NewYour coding agent can read the release notes before it upgrades.Set up the MCP server →
Go modules, #1416 by repository stars
Latest stable
v1.234.0
Last release 13 days ago
09 Nov 2026
Ships on a steady schedule
a new release about every 8 days
Some releases are documented
notes for 32 of the last 60 stable releases
Nothing withdrawn
no release was ever pulled
5 years old
2417 releases · first in 2021
Drag across the ruler to compare, or tap a release to open it.
Nothing published for this version
Nothing published for this version
feat(auth): add Azure AKS/ACR integrations mirroring EKS/ECR @osterman
<details> <summary>feat(auth): add Azure AKS/ACR integrations mirroring EKS/ECR @osterman (#2790)</summary>
atmos azure aks token, atmos azure aks update-kubeconfig, and atmos azure acr login, mirroring the existing atmos aws eks/atmos aws ecr integrations.pkg/auth/cloud/kube.KubeconfigManager from AWS-specific to a cloud-agnostic writer shared by EKS and AKS, with a regression suite locking in byte-identical AWS output.IntegrationSpec.Cluster/.Registry schema structs (renamed from EKSCluster/ECRRegistry to Cluster/Registry) so spec.cluster/spec.registry are reused verbatim across aws/eks+azure/aks and aws/ecr+azure/acr — no new per-cloud config keys.website/docs/cli/commands/azure/), a changelog post, a roadmap update, two new agent skills (atmos-azure-aks, atmos-azure-acr), and a PRD documenting the design (docs/prd/azure-aks-acr-integrations.md).google.golang.org/grpc (xDS RBAC auth bypass / HTTP2 rapid-reset bypass), and three transitive website npm packages (fast-uri, svgo, dompurify).kubectl and Docker credentials in one step via atmos auth login. Azure had the same auth foundation (providers, identities) but no equivalent for AKS/ACR, so Azure users still needed the az CLI — and for AAD-enabled clusters, the separate kubelogin binary — outside of Atmos entirely.atmos azure aks token instead of kubelogin; ACR login is a plain OAuth2 token exchange, matching what az acr login does under the hood.docs/prd/azure-aks-acr-integrations.mddocs/prd/eks-kubeconfig.md), ECR authentication PRD (docs/prd/ecr-authentication.md)Exercised end-to-end against a live AAD-enabled AKS cluster (Azure CNI Overlay + Cilium, AAD + Azure RBAC, local accounts disabled) — the live path that PRD Success Metric #2 had previously left to unit tests only. This surfaced, and fixed, a registration gap.
Bug found + fixed. atmos azure aks update-kubeconfig --integration <name> failed with unknown integration kind: azure/aks. The pkg/auth/integrations/azure package self-registers azure/aks and azure/acr in its init(), but nothing blank-imported that package in pkg/auth/manager.go (unlike the aws and github integration packages), so init() never ran and the kinds never registered. The unit suites import the azure package directly, which registered the kinds incidentally and masked the missing production import. Fixed by adding the blank import alongside aws/github.
Integration mode — describe the cluster and write kubeconfig via the Go SDK (no az, no kubelogin):
$ atmos azure aks update-kubeconfig --integration dev/aks
✓ AKS kubeconfig: dev-aks → ~/.config/atmos/kube/config
$ export KUBECONFIG=~/.config/atmos/kube/config
$ kubectl config current-context
dev-aks
$ kubectl get pods -A
NAMESPACE NAME READY STATUS RESTARTS AGE
kube-system cilium-8trqv 3/3 Running 0 134m
kube-system coredns-5d474ff6db-pknhn 1/1 Running 0 132m
kube-system metrics-server-5b879b45fc-5nxzs 2/2 Running 0 129m
...
The kubeconfig Atmos wrote drives its exec plugin through atmos azure aks token (not kubelogin), with --server-id discovered from the cluster (here the well-known AKS AAD server app):
$ kubectl config view --raw -o jsonpath='{.users[0].user.exec.command} {.users[0].user.exec.args}'
atmos [azure aks token --cluster-name aks-dev --resource-group rg-aks-cus \
--server-id 6dae42f8-4368-4678-94ff-3960e28e3630 --subscription-id <redacted> --identity=dev]
auth exec mode — Atmos injects KUBECONFIG into the child process from the integration's Environment() (works even with auto_provision: false, which only suppresses the auto-write on login, not the env composition), so no manual export is needed:
$ atmos auth exec --identity dev -- kubectl get nodes
NAME STATUS ROLES AGE VERSION
aks-system-64934532-vmss000000 Ready <none> 139m v1.35.6
aks-system-64934532-vmss000001 Ready <none> 139m v1.35.6
Both paths mint bearer tokens through atmos azure aks token against the Atmos-managed identity — no az CLI and no kubelogin binary. (ACR login against a live registry remains unit-test-only.)
