Skip to content

ai.adoc

Latest commit

 

History

History
87 lines (67 loc) · 3.66 KB

ai.adoc

File metadata and controls

87 lines (67 loc) · 3.66 KB
Table of Contents

Artificial Intelligence

IDEasy supports to bring AI to your teams boosting their efficiency. Mainly this is just done with generic features that can be also used for AI. This guide gives some special hints to make your project and team AI ready with the help of IDEasy.

AI Tools

AI tools like a CLI for some famous AI are just another tool for IDEasy. Therefore, simply install it via ide install «tool» [«version»] to your project. If the AI tool of your choice is not available in IDEasy yet, please create a feature-request. To auto-install the tool for all your team-members on project creation, simply add it to the variable IDE_TOOLS. For further details see also software.

AI Plugins

Plugins with AI support are just another plugin for IDEasy and can be configured as described in the plugin section.

Typically, developers want to have the AI integrated into their IDE. Therefore, we already preconfigured and tagged common AI plugins for various IDEs. To give an example, lets assume you are using IntelliJ and you want to use GitHub Copilot you can find the plugin configuration here. In your forked settings ensure that all your desired plugins are activated via

active=true

p.s.: If your settings are old the property may still be named plugin_active what is the legacy name for devonfw-ide compatibility.

AI Config

You AI tools shall be configured and tailored to the needs of your project. This includes aspects like instructions, skills, hooks, etc. However, common AI tools expect them in configuration folders directly in your workspace top-level directory. Even worse there is no common standard yet and every AI tool has its own naming conventions. To avoid anti-patterns and still give you high flexibility and usability, we created a special feature to automatically create additional symbolic links after cloning of a repository.

The following repository configuration in $IDE_HOME/settings/repositories/ai.properties illustrates the idea as example:

path=ai
git_url=https://github.com/devonfw/example-ai-repo.git
workspaces=*
link=.claude,.github,.kiro
active=true
tags=ai

This will clone the specified git repository into the sub-folder ai of your workspace main and also link it to every additional workspace you have (workspaces=*). Further it will create the specified symbolic links in your workspace pointing to the ai repository.

This will result in the following structure in each of your workspaces:

/
├──/ .claude/ → ai/
├──/ .github/ → ai/
├──/ .kiro/ → ai/
├──/ ai/
│  ├──/ hooks/
│  ├──/ rules/
│  ├──/ skills/
│  ├──/ steering/
│  ├──/ CLAUDE.md
│  └──/ copilot-instructions.md
└──/ .../

This is just an example. Typically, you will only need support for a single type of AI. If you have the AI tool specific folder already in your git repository, simply configure something like this:

link=.github=.github

This will create a symbolic link called .githib pointing to the .github folder inside your git repository (e.g. ai/.github).

However, you can even configure links like:

.github/hooks=myhooks/github

So if your git repo would contain the hooks for github copilot in the folder myhooks/github those will be linked to appear as .github/hooks in your workspace. We hope this gives you ultimate flexibility to solve all problems you may have.