Give your OpenClaw agent a phone call.
AgentPing is a phone-call alerting service for AI agents. It calls you when your agent is done, stuck, fails, or needs approval — so you don't miss critical moments buried in Slack, Telegram, or email.
Install the AgentPing skill and your OpenClaw agent can call your phone when it needs approval, hits a failure, or reaches a deadline.
Free plan: 10 alerts/month, 5 call minutes, 1 verified phone.
Setup
One command. One config block.
Install from ClawHub, add your API key, and your OpenClaw agent can start calling you immediately.
01
Create account & get API key
Sign up at agentping.me, verify your phone, and generate an API key from the dashboard.
02
Install the skill
Run the install command and add your key to the OpenClaw config.
03
Your agent can now call you
Ask your agent to call you, or let it decide based on urgency. It handles the rest.
Need more detail? See the full OpenClaw setup guide
$ clawhub install agentping-phone-call-alerts{
"skills": {
"entries": {
"agentping": {
"enabled": true,
"apiKey": "ap_sk_your_key_here"
}
}
}
}That's it. Your OpenClaw agent now has the agentping_alert tool available.
What it does
Your agent decides when to call. You decide what to do.
Approvals
Before irreversible actions
Your agent needs to deploy, delete, or spend money. It calls you first and waits for your explicit go-ahead.
Example alert
“Deploy approval needed — v2.4.1 ready for production, 3 migrations pending.”
Failures
When something breaks
A pipeline crashes, a build fails, or an API starts returning errors. You get a call with the details, not a silent log entry.
Example alert
“Ingest pipeline failed — crashed after processing 12k of 50k records.”
Deadlines
When timing matters
Your agent knows a window is closing. It calls to make sure you act before the moment passes.
Example alert
“Leave for the airport now — traffic just got worse, 10 minutes left.”
Features
Built for agents that run when you are not watching.
Smart delays
Approval alerts wait 5 minutes by default — giving you time to see it in chat first. Failures call immediately. All configurable.
Severity levels
Normal alerts respect your quiet hours. Critical alerts bypass them. Your agent picks the right level based on urgency.
Keypad acknowledgement
Press 0 to acknowledge, 1 to snooze 5 minutes, or enter 2-120 followed by # for a custom snooze. Your agent knows the outcome.
Retries until resolved
If you don't pick up, AgentPing calls again. Free plan retries once, Starter retries 5 times, Pro retries 10 times.
Metadata tracking
Attach task IDs, URLs, or run IDs to alerts. Check status programmatically. Build acknowledgement into your workflow.
Multi-phone routing
Verify multiple phones and target specific numbers per alert. Route work alerts to your work phone, personal to personal.
How your agent uses it
The best pattern: schedule first, escalate second.
Your agent records a checkpoint, gives you time to reply in chat, and only calls if you haven't responded by the deadline.
Step 1 — Agent finds a problem
"The deploy pipeline failed on the staging environment. I'll notify you and wait for instructions."
Step 2 — Agent sends alert with delay
agentping_alert(title="Deploy failed on staging", severity="normal", alert_type="task_failure", delay_seconds=120)
Step 3 — You reply in chat within 2 min
"Show me the error log" → agent acknowledges the alert, no call needed.
Or — 2 min pass with no reply
AgentPing calls your phone with the failure details. Press 0 to acknowledge.
Pricing
Free to start. Scale when you need to.
Free
$0
10 alerts/mo, 5 call min
Starter
$9/mo
300 alerts/mo, 120 call min
Pro
$29/mo
2,000 alerts/mo, 500 call min
All plans include voice calls, keypad acknowledgement, quiet hours, and API access. Full comparison
Get started
Your OpenClaw agent is one install away from calling you.
Create a free account, install the skill, and test the full call flow. Takes about two minutes.