<!-- This is an auto-generated comment: release notes by coderabbit.ai -->
</details>
<details> <summary>fix(secret): inherit the component's default identity for stores @jaguer0 (#2746)</summary>
atmos secret (set/get/init/validate) now inherits the component's effective identity for store-backed secrets whose store declares no explicit identity:, instead of falling back to the default AWS credential chain (→ EC2 IMDS, which fails off-EC2).injectSecretStoreAuthResolver (cmd/secret/shared.go) called atmosConfig.Stores.SetAuthContextResolver(resolver), which passes an empty identity to every store, so an identity-less store fell back to the AWS default chain → EC2 IMDS and failed off-EC2 (e.g. no EC2 IMDS role found ... dial tcp 169.254.169.254:80: connect: host is down).cmd/terraform/utils.go, internal/exec/terraform_execute_helpers.go) already call SetAuthContextResolverWithDefaultIdentity; the secret CLI even computed the same DefaultIdentity (into SecretsAuth) but never applied it to the stores.website/docs/cli/configuration/secrets.mdx: "When omitted and the secret is resolved within a component scope, the component's effective identity is inherited."identity, and an explicit --identity, are unaffected (defaultIdentityForStore only fills empty-identity stores). atmos terraform and atmos secret list behavior is unchanged.Suggested label: patch (user-visible bug fix, no new surface; no blog/roadmap required).
docs/fixes/2026-07-13-secret-cli-inherit-default-identity.md<!-- This is an auto-generated comment: release notes by coderabbit.ai -->
atmos secret operations now inherit the component’s effective default identity when no store identity is configured.<details> <summary>fix(merge): resolve deferred YAML functions losing data on merge (#2888) @osterman (#2892)</summary>
Fixes #2888 — deferred YAML functions silently losing data on merge
ApplyDeferredMerges passed processor = nil, so deferred YAML
functions (!template, !terraform.output, !terraform.state, !store, !exec, !env) were
never actually resolved-and-merged — they silently lost data whenever a concrete value at another
config layer collided with them. On top of that, !labels/!tags/!labels.keys/!labels.values
weren't in the defer list at all, which is the literal scenario reported in the issue.internal/exec/deferred_contexts.go, plus changes across
internal/exec/stack_processor_*.go, internal/exec/yaml_processor.go,
internal/exec/yaml_func_tags.go, pkg/merge/deferred.go, pkg/merge/merge_yaml_functions.go)
that resolves deferred functions per-invocation (auth- and template-context-aware) and deep-merges
the result against any concrete override at the same path — including the mirror-precedence
direction (a concrete value at a lower-precedence layer than the function), which the original
design didn't handle.ApplyDeferredMerges
now processes deferred paths ancestor-before-descendant, so a descendant leaf can never be
clobbered by a later wholesale replace of its ancestor map (see
docs/fixes/2026-08-07-deferred-merge-nested-function-collision.md).Fixes a double-execution regression introduced by the Stage 3 pass
--process-functions=true, the
document-wide ProcessCustomYamlTags pass already resolves each surviving function, and Stage 3
then re-resolved every deferred path unconditionally — so each deferred function ran twice
per component. Harmless for pure/cached functions (!template, !terraform.output/state,
!labels, !tags, !env), but !exec (uncached — runs the shell again) and !store
(an extra backend read) were executed twice. Confirmed live: a non-colliding vars.foo: !exec
ran the shell 2× on this branch vs 1× on main.pkg/merge/merge_yaml_functions.go): for a single-contribution (no-collision) deferred path
whose value is already resolved in the result, ApplyDeferredMerges now reuses that value instead
of re-invoking the processor. Tightly guarded so genuine collisions (len > 1) still fully
resolve-and-merge — the #2888 fix is untouched. See
docs/fixes/2026-08-13-deferred-merge-double-execution.md.Housekeeping
StackComponentDeferredContexts and AllStacksDeferredContexts in place
of the raw map[string]map[string][...]ComponentDeferredContexts signatures threaded through the
stack processor (readability only; no behavior change).pkg/toolchain/registry/*), and a stack-completion fix so completion lists all project stacks
including local (cmd/emulator/completions.go).vars.tags: !labels (and other deferred functions) silently lost data when another config layer
set a conflicting value at the same path — a correctness bug with no error or warning, so it was
hard to detect in real stacks.!exec, !store) from
running twice, which could surprise users with duplicated side effects or extra load.!exec fixture and end-to-end assertions through ExecuteDescribeComponent).<!-- This is an auto-generated comment: release notes by coderabbit.ai -->
New Features
local.Bug Fixes
Tests
</details>
<details> <summary> Fix const variable interpolation in Terraform module sources @gitbluf (#2914)</summary>
Updates terraform-config-inspect to support static (const = true) variable
interpolation in Terraform module.source values.
