AWS Context Switch with Tab Autocompletion

How to add tab autocompletion for a custom awsctx command in Zsh (like kubectx)

If you frequently switch between multiple AWS profiles, you’ve likely wished for a quicker, more intuitive way to manage them from the command line. Inspired by tools like kubectx for Kubernetes, this guide walks you through setting up a custom awsctx command with tab autocompletion using Zsh and Bash completion utilities.

By the end of this guide, pressing <Tab> after typing awsctx will autocomplete AWS profile names—dramatically improving speed and accuracy when working with the AWS CLI.

Step 1: Create the awsctx Function

In your ~/.zshrc, define the awsctx function. This will let you switch AWS profiles using either an argument or an interactive FZF-based selector:

function awsctx() {
  if [ -z "$1" ]; then
    profile=$(aws configure list-profiles | fzf)
    if [[ $profile != "" ]]; then
      export AWS_PROFILE=$profile
    fi
  else
    export AWS_PROFILE=$1
  fi
  echo -e "\033[32m$AWS_PROFILE\033[0m selected"
}

• If no profile name is given, it launches fzf for fuzzy selection.
• Otherwise, it sets AWS_PROFILE directly.
• The selected profile is displayed in green for clarity.

Step 2: Enable Bash Completion in Zsh

To use traditional Bash completion utilities inside Zsh, you need to enable them:

autoload -Uz compinit && compinit
autoload -Uz bashcompinit && bashcompinit

Place these in your ~/.zshrc before sourcing the completion script.

Step 3: Write the Autocompletion Script

Create a file named ~/.zsh/completions/_awsctx_completion with the following contents:

#/usr/bin/env bash

_awsctx_completion() {
  if [ "${#COMP_WORDS[@]}" != "2" ]; then
    return
  fi

  local profiles=($(compgen -W "$(aws configure list-profiles)" "${COMP_WORDS[1]}"))

  if [ "${#profiles[@]}" == "1" ]; then
    local profile=$(echo ${profiles[0]/%\ */})
    COMPREPLY=("$profile")
  else
    COMPREPLY=("${profiles[@]}")
  fi
}

complete -F _awsctx_completion awsctx

Then make it executable:

chmod +x ~/.zsh/completions/_awsctx_completion

This script leverages compgen, a Bash builtin used to generate possible completion matches. It filters AWS profile names using the current partial input.

Step 4: Source the Script in .zshrc

Still in your ~/.zshrc, source the autocompletion script so it’s loaded into your shell session:

source $HOME/.zsh/completions/_awsctx_completion

Your final .zshrc additions should look like this:

function awsctx() {
  if [ -z "$1" ]; then
    profile=$(aws configure list-profiles | fzf)
    if [[ $profile != "" ]]; then
      export AWS_PROFILE=$profile
    fi
  else
    export AWS_PROFILE=$1
  fi
  echo -e "\033[32m$AWS_PROFILE\033[0m selected"
}

autoload -Uz compinit && compinit
autoload -Uz bashcompinit && bashcompinit
source $HOME/.zsh/completions/_awsctx_completion

Test It Out

Restart your terminal or reload your .zshrc:

source ~/.zshrc

Then type:

awsctx <Tab>

You should see all available AWS profiles populated dynamically!

How It Works

• compgen dynamically generates matching profile names using your current input.
• COMP_WORDS contains the current command-line arguments; this is used to determine context.
• COMPREPLY returns the list of completion options to Zsh.
• complete -F registers your function for the awsctx command.

Although written in Bash, this works in Zsh via bashcompinit, which bridges Bash-style completion scripts into Zsh environments. For more background, check out resources like:

• compgen and tab completion in Bash/Zsh
• How to use Bash completion in Zsh
• Understanding COMP_WORDS in completion scripts

Bonus Tips

• Use a plugin manager like zinit, antidote, or oh-my-zsh to manage your custom functions and completions more cleanly.
• For cross-shell compatibility (Bash, Zsh, etc.), abstract completion into reusable scripts.
• Combine with tools like direnv to auto-load profiles per project directory.

With this setup, switching AWS profiles becomes just as easy as working with Kubernetes contexts using kubectx. Give it a try—and never mistype an AWS profile again.