CLOUDFLARE has introduced a real-time, built-in error monitoring feature for Cloudflare Workers called Issues, currently in open beta. The service aims to streamline how production failures are handed off to an agent by automatically grouping repeated incidents and collecting the relevant context, logs, traces, and the Worker version alongside the error.
Issues is designed to reduce the need for teams to manually sift through raw telemetry to understand the scope and cause of a fault, instead surfacing a consolidated view for quicker triage and remediation.
Key capabilities include automatic detection of repeated exceptions, 5xx responses, and error logs, with the option to forward the error data to a configured coding agent. It supports sending error details, stack traces, logs, and traces, and can trigger the agent’s workflow—from triage to data queries to opening a pull request.
The system also contextualises failures by attaching user-supplied identifiers via the OpenTelemetry API, enabling teams to determine whether issues cluster around particular accounts or sessions. Automations can route issues to built-in agents (Claude Code, Cursor, Devin), generic webhooks, or chat and incident-management tools, with the agent able to query related logs and traces through Cloudflare MCP if deeper investigation is needed.
The article provides practical steps to enable Issues and set up automations, and cites internal use where Issues helped resolve two Workflows bugs by surfacing and routing incidents directly for automated analysis.