Adds regression coverage for:
source = "./mods/${var.org}".*tfconfig.Module in the OpenTofu
interpolation test.Atmos previously failed while parsing valid Terraform 1.15+
configurations that interpolate a static variable in module.source:
variable "org" {
const = true
type = string
default = "myorg"
}
module "greeting" {
source = "./mods/${var.org}"
}
The previous terraform-config-inspect version evaluated module.source without an HCL evaluation context and returned Variables not allowed before Terraform or OpenTofu was invoked.
Fixes: #2913
<!-- This is an auto-generated comment: release notes by coderabbit.ai -->
New Features
Bug Fixes
Documentation
Tests
Your coding agent can read these notes before it upgrades. Set up the MCP server →
Nothing published for this version
Nothing published for this version
Nothing published for this version
Nothing published for this version
feat(config): auto type inference + provenance/merge bug fixes @osterman
<details> <summary>feat(config): auto type inference + provenance/merge bug fixes @osterman (#2897)</summary>
atmos config set and atmos stack set now default --type to auto: infer from the Atmos config schema, then from the type of the value already at the path, falling back to a string (with a warning) only when neither source has an answer. atmos stack set previously never inferred at all — every value was stored as a string unless --type was passed explicitly.PickProvenanceFile (shared by atmos stack set/get/delete/list and the AI/MCP tools layer) always picking the last provenance entry, which for any value defined only in an imported catalog file was a phantom Line:0 entry pointing at the wrong (importing) manifest instead of the file that actually defines the value.MCPSettings.Enabled (a bool tagged omitempty) silently disappearing from the merged config when atmos.yaml and an atmos.d/ fragment both set it to different values, instead of the explicit value winning.<placeholder> (e.g. pass --config <file>.) being silently stripped by the terminal markdown renderer, which parses unescaped angle brackets as inline HTML.--config a.yaml,b.yaml on atmos config get/set/delete/format now warns that only the first file is targeted, instead of silently dropping the rest.atmos config get on a key defined only in an atmos.d/ fragment now hints to check atmos describe config instead of just reporting "not found".unset alias (for config/stack delete) now shows up in --help output, alongside del.atmos config and atmos stack/atmos stack config surfaced these as real, reproducible bugs and gaps — silent type corruption, a provenance-resolution bug that broke edits for the standard catalog-import stack pattern, and a config value that could vanish entirely on merge.<!-- This is an auto-generated comment: release notes by coderabbit.ai -->
config set and stack set automatically infer value types, with warnings when falling back to strings.<file> in formatted output.Nothing published for this version
…nanoid infinite-loop DoS (GHSA-2v37-7h3g-55p8 / CVE-2026-67213).
<details> <summary>Add task-runner dependencies, freshness checks, and preconditions to custom commands and workflows @osterman (#2882)</summary>
dependencies.commands/dependencies.workflows to custom commands and workflows: named, parameterized, concurrent-by-default dependency ordering across units, with automatic dedup of identical invocations.inputs/artifacts step fields: skip a step when its declared sources haven't changed since the last successful run (implicit when: checksum.changed), exposing checksum.changed/timestamp.changed/sources/artifacts as when: CEL facts.preconditions step field: skip a step when a required tool is already on PATH (implicit when: "!preconditions.success"), resolved via exec.LookPath — no shell involved. Pluralized (precondition → preconditions) to match the block-of-checks convention already used by inputs/artifacts/dependencies.continue: always step field, mirroring GitHub Actions' continue-on-error: a step's own failure is forgiven, later steps still run, overall exit status unaffected.type: parallel/type: matrix steps silently failing in custom commands (only workflows supported them before).platforms via when: CEL facts (os/arch/platform), native per-command aliases:/internal:, and a values: constraint on flags/arguments with an interactive picker.parallel/matrix, and verbatim CmdLine construction for cmd.exe /C.UnitDependencies string-shorthand schema gap.cmd/custom_command_dependency_adapter.go and cmd/custom_command_values.go into pkg/taskgraph/adapters and pkg/flags respectively, so this logic is unit-testable in isolation instead of coupled to cmd's live command registry.atmos/manifest, config/global, stacks/stack-config) accordingly.Atmos workflows and custom commands already covered most of what a task runner needs, but a handful of real gaps kept teams running go-task alongside Atmos: no dependency ordering between named commands/workflows, no up-to-date checking, no continue-on-error, no precondition shortcut, and custom commands couldn't even use parallel/matrix steps — the exact recipe the project's own go-task migration guide recommends for concurrent dependents. This closes those gaps using the existing when:/CEL condition engine and scheduler rather than inventing a second mechanism.
