Operations | Monitoring | ITSM | DevOps | Cloud

The faster you ship with AI, the wider your test coverage gaps get. Chunk scans your codebase, finds what's untested, writes the tests, and opens a PR.

AI coding tools help you ship features faster than ever, but test coverage often can't keep up. In this video, we show you how CircleCI's Chunk autonomous CI/CD agent finds untested code in your codebase, writes tests to cover it, and opens a pull request for your review. What you'll learn: Chunk works directly inside your CI/CD pipeline, giving it access to your build history, test results, and coverage reports. That means smarter tests, not just more tests.

AI agents write the code. CircleCI makes sure it actually ships. The CircleCI plugin for Codex is live.

RalphCI is an open-source, CI-enabled agentic coding loop built by the Loop Lab at CircleCI. You write a spec, and the agent breaks it down into tasks, builds your application step by step, commits to GitHub, and runs your full CI pipeline on every iteration. If anything fails—linting, tests, security scans, missing files—a CI Doctor sub-agent detects the failure, reads the stack trace, and fixes it automatically. In this video, Ryan Hamilton demos RalphCI by building a classic Snake game end-to-end with zero manual coding.

Hitting your PR count feels productive, but maybe it's not. Sometimes you have to slow down to 10x.

You've been staring at the problem too long. Fresh eyes break what you can't.

Chunk analyzes your entire codebase, refactors what matters, and validates every change before opening a PR.

Shipping feels slow. Reviews take forever. Things break more than they should. That’s the bottleneck.

Claude Livecaster is now public on CircleCI Research. In this update, Ryan Hamilton walks through the newly open-sourced repo, seven built-in simulation scenarios, and a new two-voice broadcast format featuring an anchor and a field correspondent narrating the action together. The demo scenario: Pipeline Wars, six CI pipelines racing across three providers, with Claude providing live color commentary on every Docker build failure, OOM kill, and production rollout.

What if you didn't have to stare at logs while your AI agent worked? In this Loop Lab experiment, Ryan Hamilton built Claude Livecaster, a tool that gives Claude a live voice to narrate long-running agentic processes like a sports commentator. The demo: six AI models (GPT, Gemini, and Claude variants) race through a CI/CD benchmark, and Claude calls the whole thing play-by-play. Rate limit hits, comeback stories, photo finishes, all of it, out loud.