Get the Routines Operator Playbook
Anthropic just shipped Claude Code Routines — one /schedule command that turns Claude Code from a tool you press enter on into a teammate that reads your repo, notices when something breaks, and opens a PR before you have opened your laptop. Maya from Anthropic's Applied AI team unveiled it at Code with Claude, alongside the proof: weekly Claude Code PRs are up 200% and one Anthropic engineer was drowning in documentation — until two Routines closed the gap. This is the plain-English operator playbook: the only mental model you need (the 3-decision framework — Trigger, Context, Steerability), the trigger menu (schedule vs native GitHub events vs custom webhook — and exactly when to pick which), the 5 native GitHub events Anthropic ships out of the box (PR merged, PR opened, issue opened, label added, release cut), a complete custom-webhook payload example, the connector strip (repos + Slack + Drive + GitHub MCP + Datadog/Grafana + Twilio — "context is the ceiling of success"), the three steerability patterns (agent-on-agent generator-critic review, the live web session you can interrupt mid-run and resume a week later, and the verify-outputs habit), three paste-ready /schedule prompts you can copy tonight (Docs Sync — the exact internal Anthropic routine; Deploy Verifier — webhook from CI/CD plus monitoring plus Slack with a go/no-go rollback recommendation; PM Backlog Triage — weekly schedule plus GitHub issues plus Slack ranked by revenue and churn impact with draft PRs for the top three), the generator-critic diagram (Routine A opens the PR, Routine B triggers on that PR being created and reviews the work before any human reads it), and the 5 fatal mistakes that kill a Routine (running on local cron, no steerability plan, too-narrow context, no verification step, skipping the critic routine) with the one-line fix for each.
Free. No spam. Unsubscribe anytime.