website/blog/2026-08-05-taskfile-convergence.mdx<!-- This is an auto-generated comment: release notes by coderabbit.ai -->
New Features
Bug Fixes
Documentation
</details>
<details> <summary>feat(provisioner): Azure (azurerm) backend auto-provisioning @aknysh (#2911)</summary>
Adds automatic provisioning for the azurerm Terraform state backend — the Azure counterpart to the existing S3 backend provisioner. When provision.backend.enabled: true on a component using an azurerm backend, Atmos creates the resource group (if missing), storage account, and blob container before terraform init, with opinionated secure defaults.
Previously, atmos terraform backend create returned create not implemented for backend type: azurerm and provision.backend.enabled silently skipped for Azure.
StorageV2 / Standard_LRS, TLS 1.2 minimum, HTTPS-only, public blob access blockeduse_azuread_auth: trueName + ManagedBy=Atmos tagsNo lock resource is created — the azurerm backend serializes concurrent writes with native Azure Blob Storage blob leases (the DynamoDB / native-S3-locking analog, built into Blob Storage).
create/delete/exists/name into the shared backend registry via init(), so the before.terraform.init hook and atmos terraform backend create/delete pick up azurerm with no wiring changes to the hook or CLI.azureBackendAPI interface hides the ARM SDK pollers behind synchronous methods for testability, mirroring the S3 client factory and the existing azurerm state-reader wrapper. A test-injectable client factory (SetAzureBackendClientFactory/ResetAzureBackendClientFactory) mirrors SetS3ClientFactory.backend block — it is not a valid azurerm backend argument and Terraform would reject it in backend.tf.json.armresources + armstorage SDK deps (azcore/azidentity/azblob were already present).azurerm backend init under Atmos-managed CLI authField-testing the provisioner end-to-end surfaced a pre-existing bug that blocked azurerm backends whenever an Atmos profile was active. For CLI / device-code / interactive auth, Atmos exported both ARM_SUBSCRIPTION_ID and ARM_TENANT_ID to the Terraform subprocess. OpenTofu's azurerm backend authenticates via the Azure CLI and runs az account get-access-token --subscription <id> --tenant <id>, which the CLI rejects with Please specify only one of subscription and tenant, not both. It fails at argument validation — before the CLI even checks the session — so terraform init cannot list existing workspaces or read state at all:
Error: Failed to get existing workspaces: error listing blobs:
AzureCLICredential: ERROR: Please specify only one of subscription and tenant, not both
The tenant is now exported only for OIDC (service-principal / federated) auth, which needs it and does not shell out to the Azure CLI. On the CLI path only ARM_SUBSCRIPTION_ID is exported; the tenant is already fixed by the MSAL session Atmos seeds (and the active subscription), and the azurerm/azapi/azuread providers auto-detect it. oidc.go re-adds the tenant in its OIDC override, so service-principal auth is unchanged.
pkg/auth/cloud/azure/env.go: PrepareEnvironment gates the tenant export on UseOIDC.pkg/auth/providers/azure/oidc.go: OIDC override re-adds ARM_TENANT_ID / AZURE_TENANT_ID.Brings Azure to feature parity with AWS for backend bootstrapping and eliminates the chicken-and-egg problem of needing remote state before Terraform can run — replacing the bespoke cold-start storage-account component teams currently hand-roll on Azure.
