Anthropic Debuts Repeatable Routines in Major Claude Code Automation Update Anthropic has introduced a new repeatable routines feature for Claude Code, enabling developers to automate tasks on a schedule without requiring their local machines to remain online. The feature, launched as a research preview on April 14, 2026, runs automations on Anthropic’s cloud infrastructure, allowing routines to execute even when a user’s Mac is closed or disconnected from the internet. A routine is defined as a saved Claude Code configuration consisting of a prompt, one or more repositories, and a set of connectors. Once packaged, it can be triggered on a schedule or via a webhook. This approach differs from traditional automation tools like cron jobs or GitHub Actions, which run static scripts without dynamic AI input. Claude Code routines leverage the AI model to assess context and make decisions during execution, enabling more adaptive workflows. According to Anthropic, the feature is particularly useful for tasks such as verifying software deployment by scanning CI/CD output for errors and generating reports, or triaging alert messages by analyzing incoming notifications and determining appropriate responses. Routines are designed to reduce the operational overhead previously associated with managing autonomous agents, infrastructure, and external tooling like MCP servers. Access to the routines feature is tiered by subscription plan. Pro users can run up to five routines per day, Max users up to fifteen, and Team or Enterprise users up to twenty-five routines daily. These limits are intended to balance accessibility with resource management on Anthropic’s infrastructure. The rollout of repeatable routines coincides with a broader redesign of the Claude Code desktop application. The updated interface now supports parallel sessions, allowing users to run multiple Claude Code instances side by side within a single window, managed through a new sidebar. This enhancement improves workflow efficiency for developers juggling multiple projects or environments. Anthropic positions Claude Code routines as a middle ground between static automation scripts and fully autonomous AI agents. Unlike agents, which maintain state and engage in ongoing interactions with tools and data, routines are short-lived, trigger-driven executions that do not retain context between runs. This design makes them suitable for discrete, repeatable tasks while avoiding the complexity and unpredictability associated with persistent agent behavior. The feature builds on prior developments in the Claude ecosystem, including the introduction of auto mode the previous month and the recent graduation of Claude Cowork from research preview with added enterprise capabilities. Together, these updates reflect Anthropic’s strategy to deepen integration of its AI models into software development lifecycles through practical, controllable automation tools. As organizations seek to streamline DevOps processes and reduce manual intervention in routine operations, Claude Code’s repeatable routines offer a cloud-based, AI-augmented alternative to conventional scheduling systems. By executing on Anthropic’s managed infrastructure, the feature aims to deliver reliability and scalability without imposing operational burdens on end-user devices.
Related reading