Intel / blog

Detect and send production issues straight to your agent

Source: blog.cloudflare.com
Date: Published Sep 30, 2026

Context

Cloudflare's Issues feature turns repeated production failures into a structured assignment for a coding . For applications running on Workers, it groups exceptions, server-error responses and error logs, then attaches diagnostic such as stack traces, surrounding logs and the deployed Worker version. That gives the agent a starting point for investigation instead of a collection of disconnected failures. The announcement describes the open beta.

A configured automation can send an issue when failures cross a chosen threshold or return after a quiet period. The agent can investigate and propose a pull request; the described workflow leaves review and deployment with the team. The change is an automatic handoff from error detection to investigation, with the evidence needed to locate the problem. Workflow details.

Key takeaways · AI-distilled

Issues is built into the Workers runtime and enabled with one config line, with no SDK. It records uncaught exceptions, failed invocations, 5xx responses, console.log and console.error output, and logs with stack traces, and flags runaway alarms and log-heavy loops.
Automations fire when an issue crosses an occurrence threshold or returns after a quiet period, sending the source-mapped stack trace, surrounding logs and traces, Worker version and app context to Claude Code, Cursor, Devin, a generic webhook, or chat and on-call tools.
Cloudflare recommends adding user, account or session identifiers through the runtime's built-in OpenTelemetry API, so each occurrence shows whether failures cluster in one account or session before the issue goes to an agent.
Turning Issues on for Cloudflare Workflows surfaced two bugs within a day: a control-plane migration stuck retrying on a SQLite foreign key error, and instance deletion that could exceed a Workers subrequest limit. Cloudflare OS traced both into the code and proposed fixes.

Terms in this piece · Glossary

AI agent — An AI system that doesn't just answer once but works toward a goal in a loop — taking actions, reading the results, and deciding what to do next.
context window — The maximum amount of text a model can consider at once — its working memory for the current conversation or task.

Read the source blog.cloudflare.com