The CLI-auth fix is what makes the provisioned backend actually usable under Atmos-managed Azure auth: without it, terraform init against any azurerm backend fails while a profile is active.
docs/prd/azurerm-backend-provisioner.mddocs/fixes/2026-08-11-azurerm-backend-cli-subscription-tenant-conflict.mddocs/prd/s3-backend-provisioner.mdpkg/provisioner/backend/azurerm_test.go, azurerm_wrappers_test.go): table-driven, mocked azureBackendAPI and Azure SDK fake servers — config extraction/precedence, use_azuread_auth parsing, full-create, resource-group reuse, existing-account warning, location-required error, every error path, existence checks, delete safety, registry wiring, and the ARM passthrough wrappers (404→exists mapping, poller completion, error propagation). 95.1% package coverage.pkg/auth/cloud/azure/{env,setup}_test.go, pkg/auth/providers/azure/{cli,device_code,oidc}_test.go): both branches of the tenant gate verified — omitted for CLI/device-code/interactive, present for OIDC. env.go and oidc.go PrepareEnvironment are both 100% covered.go build ./..., go vet, gofmt, and all pre-commit hooks (go-fumpt, golangci-lint, go.mod tidy) pass.[!NOTE] The auto-provisioned account uses secure-but-simple defaults (
Standard_LRS, Microsoft-managed keys, public network access with Entra ID/RBAC gating) intended for dev/test/bootstrap — not production. For production, import the resources into a managed module (e.g.Azure/avm-res-storage-storageaccount); the provisioner is idempotent, soprovision.backend.enabled: truecan be left in place.
<!-- This is an auto-generated comment: release notes by coderabbit.ai -->
New Features
azurerm backend provisioning with resource groups, storage accounts, and private blob containers.Bug Fixes
Documentation
</details>
<details> <summary>fix(terraform): prevent concurrent output corruption @zack-is-cool (#2898)</summary>
Prevent concurrent Terraform runs from interleaving provisioner and lifecycle UI with component-prefixed output.
JIT provisioning, backend provisioning, post-init provider locking, and clear or spin step hooks could bypass the scheduler's concurrent-output suppression. Terminal control sequences could corrupt output from other components.
go build ./...atmos lint --changedinternal/exec suite completed in a clean worktree.tenv, a non-linked Git worktree for one sandbox test, and an environment without inherited GitLab tokens.<!-- This is an auto-generated comment: release notes by coderabbit.ai -->
<details> <summary>fix: preserve trailing newlines in text-based 3-way merges @jorrite (#2891)</summary>
atmos scaffold generate --update unconditionally stripping the trailing newline from every file it 3-way-merges, whether or not the file actually changed.pkg/generator/merge/text_merger.go: TextMerger.Merge() now appends one newlineSeparator ("\n") to each of ours/base/theirs before handing them to diff3.Merge, so diff3's guaranteed loss of exactly one trailing newline cancels out and the original count survives.pkg/generator/merge/text_merger_test.go: consolidated the trailing-newline regression coverage into a single table-driven test, TestTextMerger_TrailingNewlinePreservation, asserting exact byte-for-byte output across a no-op merge (0/1/2/3 trailing newlines, plus an internal blank line) and a genuine template change (theirs with 0/1/2 trailing newlines).ConflictStrategy handling — out of scope, and unaffected since the appended newline is identical across all three inputs.TextMerger.Merge() delegates the actual 3-way merge to epiclabs-io/diff3, which reads each of base/ours/theirs line-by-line via bufio.Scanner (ScanLines, Go's standard-library default split function) and rejoins the merged lines with strings.Join(lines, "\n").ScanLines strips every line's terminator — including the last — and gives no way to tell afterward whether the original input ended with a trailing newline or not. Concretely: for content ending in N trailing newlines, the round-trip through GetLines + Join always reconstructs exactly N-1 (it loses exactly one, regardless of how many there were; for N = 0 there was nothing to lose in the first place). Verified directly: generating a file with 3 trailing newlines and running --update with nothing changed on the template side reproducibly comes back with 2.theirs: appending it only to theirs makes an otherwise-identical ours/theirs pair (a very common no-op shape) differ by one trailing newline as far as diff3 is concerned, which turns a no-op into a spurious detected change/conflict instead of fixing anything.<!-- This is an auto-generated comment: release notes by coderabbit.ai -->
Bug Fixes
Tests
Documentation
<details> <summary>fix(scaffold): preserve source in scaffold config @jorrite (#2869)</summary>
