generate-skills command, you can equip AI coding assistants such as Cursor, Claude Code, Windsurf, or GitHub Copilot to create, configure, and inspect your Minimus private images using natural language prompts.
Installation & authentication
Download the latest binary and authenticate to minicli. See minicli installationCommands
1. version
Display the current version of the minicli binary. Does not require authentication.2. generate-skills
minicli is designed to be natively AI-agent ready. You can empower your favorite AI coding assistants (such as Cursor, Windsurf, GitHub Copilot, or other local LLM tools) to natively understand and interact with minicli.
Because different AI tools look for project instructions in different directories, use the --output flag to add the skills in the appropriate folder for your specific assistant.
Example: Equipping Cursor
Cursor looks for custom skills in the.cursor/skills directory of your project.
Example AI workflows with minicli
Once the skills have been generated, you can open your AI chat interface and start prompting. Here are a few ways to interact with your agent:Create a new private image
Create a new private image
Ask the agent to scaffold a completely new image using a specific starter image and inject environment variables and packages.
“Using minicli skills, create a new private image calledmy-test-image-skillsbased on thegostarter image, with packagescurlandgit, and environment variableENV=production. Walk me through the steps.”
Modify an existing image
Modify an existing image
The agent can read your existing image recipes, update them, and re-apply the changes without you needing to manually edit the YAML or run the CLI commands yourself.
“I have an existing private image calleddemo-image. Export its config, add theagettyandagetty-openrcpackages to it, as well as the env varSOMETHING=INBARand the file bundlebb-2, and re-apply.”
Query and inspect images
Query and inspect images
Let the agent do the querying for you when you need to check what’s currently running or inspect the exact configuration of a specific image.
“Ok, now list the images, then get all details on the demo-image image and print them to me.”
3. image init
Create a local YAML template with all supported image fields, pre-populated with empty values for you to fill in.
The generated file will include notes to help clarify its use. Below is the default output and the bare structure stripped of the notes:
image recipe example
4. image apply
Create or update a private image recipe from a local YAML file. Follow up with the commandimage build submit to trigger a build of your updated image.
If the YAML is invalid, minicli returns a detailed list of errors, for example:
5. image build submit
Trigger an asynchronous build for a private image. Returns immediately without waiting for the build to complete.- You can run
minicli image build status --name {private-image-name}to track progress. - Once the build is complete, you can view the image’s detailed build report in the Creator UI. This is particularly helpful if the build did not succeed. The image build report will show you which packages were incompatible. Learn more
6. image build status
Look up the current status of a private image build.7. image export
Retrieve the configuration of a remote private image and output it in YAML file. You can export private image recipes to track changes in git.
YAML output example:
8. image list
List your Minimus private images:
You can also list the starter images available in your license for building private images:
9. image get
Get the full details of a private image.10. bundle list
List all file bundles in your account. File bundles are used to include public keys for internal PKIs (Public Key Infrastructure certificates) and override configuration files and more. Learn more11. image delete
Delete a private image from your account. Prompts for confirmation unless--force is passed. The image will also be removed from Creator and will no longer be available for use.