Helm, AICR, OCI, YAML, Argo, Flux, Sveltos and more

Configuration catalog for Agents and Kubernetes

ConfigHub Workshop gives agents, and the people working with them, a catalog of tested configuration as data, tools to act on it, and a ConfigHub on-ramp.

Start with your own chart or app. Use cub directly or ask your agent to inspect the rendered configuration, compare changes and run checks. Local checks need no account.

I have an existing app Check my charts and values

When your team needs to keep and share the result, ConfigHub adds version history, approvals and releases. Argo CD or Flux still delivers it using your existing setup.

Build a platform or fleet Set up my AI agent
use cub open source, or upload to ConfigHub Server
# what will this chart install?
$ cub config check redis
14 objects: 3 ConfigMap, 1 NetworkPolicy, 2 PodDisruptionBudget,
1 Secret, 3 Service, 2 ServiceAccount, 2 StatefulSet

# give a whole stack: an inference platform
$ cub stack sandbox eks-inference
=> CHECKED  130 objects, no conflicts across 8 components

# Use ConfigHub to store, place and publish default redis config
$ cub variant upload --component redis --variant base redis.yaml
$ cub variant create prod redis-base --target prod/target
$ cub release publish redis-prod
# note: your Argo CD or Flux pulls that digest and deploys it
For your agent

Prompt Claude or Codex to get started

Paste this into Claude Code, Codex, or any agent that can run a shell.

Use the ConfigHub Workshop catalog for Kubernetes config work. It holds configurations rendered to exact objects, each with a record of what was checked and what was not, across Helm, AICR (AI infrastructure), Timoni, Kubara, plain YAML, and OCI. Start at https://confighub.github.io/helm-expt/site/llms.txt, search only enough of the relevant section or listings index to identify one entry, then read that entry's listing and one relevant Guide. Do not load whole indexes or the whole catalog into model context. Do the work with `cub` and the cub workshop plugin (https://github.com/confighub/cub-workshop): `cub config`, `cub config values` and `cub config diff`, `cub stack check` and `cub stack sandbox`, `cub app match`. Prefer exact versions and digests.

Or install the skill in your project. Use with your AI has the fuller prompt.

npx skills add https://github.com/confighub/helm-expt/tree/main/skills/config-workshop

New to cub and ConfigHub? Install the cub CLI first. The public catalog uses OCI and standard config formats. You don’t need to sign up until you want to save changes in ConfigHub

The Catalog

What the Catalog holds

  • Configs: 528 configurations. 357 are tested and rendered to the exact objects they install. 133 are AICR recipes rendered as Argo CD Applications, and they have not been published or run. 38 are NIMService model variants kept as exact objects with their plans, and 37 of them are published as OCI.
  • Stacks: 12 stacks, sets of configs checked together before anything runs.
  • Apps: 25 worked example apps, plain or delivered by Argo CD, Flux or a generator.
  • Plugins: 12 cub plugins, each marked by its state.
  • Guides: 58 known paths an agent walks with you, the five journeys first.
  • Docs: how the model works, why you can trust it, and every reference.
The short version

What is the Workshop?

The Catalog supplies configurations and evidence. Workshop combines it with Guides and tools. cub is the command-line interface that you and your agent use. ConfigHub Server keeps shared configuration and manages its releases. You can run local checks without a ConfigHub account. See the full detail.

Other catalogs give you charts. We give your agents the rendered objects, digests and recorded checks to assess them before use.

Configuration as data starts with flattened, explicit objects you can inspect and compare. OCI is a shared transport for this configuration. Each entry records its source-specific processing and checks, including whether flattening is supported. How the model works gives the three steps.