Anthropic’s Claude Code Routines

Anthropic has added “Routines” to Claude Code, letting developers save a prompt, attach repositories and connectors, and run the job automatically in Anthropic’s cloud. (code.claude.com) Anthropic’s documentation says a routine can start on a schedule, from an application programming interface call, or from GitHub events such as pull requests and releases. The feature is marked “research preview,” and Anthropic says its behavior, limits, and application programming interface surface may change. (code.claude.com) The company says routines are available on Pro, Max, Team, and Enterprise plans when Claude Code on the web is enabled. Anthropic also says users can create and manage routines in the web interface or from the command line with `/schedule`. (code.claude.com) Claude Code is Anthropic’s coding agent for terminals, integrated development environments, and GitHub. Anthropic says the tool can read a codebase, edit files across repositories, run tests, and use tools such as the GitHub command line interface. (anthropic.com) (github.com) Routines extend that model from one-off coding sessions to repeatable background jobs. Anthropic’s examples include nightly pull request review, weekly backlog maintenance, deploy verification after production releases, and documentation updates after merged code changes. (code.claude.com) Anthropic says the runs happen on Anthropic-managed infrastructure, so they continue when a developer’s laptop is closed. That shifts Claude Code closer to continuous integration tools and workflow automation products that watch repositories and trigger actions without a person sitting at the keyboard. (code.claude.com 1) (code.claude.com 2) The pricing question is built into the product design. Anthropic’s docs say each plan includes a number of daily routine runs, and organizations that enable extra usage can keep routines running on metered overage after they hit a daily cap or subscription limit. (code.claude.com) Anthropic has been moving Claude Code toward longer-running autonomous work for months. In September 2025, the company said it was adding checkpoints for autonomous operation, and in February 2026 it said Claude Code users could assemble agent teams to work on tasks together. (anthropic.com 1) (anthropic.com 2) Anthropic also says most of its own code is now written by Claude Code, and it has used customer case studies to argue that the product is moving from an assistant inside a terminal to a system that can handle larger parts of engineering work. Routines package that pitch into a feature teams can wire into GitHub, release pipelines, and issue trackers. (anthropic.com)