Running the curl install listed on the homepage of atmos, immediately ran into the following. Is this supposed to run as sudo? First time posting here, so let me know if this is the wrong way to go about reaching out.

cat /etc/lsb-release
DISTRIB_ID=Ubuntu
DISTRIB_RELEASE=24.04
DISTRIB_CODENAME=noble
DISTRIB_DESCRIPTION="Ubuntu 24.04.4 LTS"

Does anyone tried atmos.tools/changelog/tags-and-labels ?

For some reason, these doesn't work for me (I located them in most top _defaults.yaml):

  metadata:
    labels:
      some_tag: some_value
  vars:
    tags: !labels

Inside generated .tfvars.json after atmos terraform plan:

"tags": {
    "unrelated_tag": "foo"
  },

atmos v1.223.0

Hi, there. I surprised that it not covered from day 1 of hooks introduction
Can you please take a look?
github.com/cloudposse/atmos/issues/2799

really enjoying DAG, seems to be working decently with a PoC for targeted stack precedence

Trying to implement a blue/green pattern in atmos. DAG doesn't work across multiple stacks yet does it?

The recent changes to --help is very welcomed. It's a much better DX then providing all the help information.

Hey everyone! Wondering if anyone else has seen this behavior:

We noticed that after a recent update to cloudposse/github-action-setup-atmos@v3, our atmos terraform deploy and atmos terraform apply commands in GitHub Actions started completing instantly (0 seconds, exit code 0) with no terraform output — no init, no plan, no apply.

The install path changed from /home/runner/work/_actions/cloudposse/github-action-setup-atmos/atmos to /opt/hostedtoolcache/atmos-wrapper/1.222.0/x64. Actions runs using the old path on July 3 worked fine (5-minute deploys); runs using the new wrapper path on July 4+ silently no-op.

Workaround: Setting install-wrapper: "false" resolves it — deploy commands run terraform as expected.

Are we missing a configuration step for the wrapper, or is this a known issue with the latest release? Happy to provide workflow snippets or run logs if helpful. Thanks!

@Erik Osterman (Cloud Posse) Do you know if you can set authentication for an oci JIT source component? Or if it's documented somewhere specific? Specifically a terraform.$component.source.uri

so sick !

Hi. Can you look into github.com/cloudposse/atmos/issues/2376 bug? It still exist in 1.222.0

Hello, I ran into an issue with Native CI in JIT vendored components, added a bug report for it: github.com/cloudposse/atmos/issues/2684
Basically if you source a component direct from github to use directly, and specify a workdir, the native ci planfile logic looks in the wrong spot for the planfile

Hi, do you accept issues made by AI? I found a bug it seems and made codex to file the bug report here: github.com/cloudposse/atmos/issues/2680

Is there some command to apply all components in a specific stack? All flag applies all components in all stacks, not specific one.

Hey, team. I think there might be a documentation snafu.

Here in the Native CI docs, it shows an example of how one can set up Github OIDC auth. This did not work. Atmos complained that the profile was not valid.

auth:
  providers:
    github-oidc:
      kind: github/oidc
      region: us-east-1
  identities:
    plat-dev/terraform:
      provider: github-oidc
      role_arn: arn:aws:iam::111122223333:role/atmos-terraform
      default: true   # Use this identity unless a component overrides it

Fortunately, atmos auth validate was pretty helpful and suggested useful changes, which I was later able to confirm in the more thorough auth section in the Atmos docs. I would humbly suggest that the example for the Native CI section should look something like this.

auth:
  providers:
    github-oidc:
      kind: github/oidc
      region: us-east-1
      spec:
        audience: <http://sts.amazonaws.com|sts.amazonaws.com>
  identities:
    plat-dev/terraform:
      kind: aws/assume-role
      via:
        provider: github-oidc
      principal:
        assume_role: arn:aws:iam::111122223333:role/atmos-terraform
      default: true   # Use this identity unless a component overrides it

Hey, everyone. Random unimportant question.

I think I know the answer, considering the obvious focus of Atmos and Atmos Pro using Github/Gitlab. But I'm curious. Has the team ever considered a Kubernetes Operator approach similar to tofu-controller or burrito? We've been using Atmos for some time now, but we're also a heavy Kubernetes + ArgoCD shop. So, naturally, I found the approach intriguing.

hello, anyone able to confirm if planfile artifact storage to gh artifacts has been implemented as of v1.221.1?

