Incident Scribe Agent with Codex
A scribe's value decays by the minute, so the run has to finish while people are still asking what is happening. Codex is quick on tightly specified output, and this output is about as specified as it gets: five labelled sections, one or two sentences each, no threads and no code blocks. The PagerDuty side is a read of the incident, its alerts and past incidents, then one note back.
Watch it work before it's live
Run a staged conversation — no account needed. The agent handles it for real while a simulated world answers its tool calls; nothing touches real accounts, and nothing is actually sent.
P1: Checkout API error rate 11% for 6 minutes
TriggeredMonitor checkout-api 5xx ratio triggered at 14:02 UTC against a 2% threshold, currently 11.4%, with 340 of the last 2,980 requests returning 502 from the payments upstream. Two alerts merged into this incident, both from eu-west-1. Escalation policy Payments Primary, nobody has acknowledged yet.
Set up in minutes
Using this template drops you into a guided setup. It asks exactly this, nothing else:
Connect PagerDuty
One sign-in. The agent acts through your account, scoped to what this template uses.
Connect Slack
One sign-in. The agent acts through your account, scoped to what this template uses.
Runbook notes
What your team needs said about each service - what it does, who feels it when it breaks, and any standing context you want repeated in an update.
Comms style
How your incident updates should read and how often - the tone, who is reading, and the cadence that sets the next update time, for example every 30 minutes while a P1 is open.
Runs on Codex
Preselected for this page — connect your Codex account during setup, or switch to NoClick's built-in models with one click.
Watch it handle a test run
A staged conversation against a simulated world — then it’s live.
Why Codex for this agent
Fast enough to matter
The first update lands while the responder is still reading the alert, which is the only window in which a scribe changes anything.
A fixed output shape
The five sections never vary, and holding a rigid format is an easier job than holding an opinion.
Before you fork
What happens with a flapping alert that keeps reopening?
You get one run per incident event you forward, so a service that flaps produces repeated updates in the channel. Filter that at PagerDuty by pointing the trigger at the services that need public communication rather than every service. The staged flapping alert exists so you can see what that noise reads like first.
Run it with a different agent
Put Incident Scribe Agent to work on Codex
Free to start. Guided setup, a test run against staged conversations, and it's live.