GitHub Issue Triage Agent with Codex
Triage is a short, well defined job repeated once per issue: label it, search for the duplicate, leave one comment, page Slack only for a genuine fire. Codex is fast on concrete work with clear edges, and the GitHub side of this is mostly precise API calls. For a team already on OpenAI tooling it is the least surprising harness to put on a public repo.
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.
Webhook retries fire twice when the first attempt times out
Openopened by Priya Raghunathan
On v4.2.1 self-hosted, a webhook whose first delivery hits the 10 second timeout is retried, but the receiver gets both the retry and the original once it finally completes. Steps: create a webhook pointing at an endpoint that sleeps 12 seconds, trigger it once, watch the receiver log two deliveries with the same event id about 40 seconds apart. Expected one delivery per event, or at least the same event id deduped. This started for us after upgrading from 4.1.6, where the same endpoint only ever received one call.
Set up in minutes
Using this template drops you into a guided setup. It asks exactly this, nothing else:
Connect GitHub
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.
Triage guide
Your label taxonomy and who owns what - the labels you actually use, what each one means, which team or person owns each area, and what counts as critical. The agent only ever labels and routes from this.
Maintainer notes
The repro details you always end up asking for, the parts of the codebase that break most often, and the tone you want used with contributors.
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
Seconds after the issue opens
The point of automated triage is that the reporter hears back while they still have the tab open. A quick harness makes that the normal case rather than the best case.
Exact with labels and links
Applying only labels that already exist and citing an issue by number are precision tasks. Careful API handling is where Codex is dependable.
Before you fork
What does this cost on a repo taking a lot of issues?
Each run covers one issue: a read, a search, a comment. Twenty new issues a day is twenty short runs a day, billed as your OpenAI token spend plus NoClick's hosted sandbox time, and it does not scale with how large the codebase is. Watch a busy public repo for the first day, mostly to check the comment tone rather than the bill.
Run it with a different agent
Put GitHub Issue Triage Agent to work on Codex
Free to start. Guided setup, a test run against staged conversations, and it's live.