Your AI App Passed the Vibe Check. Now What?
Run a few prompts. Looks good. Ship it. According to Laurie Voss, that's still how a surprising amount of AI software gets tested.
Operations | Monitoring | ITSM | DevOps | Cloud
The Place Where Modern Operations & Technology Come Together
What Every ERG Leader Learns
People want to support your community. Sometimes you just have to invite them in.
What is mutation testing?
A test suite can be all green and hit 100% line coverage and still miss bugs. Coverage measures which lines ran during the tests, not whether the assertions actually caught a defect. A test that calls a function but never checks the return value still counts toward the coverage number. The bug it would have prevented still ships.
Why you should use Language Server Protocol (LSP) with Claude Code
Agentic coding tools like Claude Code can write, refactor, and debug across an entire codebase, but by default they read code as plain text, the way grep does. The Language Server Protocol (LSP) changes that: it’s the same code-intelligence layer an IDE uses, and wiring it into an agent lets it read code by meaning instead of by string match. The bigger the codebase, the more a wrong guess about a symbol costs, and the more that structural view pays off.
Cut your environment setup time in half with Chunk sidecar snapshots
When you’re building with AI, you can get a lot done in 30 seconds. Waiting minutes for CI feedback on your latest change can feel like an eternity. Chunk sidecars are designed to give you feedback fast, running your full test suite against the same Linux environment as CI, directly inside the agentic loop. Traditional CI pipelines can take five or ten minutes to catch a basic lint error or failing unit test.
Isolating Risk: Why 5,000-Line PRs Are Dead (Even With AI)
The real unlock is flagging what actually needs human review, not shipping 5,000-line PRs and hoping for the best.
Why AI Is Following the Same Path as Software Development
Devs have a habit of solving their own problems before anyone else's and the AI boom is repeating that pattern.
Chunk sidecars: Inner Loop Validation for AI Coding Agents
Your agent writes code fast, but you shouldn't have to see it until it's right. Chunk sidecars are lightweight microVMs that work inside the agent loop, requiring agents to pass pre-push validation in a CI-like environment before they declare they're "done." That means no massive CI pile-ups, no long round-trips that risk resetting your agent's context. You're sending code you already know is good.
Untested AI is unshippable AI, with Laurie Voss of Arize
Most AI applications in production right now were shipped on vibes. A developer ran a few queries, liked what they saw, and pushed to prod. Laurie Voss argues that's the core reason so many AI products feel broken, and the fix is simpler to name than it is to do: write the tests.
We told Claude to break our tests. Here's what happened.
Catch failing tests before they ever leave the machine with just two hooks and Chunk sidecars.
The engineers who handle change best aren't keeping up with every tool
Hywel Carter on the mental model that travels with you no matter what ships next.
Agent Hooks + Chunk sidecars: Stop Broken AI Code Before It Hits CI
AI agents write code fast, but the feedback loop usually can't keep up. In this tutorial, you'll see how to wire Chunk sidecars into your agent's hooks so basic failures get caught before they ever reach your CI pipeline. We'll walk through the two hooks that chunk init writes automatically: Both hooks return exit 2 on failure, blocking the commit or keeping the turn open so the agent can fix its own mistakes with no manual prompting required.
Chunk sidecars: CI Feedback Before the Push
Chunk sidecars give your agent CI-grade feedback in seconds. No push, no pipeline wait. Catch failures and fix them while the agent is still in context.
Agentic validation needs different infrastructure
Previously, I described some core approaches to validating agent written code: feedforward and feedback techniques. Feedforward techniques are about avoiding errors up front, for example by coming up with better prompts and planning strategies. Feedback gives agents a signal that they have actually achieved a task. Feedback is a key part of common agentic patterns like Ralph loops or the /goal commands in Codex and Claude Code: keep working until some known condition passes.
How to set up blue-green deployments with CircleCI
How to set up blue-green deployments with CircleCI.
The skill gap AI can't close, with Hywel Carver of Skiller Whale
If 90% of code is now AI-generated, what happens to the engineers who never had to write it themselves? The ability to review code well, to catch what the LLM missed, depends on mental models that only come from practice. That tension sits at the center of this conversation.
Run CI Tests Without Pushing: Microbuilds with Chunk sidecars
AI coding agents write code faster than your pipeline can catch mistakes. What if the agent could validate against CI before you ever push? In this 5-minute demo, we set up CircleCI's Chunk CLI and run a microbuild using Chunk sidecars, secure Linux microVMs that spin up in ~1 second in your CircleCI account, mirror your working directory (no git push required), and give your agent CI-grade feedback while it's still in context.
Patterns of validation
This post covers some techniques to get agents to validate their work. We’ll cover why this is important, a variety of types of checks you can employ, and mechanisms you can use to enforce them.