ApprovalGate

approvalgate

Approval layer

A webhook almost refunded $4,200 to the wrong customer

Put a human in front of risky webhooks.

A Stripe webhook fired, my automation ran, and I caught it 2 seconds before it went through. Pure luck.ApprovalGate pauses risky requests, shows you the payload, and waits for your OK.

Start FreeSee how it works

No SDK. Swap your webhook URL. Define rules like amount > 500. Done.

approvalgate.devgate.run

incoming webhook

{
  "event": "charge.refunded",
  "amount": 4200,
  "customer": "wrong_acme"
}
rule matched: amount > 100awaiting approval

slack notification

[approve] [reject]
refund $4,200 to wrong_acme?

The problem

Webhooks are fire-and-forget

The duplicate refund

Retries fired twice

Money went out. Now you're unwinding it.

The partner sync

Bad payload touched prod

Thousands of records updated.

The delete endpoint

Archive vs delete

One verb, very different outcome.

The AI agent

Wrote to production

LLMs hallucinate. Your data shouldn't.

Automation is great until it isn't.
The risky 5% deserves a human checkpoint.

How it works

5 minutes to set up. No code changes.

It sits between your trigger and your action.

1
Point your webhook source at ApprovalGatehttps://approvalgate.dev/g/xK9mP2
2
Define your rulesif amount > 500 or action = "delete" → require approval
3
Get notified in Slack with full context

See the payload. Approve or reject in one click.

Approved → forwards to your real endpoint. Rejected → dropped.

Use cases

For the 5% of webhooks that can hurt you

Refunds, chargebacks, payouts

Stop money from leaving on a bad payload.

if amount > 100

Partner syncs and imports

One bad payload can touch thousands of rows.

if records > 500

AI agents writing to prod

Let agents draft. You approve what ships.

if action = "write"

Mass emails and notifications

Wrong segment = instant trust damage.

if audience > 1000

Being honest

What ApprovalGate is not

Not a filter

Filters auto-approve or reject. This pauses and asks a human.

Not a queue

It doesn't fix retries or ordering. It's about judgment.

Not for latency-sensitive flows

If it must be instant, don't gate it. Gate the risky 5%.

Pricing

Free to start. Upgrade when the risk is real.

Free

$0

  • 1 gate
  • 20 requests/month
  • Slack notifications
  • 24h timeout

For testing and side projects

Popular

Pro

$9/mo

  • 10 gates
  • 1,000 requests/month
  • 5 rules per gate
  • Priority support

For revenue workflows

Team

$29/mo

  • Unlimited gates
  • 10,000 requests/month
  • Unlimited rules
  • Team members Soon

For teams on production pipelines

Had your own "almost" moment?

Add a checkpoint before it hits prod.

5 minutes to set up. No SDK. No code changes.

Start Free