atmos scaffold generate/atmos init recording a dangling, already-deleted temp-directory path in spec.source of .atmos/scaffold.yaml whenever the template source is remote (git::... or a bare https://... URL).pkg/generator/source/resolver.go: in resolveRemote(), set conf.Source = src (the original source string the caller passed in) after loading the template configuration from the temporary download directory, instead of leaving Configuration.Source as whatever LoadConfigurationFromDir was given (the temp dir itself).pkg/generator/source/resolver_test.go: added/extended tests asserting Configuration.Source holds the original source for both local and remote paths, plus a dedicated regression test (TestResolve_RemoteRecordsOriginalSource) that fails on the pre-fix code and passes after.resolveLocal), which already recorded the correct value.resolveRemote() downloads the template into os.MkdirTemp("", "atmos-scaffold-"), then loaded the config with that temp dir passed in as the "source" — so Configuration.Source, and
therefore the persisted spec.source, ended up holding something like /var/folders/xx/.../atmos-scaffold-1234567890. That directory is removed by cleanup() immediately after the command finishes, so the recorded provenance is a dangling reference to nothing as soon as generation completes — useless for anything that might want to read it back later (e.g. a future --update/re-resolve flow), and directly contradicts SaveProjectRecord's own doc comment: "spec.source and spec.baseRef record provenance for future updates."file://...) were correct only by accident of not having a temp-dir indirection step in resolveLocal, not because anything special-cased provenance for them.atmos scaffold generate "git::https://.../scaffold-template.git" ./out --defaults, then cat ./out/.atmos/scaffold.yaml shows a /var/folders/...//tmp/... path for spec.source, and that path no longer exists on disk.cloudposse/atmos + spec.source/scaffold, nothing matched as of 2026-08.<!-- This is an auto-generated comment: release notes by coderabbit.ai -->
Bug Fixes
Documentation
<details> <summary>fix(version): exclude draft GitHub releases from Version Tracker resolution @osterman (#2900)</summary>
pkg/github.GetReleases now excludes draft GitHub releases unconditionally, alongside the existing prerelease filter. This fixes the Version Tracker's github-releases datasource resolver, atmos version list, and GetReleaseVersions, which all share this function.atmos version track render subcommand. It was marked Deprecated/Hidden in the same commit that introduced it and has never had a non-deprecated existence in any release, so no migration path is needed. The shared renderTemplate helper moves into apply.go, its only remaining consumer.docs/prd/atmos-version-management.md, surfaced once the file was touched by this branch's --affected validation.version.files and !version function docs with worked before/after examples for the marker and github-actions file managers (including SHA pinning) and adds Helm/Container-component examples alongside the existing Terraform one.atmos version track with datasource: github-releases and desired: latest could resolve to an unpublished draft release instead of the actual latest published release, whenever the GitHub token had repo write access (e.g. secrets.GITHUB_TOKEN in a repo's own CI). Reproduced against cloudposse/atmos itself: atmos version track lock resolved to v1.226.0, which gh release view v1.226.0 --repo cloudposse/atmos --json isDraft confirmed was a draft, when the real latest published release was v1.225.0. Unlike prerelease, there's no legitimate case for ever resolving to a draft, so it's excluded unconditionally rather than gated behind a new opt-in policy field.atmos version track render was superseded by apply/the file-managers architecture within its own introducing PR and has been carried as dead weight across multiple releases; removing it avoids maintaining a command with no live users.!version behavior easier to learn from concrete examples rather than a single terse case.<!-- This is an auto-generated comment: release notes by coderabbit.ai -->
New Features
Bug Fixes
Removed
version track render command and associated rendering behavior.
<!-- end of auto-generated comment: release notes by coderabbit.ai -->
</details><details> <summary>fix(deps): update github.com/epiclabs-io/diff3 digest to 3b16698 @renovate[bot] (#2917)</summary> This PR contains the following updates:
| Package | Type | Update | Change |
|---|---|---|---|
| github.com/epiclabs-io/diff3 | require | digest | 280ec18 → 3b16698 |
📅 Schedule: (UTC)
🚦 Automerge: Disabled by config. Please merge this manually once you are satisfied.
♻ Rebasing: Whenever PR is behind base branch, or you tick the rebase/retry checkbox.
🔕 Ignore: Close this PR and you won't be reminded about this update again.
This PR was generated by Mend Renovate. View the repository job log. <!--renovate-debug:eyJjcmVhdGVkSW5WZXIiOiI0NC4xMi4wIiwidXBkYXRlZEluVmVyIjoiNDQuMTIuMCIsInRhcmdldEJyYW5jaCI6Im1haW4iLCJsYWJlbHMiOlsibm8tcmVsZWFzZSJdfQ==-->
</details>
<details> <summary>fix(deps): update kubernetes monorepo to v0.36.3 @renovate[bot] (#2918)</summary> This PR contains the following updates:
| Package | Change | Age | Confidence |
|---|---|---|---|
| k8s.io/apimachinery | v0.36.2 → v0.36.3 |
||
| k8s.io/client-go | v0.36.2 → v0.36.3 |
<details> <summary>kubernetes/apimachinery (k8s.io/apimachinery)</summary>
v0.36.3</details>
<details> <summary>kubernetes/client-go (k8s.io/client-go)</summary>
v0.36.3</details>
📅 Schedule: (UTC)
🚦 Automerge: Disabled by config. Please merge this manually once you are satisfied.
