Agent Ops

Most agents look great in a demo and break in week two. Agent Ops is the operating discipline that prevents that: evals you trust, logs you read, retries you tune, and a way of knowing when an upgrade silently regressed a workflow that has been quietly running for months.

Below: everything we have published on agent ops.

Start here

Coverage of evals, observability, and the day-two operating discipline is on the roadmap. The first workflow article in this topic lands once we publish our internal eval harness.

All workflows

• May 5, 2026 Run autonomous workflows 24/7 with Claude Code Routines Set one recurring task to run on Anthropic's servers, with your laptop closed.