Clio is an AI-powered copilot designed to help you with DevOps-related tasks using CLI programs. It leverages OpenAI's capabilities to provide intelligent assistance directly from your command line.

Note: Clio is designed to safely perform actions. It won't do anything without your confirmation first.

• Kubernetes Management: Interact with your Kubernetes clusters using kubectl, helm, and other CLIs.
• AWS Integration: Manage your AWS resources using the aws CLI.
• Azure Integration: Manage your Azure resources using the az CLI.
• Google Cloud Platform Integration: Manage your GCP resources using the gcloud CLI.
• DigitalOcean Integration: Manage your DigitalOcean resources using the doctl CLI.
• EKS Management: Manage your EKS clusters in AWS using eksctl and aws CLI.
• GitHub Integration: Interact with your GitHub repositories using the gh CLI.
• Easily Customizable: Add new capabilities with no code.

To install Clio, you can use Homebrew:

brew install gptscript-ai/tap/clio

Alternatively, you can clone the repository and build the project manually:

git clone https://github.com/gptscript-ai/clio.git
cd clio
make build

To start Clio, simply run:

clio

Clio will prompt you to authenticate with GitHub to allow access to the AI model powering Clio. You can also set a custom personal OpenAI API key and base URL using environment variables, refer to clio --help for specific environment variable names.

Clio is composed of multiple internal agents. There are several built-in agents that provide functionality for interacting with Kubernetes, AWS, GCP, GitHub, etc., but you can easily add your own agents to extend the functionality of Clio. The built-in agents are located in the agents directory, with each file being a separate agent. To create a custom agent, you must write a new GPTScript and place it in the $XDG_CONFIG_HOME/clio/agents directory.

A custom agent is any GPTScript with the requirement that it minimally must contain the following lines

chat: true
context: github.com/gptscript-ai/clio/context

You can refer to the GPTScript documentation for all the capabilities of GPTScripts, but for now the below example is typically all you need to know.

For this example, we will add a custom agent that is specialized for GoReleaser 2. We will create a file called goreleaser.gpt. The finished example is available in the examples directory.

The GPTScript starts with a metadata block that defines the name of the agent, a description, and the required context and chat fields. It is then followed by the prompt that will tell Clio how to behave when this agent is invoked.

Name: GoReleaser
Description: Agent for GoReleaser 2 using the goreleaser CLI
Chat: true
Context: github.com/gptscript-ai/clio/context

You are an expert at goreleaser. You can run the goreleaser CLI and help manage the goreleaser config file.

Rules:
1. Before changing the config, always show goreleaser config to the user for confirmation. After they agree, then write to disk.
2. Make sure "version: 2" line is always in the goreleaser config.
3. If the user asks to build, do a snapshot build.
4. Always search the internet for relevant information when asked a question or to do a task.

First ask the user what would they like to do with regards to GoReleaser.

To test this agent out you can run clio goreleaser.gpt. After testing the agent, you can modify the text if you don't like the exhibited behavior. There is no defined format for the prompt. The fact that the example has the structure with "Rules" in it is just a convention but not technically required.

To make the agent even more useful, you can extend the agent to have dynamic contextual information. To do this, we are going to add another "context tool" to the agent. Context tools add capabilities to the agent by prepending the output of the tool to the prompt.

In the below example, we add a new line for context: additional-environment to the metadata block. We then define the additional-environment context tool. This tool will show the user the current goreleaser version, the JSONSchema for the goreleaser config file, and the help output for the goreleaser CLI and the build subcommand. It will also show the user the current goreleaser config file if it exists.

Name: GoReleaser
Description: Agent for GoReleaser 2 using the goreleaser CLI
Chat: true
Context: github.com/gptscript-ai/clio/context
Context: additional-environment

You are an expert at goreleaser. You can run the goreleaser CLI and help manage the goreleaser config file.

Rules:
1. Before changing the config, always show goreleaser config to the user for confirmation. After they agree, then write to disk.
2. Make sure "version: 2" line is always in the goreleaser config.
3. If the user asks to build, do a snapshot build.
4. Always search the internet for relevant information when asked a question or to do a task.

First ask the user what would they like to do with regards to GoReleaser.

---
Name: additional-environment

#!/bin/bash

if ! command -v goreleaser; then
    echo 'Inform the user goreleaser is not installed or available on the path'
else
    goreleaser --version || true

    echo 'The JSONSchema for .goreleaser.yaml is as follows:'
    goreleaser jsonschema || true

    echo Additional CLI help
    echo
    goreleaser --help || true
    goreleaser build --help || true

    for i in .goreleaser.yaml .goreleaser.yml; do
        if [ -e $i ]; then
            echo
            echo "The current goreleaser config in ./$i:"
            echo
            echo '```yaml'
            cat $i
            echo '```'
        fi
    done
fi

That is now our finished agent. You can place the goreleaser.gpt file in the $XDG_CONFIG_HOME/clio/agents directory and then the next time you run clio you will see your agent in the list of agents, and it can be referenced by doing @goreleaser <your question>.

This project is licensed under the Apache License 2.0. See the LICENSE file for details.