♻ Rebasing: Whenever PR is behind base branch, or you tick the rebase/retry checkbox.
🔕 Ignore: Close this PR and you won't be reminded about these updates again.
This PR was generated by Mend Renovate. View the repository job log. <!--renovate-debug:eyJjcmVhdGVkSW5WZXIiOiI0NC4xMi4wIiwidXBkYXRlZEluVmVyIjoiNDQuMTIuMCIsInRhcmdldEJyYW5jaCI6Im1haW4iLCJsYWJlbHMiOlsibm8tcmVsZWFzZSJdfQ==-->
</details>
Nothing published for this version
Nothing published for this version
Nothing published for this version
The security fix was picked up automatically after pushing this branch (GitHub reported the alerts against the default branch) and is bundled here per…
<details> <summary>fix(ai): close DX gaps found in atmos ai field test @osterman (#2903)</summary>
atmos ai skill update [name], a new command that compares each installed bundled skill's recorded version against the catalog embedded in the running binary and reinstalls only the ones that are actually outdated — closes the "no update command" gap the fixes below originally left deferred. See the blog post for the full story.compatibility.atmos version-compatibility gate for bundled and multi-skill Git package skill installs, not just single-skill Git clones (it was previously skipped entirely for those two paths).atmos ai skill install/uninstall/update --client values instead of silently no-op'ing, by extending pkg/flags's WithValidValues to work on string-slice flags generically (this also fixed a latent bug where WithValidValues was silently dead for every command that binds flags via BindFlagsToViper without calling the full Parse() pipeline).--path is combined with --client/--scope/--global/--all-clients on skill install, since --path skips auto-distribution and those flags are otherwise silently ignored.skill list --detailed, and flags when an installed skill has a newer catalog version available (using the same comparison update now acts on).agent-skills/skills/atmos-ai/SKILL.md doc drift (it never documented skill install at all) and removes a phantom info subcommand from atmos ai skill --help.file:// support to skill source parsing and the downloader.--scope/--global precedence on skill install/uninstall/update.atmos ai exec/ask --session actually persist and resume conversations — previously a documented flag that was a complete no-op.Model from the constructed AI client instead of an independent config lookup, fixing sessions export/import for the default zero-config claude-code provider path (previously exported checkpoints for that path could never be re-imported).--mcp server filtering for CLI providers (claude-code/codex-cli/copilot-cli/gemini-cli) too — it was silently ignored, so all configured MCP servers were always passed through regardless of the flag.--format values on ai exec instead of silently falling back to text.ai exec can now return exit code 2 for a genuine infrastructure-level tool failure (e.g. an unregistered tool) immediately, without waiting on the 25-iteration tool-call loop to exhaust as it did before.sessions clean --older-than 0d now deletes all sessions immediately, distinguished from the flag not being passed at all (which still defaults to 30 days); negative durations are now a hard parse error instead of silently falling back to the default.js-yaml (GHSA-5p4m-2wfm-xmqj, quadratic CPU consumption in !!omap resolution) and mermaid (5 advisories), via pnpm.overrides bumps within their existing major versions. No CodeQL alerts were open.TestManager_ExportSession_WarnsOnUnimportableCheckpoint CI failure: the test asserted on raw ANSI-styled ui.Warning() output, which the formatter renders as two adjacent styled runs under CI=true — same visible text, different byte layout, so the test passed locally and failed in CI. Strips ANSI before asserting on content now.atmos ai command surface — reading the real implementation, hypothesizing plausible misuse an automated test wouldn't catch, and executing for real against isolated fixtures — rather than a spec change or feature request. Most are silent DX gaps (a flag that looks like it works but doesn't, an error with no way forward, a validation check that only applies on some of the paths that need it).0d instead of doing what it says is a footgun in the other direction (a user who deliberately asks to delete everything gets nothing, silently).atmos ai skill update exists because, once asked, leaving "no update command" as a documented gap wasn't the right call — bundled skills going stale after a binary upgrade is exactly the kind of silent drift this whole PR is about fixing elsewhere.atmos ai skill update command that was initially scoped out and then built once asked for.</details>
<details> <summary>fix(config): honor --config across internal reloads and multi-file merges @osterman (#2875)</summary>