We've published a media kit for atmos! If you've wanted some atmos logos, we got 'em.

https://atmos.tools/media-kit

I’m looking forward to Atmos secrets management. We use Sealed Secrets for our application layer since everything runs on Kubernetes. Managing secrets as code makes deployments much more portable across regions and accounts.

However, for our infrastructure layer, secrets are stored in Secure SSM Parameters and Secrets Manager. The lack of portability has become a small pain point when managing a DR region, so I’ve been considering a move to SOPS. After another practice run of our DR runbook last week, I’m definitely making the switch soon. The timing of Atmos secrets management with SOPS support couldn’t have come at a better time.

Btw, it's a really minor issue, but atmos ai skill does not support vscode natively, as it expect skills in some specific folders.
I've had to make it available this way:

cd <my VScode workspace>
atmos ai skill install cloudposse/atmos -y --force
mkdir -p .github/skills && for d in ~/.atmos/skills/cloudposse/atmos/*/; do ln -sfn "$d" ".github/skills/$(basename "$d")"; done

I recorded a 20-minute overview of the Atmos Changelog covering all of the latest features that have come out over the past month.

https://us02web.zoom.us/clips/share/tQmxbpnfQxGWs-bpWl4sAA

• Atmos Git for GitOps — Manage repositories as first-class Atmos resources; clone, diff, commit, push, and publish generated artifacts through lifecycle hooks.
• Secrets Management — Declare component secret dependencies and manage secrets natively in Atmos.
• CI Caching — Cache toolchains, providers, modules, vendored components, and remote imports to accelerate local and CI runs.
• Terraform Registry Cache(Experimental) — Cache providers and modules for faster, reproducible, and more resilient builds.
• Terraform/OpenTofu RC Management(Experimental) — Define CLI runtime configuration in atmos.yaml; Atmos manages temporary RC files automatically.
• Terraform DAG Concurrency — Dependency-aware execution with concurrent planning and deterministic ordering.
• Interactive Workflows & Commands — 25+ interactive step types for building CLI wizards and guided workflows.
• Atmos Pro STS Tokens — Just-in-time GitHub tokens for CI access to private resources.
• Git Repository Metadata Functions — YAML functions exposing repository metadata.
• atmos.Resolve Template Function — Evaluate YAML functions inside templates.
• Install from Branches & Tags — --use-version supports Git refs directly.
• Dotenv Support in !include — Include .env files directly in YAML.
• !append YAML Function — Extend inherited lists.
• !unset YAML Function — Remove inherited configuration keys cleanly.
• ECR Public Authentication — Authenticated pulls from public.ecr.aws to avoid rate limits.
https://atmos.tools/changelog

hello, we're trying to standardize on making every component declare a standard instance and a dr instance of said component. So under stacks/catalog/network/default.yaml for example, we'd declare our VPC twice referencing components/terraform/network as the component. Now under stacks/deploy/dev.yaml we import our stacks/catalog/network and in the dev.yaml, we'd like to be able to toggle that DR instance on or off. The idea is that every environment has to opt out of provisioning a DR environment. What approach would you recommend to address this? We've had trouble getting a variable passed over to the metadata.enabled because anything we pass gets converted to a string and enabled needs a bool

Hi team, I recently started using the required_providers, required_version and providers atmos fields and wanted to share some findings and get a sense of whether I'm on the right track.

In 1.221.0 you can now define your own custom components. https://atmos.tools/changelog/custom-component-types

In atmos+pro, you can now mint short-lived narrowly scoped GitHub tokens that span repos. Think of it like AWS STS but for GitHub.

https://atmos-pro.com/changelog/2026-05-29-cross-repo-sts

This makes it trivial to read stack configs, source components, or really anything you do that needs a short-lived cross-repo token.

I found some interesting design differences between my own GCP components vs. Cloud Posse AWS components. I designed my GCP components to always take explicit var.project_id , this concept of account_map is quite confusing. 😃

Especially when tofu allows to use variables within provider blocks nowadays.