mirrord: a tool that runs local processes inside a live Kubernetes cluster to bypass the deploy cycle

What it solves

mirrord eliminates the need to deploy code to a Kubernetes cluster to test it. It allows developers and AI coding agents to run local processes as if they were inside a live cluster, providing immediate feedback without the slow deploy cycle or disrupting other users in the cluster.

How it works

When a user selects a target pod to impersonate, mirrord launches a temporary pod on the same node. This agent pod acts as a bridge, mirroring incoming traffic, routing outgoing traffic, and syncing environment variables and file system access between the local process and the target pod in the cluster.

Who it’s for

Software developers and AI coding agents (such as Claude Code, Cursor, and Copilot) who work with Kubernetes-based applications and need to verify code changes against real cluster services and data.

Highlights

• Multi-platform support: Available as a VS Code extension, IntelliJ plugin, and CLI tool.
• Real-time context: Accesses real environment variables, service responses, and queue contents from the live cluster.
• AI Agent Integration: Specifically designed to work with AI coding agents to run and verify generated code without deployment.
• Zero-disruption: Provides the feedback of a deploy in seconds without actually deploying or affecting other cluster users.