeout)

The Conflict

The cron job (funnel-github-watch) runs every 30 minutes to check GitHub. The heartbeat (3×/day at 10am, 2pm, 6pm PT) also checks GitHub as part of its window tasks. Result: GitHub is being checked too often, sometimes by two systems simultaneously, with John having told Sam five times to stop the constant reporting.

---

Heartbeat

Configuration

The heartbeat is not configured in openclaw.json — it's entirely driven by HEARTBEAT.md in the workspace, which Sam reads at the start of each heartbeat poll.

The heartbeat prompt that triggers Sam is:

"Read HEARTBEAT.md if it exists (workspace context). Follow it strictly. Do not infer or repeat old tasks from prior chats. If nothing needs attention, reply HEARTBEAT_OK."

Schedule

Sam has tightened the heartbeat schedule over time. The current HEARTBEAT.md specifies:

Only act at these 3 windows (PST):

• 10:00 AM (5:00 PM UTC)
• 2:00 PM (9:00 PM UTC)
• 6:00 PM (1:00 AM UTC)

If not within 30 minutes of those windows: reply HEARTBEAT_OK and do nothing.

This was added after John complained 5+ times about constant commit reports.

Tasks at Each Window

At each of the 3 daily windows, Sam performs:

1. Check sam-chat (C0AJVMAHM50) — read last 15 messages, respond if needed
2. Check funnel-squad (C0AJ67B5VP1) — read last 10 messages, respond if needed
3. Funnel & test tracking — post to sam-chat with:
   • Overall funnel CVR (landing → purchase), target 1.5%
   • Order Bump: AOV, purchases, upsell rate vs baseline
   • SMS Disclaimer: lead view → lead CVR, baseline 74.7%
   • SEMA route tracking (reverted to SEMA 3/17)
   • Mounjaro routing tracking (new 3/17)
   • Express Pay tracking (new 3/17)
   • DM John only if >5% shift
4. GitHub funnel monitor — check funnel, funnel-cms, landing-pages, checkout repos silently; log to memory/funnel-changes.md; alert John ONLY if change correlates with >5% CVR shift

Heartbeat History Note

Before the current tight schedule, Sam was:

• Reporting GitHub commits every single heartbeat (every ~30 min)
• Posting to sam-chat continuously
• Checking PostHog data outside scheduled windows

This led to multiple complaints from John. The HEARTBEAT.md was progressively tightened with explicit STOP commands.

---

Cron Job

funnel-github-watch

{
  "id": "22b4d31f-04ec-475e-8614-6acc60c85994",
  "name": "funnel-github-watch",
  "description": "Check FuturHealth GitHub repos for new code pushes every 30 min. Alert John on changes.",
  "enabled": true,
  "schedule": {
    "kind": "every",
    "everyMs": 1800000
  },
  "sessionTarget": "isolated",
  "wakeMode": "now",
  "payload": {
    "kind": "agentTurn",
    "timeoutSeconds": 90,
    "lightContext": true
  },
  "delivery": {
    "mode": "announce",
    "channel": "slack",
    "to": "U06BQPTAE1H"
  }
}

Current state:

"state": {
  "lastRunStatus": "error",
  "lastStatus": "error",
  "lastDurationMs": 90008,
  "consecutiveErrors": 1,
  "lastError": "cron: job execution timed out"
}

Problems:

1. Timing out at exactly 90,008ms — the 90-second timeout is being hit consistently
2. Running every 30 minutes — despite HEARTBEAT.md saying to only check GitHub 3×/day
3. Conflicting with heartbeat — both systems attempt GitHub monitoring
4. Delivering to John directly — but heartbeat also posts to sam-chat. Two systems, overlapping outputs.

Why it's timing out: The cron job tries to authenticate with GitHub API, check 4 repos, compare SHAs, log to funnel-changes.md, and potentially send a Slack DM — all in 90 seconds. With any network latency or GitHub rate limiting, this fails.

Fix: Either disable this cron job (rely on heartbeat's GitHub check instead) or extend the timeout and fix the GitHub check script.

---

Delivery Queue

.openclaw/delivery-queue/
└── failed/    # Empty — no failed messages

No failed deliveries in queue. Messages from cron jobs and heartbeats have been delivered successfully to Slack when they do complete.

---

What's Not Configured (But Should Be)

Gmail PubSub

What it is: OpenClaw supports receiving Gmail push notifications via Google Cloud PubSub. Sam already has Google OAuth tokens for john@fh.co and johnlevan1@gmail.com.

What it would enable: Instead of polling Gmail every 30 minutes, Gmail pushes notifications to Sam instantly when important emails arrive.

Config needed:

"hooks": {
  "enabled": true,
  "mappings": [
    { "match": { "path": "gmail" }, "action": "agent", "agentId": "main" }
  ]
}

Webhooks / Hooks

What they are: HTTP endpoints on the gateway that external services can POST to.

What they would enable:

• GitHub webhooks → Sam gets notified of pushes instantly (no need to poll every 30 min)
• Zapier/Make integrations
• Custom event triggers from FuturHealth's own systems

Polls

What they are: OpenClaw's poll automation type — check an endpoint on a schedule and trigger only when data changes.

What they would enable: More efficient data checks without running a full agent turn for each poll.

---

Summary: Automation Problems

Problem	Impact	Fix
Cron job timing out every 30 min	Wasted compute + incomplete GitHub monitoring	Disable cron or extend timeout + fix script
Heartbeat + Cron checking GitHub simultaneously	Duplicate work, duplicate alerts	Consolidate into one system
No webhooks configured	Sam polls rather than reacts	Add GitHub + Gmail webhooks
No heartbeat config in openclaw.json	Schedule not enforced by runtime	Add agents.defaults.heartbeat.every
No Gmail PubSub	Email scanning is 30-min delayed	Set up Gmail PubSub hook

---

See also: Configuration | Known Issues | GitHub Monitoring Workflow