Production Errors Straight to Your Coding Agent: Workers Issues Enters Open Beta

Hi everyone, it's Shii-chan! This one should make debugging production a lot easier.

Cloudflare Blog blog.cloudflare.com

What was announced?

The Cloudflare Blog introduced Issues, built-in error monitoring for Workers, in open beta. It detects production failures, groups them, and sends diagnostic context to a coding agent so it can investigate and open a pull request.

Why it matters

Detecting errors, collecting logs, and finding causes takes effort. Issues captures errors at the Workers runtime level, so no extra instrumentation is needed.

What changes

Related exceptions, 5xx responses, and error logs are grouped into a single issue that can go straight to an agent. Cloudflare's Workflows team found two production bugs within a day using Issues on their own systems.

Dive Deep

What it detects:

• Uncaught exceptions and failed invocations
• HTTP 5xx responses
• console.log / console.error output and stack traces
• Runaway alarms and excessive logging in loops

What the agent receives: the failure summary, the exception or error, a source-mapped stack trace, leading and trailing logs and traces, and the Worker version. You can attach identifiers such as user, account, and session IDs through the OpenTelemetry API without extra packages.

Issues can be routed to Claude Code, Cursor, Devin, generic webhooks, or incident management and chat platforms, based on thresholds you configure. Setup is one line:

observability.issues.enabled: true

The new cf CLI can also inspect issues and create automations. Pricing was not specified in the announcement.

Wrap-up

• Built-in Workers error monitoring is in open beta
• Exceptions, 5xx, and logs are grouped, with source-mapped stack traces and more sent to agents
• One-line setup, plus cf CLI support

This is for anyone running Workers in production who wants to hand incident investigation to an agent!