kubefwd (Kube Forward) is a Kubernetes bulk port forwarding command-line tool for local development, with the goal of letting developers seamlessly access services running in a Kubernetes cluster while working on their local machine. When a local application needs to connect to services running in the development cluster, such as the database db:5432, authentication service auth:443, cache redis:6379, etc., kubefwd can expose them locally by service name, just as if they were inside the cluster, without the need to write environment-specific configuration, set up local services, or use a Docker Compose file. As long as you run kubefwd, the existing connection strings in the application will work. It is officially described as the foundational use case for "local development, connecting to Kubernetes": if the code uses in production http://api-gateway:8080，那么在笔记本电脑上使用 kubefwd will work the same way.
In terms of main capabilities, kubefwd supports bulk forwarding of all services in a namespace with a single command; it assigns each service a unique loopback IP (127.x.x.x), thereby avoiding port conflicts; it automatically writes service names into /etc/hosts and establishes port forwarding via the Kubernetes API, making services accessible by name. It provides an interactive TUI interface for real-time monitoring of service status and traffic metrics, viewing bytes in/out and HTTP activity, and supports streaming container logs in the TUI (Pod Log Streaming). When a Pod restarts or a connection drops, the tool automatically reconnects. In addition, it provides a REST API for programmatic control through HTTP endpoints, and supports MCP for integration with AI assistants such as Claude Code and Cursor.
In terms of usage, kubefwd is a command-line tool. For installation on Linux, .deb, .rpm, or .tar.gz packages can be downloaded from releases. The prerequisites are that kubectl is configured and can access the cluster, and root/sudo privileges are required to modify /etc/hosts and network interfaces. After running, it discovers the services in the namespace it is in, assigns each service a unique loopback IP, updates the service names in /etc/hosts, and establishes forwarding through the Kubernetes API; after that, services can be accessed by service name just as from inside the cluster. The official documentation is located at kubefwd.com and includes Getting Started (installation and configuration), User Guide (interface and shortcuts), Configuration (CLI options), Advanced Usage (multiple clusters, selectors), REST API, MCP integration, troubleshooting, architecture notes, and comparisons with tools such as Telepresence and mirrord.
The applicable scenario is mainly local development: developers who need their local code to connect to multiple services in a cluster and want to continue using in-cluster service names and existing connection strings. In terms of limitations, the materials do not provide platform details beyond supported operating systems, specific performance data, pricing, or fee information; this content is not provided for now. The project uses the Apache License 2.0 and is open-sourced by Craig Johnston, sponsored by Deasil Works, Inc.
