
Anthropic this week updated the widely used Claude Code to streamline common processes in coding workflows.
Claude Code Routines let developers configure once, then save, a common Claude Code automation and then invoke it automatically when needed. Automations apply to prompts, connectors and context – detail extracted from a software repository to help Claude Code understand the full codebase – not just the currently open file.
Routines run on Anthropic-managed infrastructure, so they aren’t dependent on the developer’s laptop being open and operating, the company said.
While developers are currently using Claude Code to automate development processes – scheduled jobs, infrastructure, and tools such as Model Context Protocol (MCP) servers — they’ve managed that themselves prior to Routines – which are in research preview mode.
Routines ship with access to the customers’ repositories and connectors for a simple process of building and packaging automations to run on a schedule or trigger.
Here are three ways Routines can be used by developers:
Scheduled routines: A developer gives Claude Code a prompt and a cadence (such as nightly or weekly) and it runs on that schedule. These can be applied to backlog management for triaging of new issues, as well as labeling, assigning, and posting a summary of the issues to internal communications platforms. They could also be applied to documentation “drift” to flag docs that reference changed APIs, for example.
API routines: Developers can configure Routines to be triggered by API calls. Every routine gets its own endpoint and authentication token for server access. These can be applied to code deployment verification, with Claude Code scanning error logs for regressions and posting go/no-go. They can also be used for alert triage, with Claude Code pulling a trace, correlating it with recent deployments, and drafting a fix.
Webhook routines, with a focus on GitHub: Developers will subscribe a routine to automatically kick off in response to GitHub repository events, then Claude Code will create a new session for every pull request matching filters and run a routine. For example, a GitHub routine could trigger a review of bespoke code; on pull request being opened, it can run a team’s checklist across security and performance, leaving inline comments before a human conducts their review.
With Routines, Anthropic is expanding the appeal of Claude Code for enterprise developers and IT leaders by delivering new levels of automation that align more closely with corporate development practices and processes. The continued innovation should contribute to the coding assistant’s continuing momentum.
For a 36-Hour Immersion into the FY27 Priorities that define Partner Success in the AI Era, join us at the AI Business Solutions Partner Executive Summit, running July 22-23, 2026, in Bellevue, Washington. Register today.