internal/exec, pkg/vendoring, cmd/, etc. calling InitCliConfig(schema.ConfigAndStacksInfo{}, false)) now fall back to parsing --config/--config-path/--base-path from os.Args/env instead of silently discarding the selection made at startup.--config file that sets a conflicting value for an array-typed key (e.g. stacks.included_paths) no longer aborts stack discovery for entries that still legitimately match; a real "nothing matched at all" case now returns a distinct error instead.atmos config get now reports the effective, fully-merged configuration for the invocation (all --config files, --config-path dirs, and profiles applied) instead of reading a single physical file.VendorDirAbsolutePath/WorkflowsDirAbsolutePath are now precomputed once (mirroring the existing top-level base_path resolution), so vendor/workflow path joins no longer re-derive a possibly still-relative BasePath.atmos --config <file> terraform plan/test was failing with failed to find import even though atmos --config <file> list stacks worked with the identical flag, because a downstream InitCliConfig re-invocation lost the --config selection mid-command.--config files with a conflicting array value made stacks.included_paths unusable for stack discovery, while atmos config get misleadingly reported the config as unchanged.Closes #2867 Closes #2868
<!-- This is an auto-generated comment: release notes by coderabbit.ai -->
New Features
atmos config get now reads the fully merged configuration, including multiple files, paths, and profiles.Bug Fixes
Documentation
<details> <summary>fix(steps): resolve relative paths against step.WorkingDirectory @osterman (#2880)</summary>
type: archive, file, workdir, junit, and container build step handlers to resolve relative source/destination/path/files/context/dockerfile fields against step.WorkingDirectory instead of the Atmos process's own cwd.BaseHandler.ResolveInWorkingDirectory helper (pkg/runner/step/handler_base.go) used by all five handlers; container build additionally anchors Dockerfile to the resolved Context, matching Docker's own convention.TestStepEngineRunsArchiveTypeWithRelativeWorkingDirectory) reproducing the original bug end-to-end.type: archive steps run as component lifecycle hooks ignored step.WorkingDirectory, even though the hooks engine (pkg/hooks/step_engine.go) already correctly computes and sets it to the resolved component path before dispatch — the field was just never read back out by the handler.file, workdir, junit, container build) with the identical bug: relative paths resolved via template substitution only, then silently anchored to process cwd instead of the step's configured working directory.<!-- This is an auto-generated comment: release notes by coderabbit.ai -->
Bug Fixes
Documentation
Tests
<details> <summary>fix(auth): cover legacy ARM audience and seed refresh token in Azure CLI cache @aknysh (#2890)</summary>
target field — the modern scope (https://management.azure.com/.default) plus the legacy audience forms (https://management.core.windows.net/.default and the double-slash variant), with matching forms for the US Government and China clouds (new LegacyManagementScopes field on CloudEnvironment).~/.azure/atmos/<realm>/msal_token_cache.json) into the Azure CLI cache after login (new CopyAtmosRefreshTokensInto; UpdateAzureCLIFiles gains a realm parameter). Skipped for service principals, empty realms, or unmatched home account IDs.pkg/auth/providers/azure/token_audience_test.go).docs/fixes/2026-08-06-azure-cli-cache-legacy-audience-refresh-token.md.atmos auth login, Terraform providers that authenticate via AzureCLICredential request an ARM token for the legacy audience https://management.core.windows.net/ (the azidentity/azapi default). The cache write-back only seeded the modern scope, so MSAL's cache lookup missed and azapi-based modules (all modern Azure Verified Modules) failed mid-apply with AzureCLICredential: ERROR: Can't find token from MSAL cache — while azurerm resources in the same apply succeeded. Observed in a real cold-start apply of a state backend component.target, and ARM accepts both audiences interchangeably, so a single entry carrying every form satisfies every lookup.az-side lookup failed the same way. Atmos authenticates with the Azure CLI's own public client ID, so the refresh token in the Atmos realm cache is directly usable by az — seeding it lets az self-mint tokens for any audience and survive access-token expiry.az login alongside atmos auth login, defeating the purpose of single-command auth.Manually verified end-to-end on a real Azure tenant:
az logout, az account clear, removed ~/.azure/msal_token_cache.json and ~/.azure/atmos/ (confirmed with az account show failing).atmos auth login alone — no az login at any point.jq '.RefreshToken | length' ~/.azure/msal_token_cache.json returned 1 (previously 0).azidentity/azapi make: az account get-access-token --resource https://management.core.windows.net/ succeeded (previously failed with Can't find token from MSAL cache). Its expiry matched the login session's, proving MSAL served it from the seeded multi-audience entry via subset matching rather than minting a new token.atmos terraform plan on an azapi-heavy component (the exact field failure): refresh and plan completed clean with no MSAL errors.azure/interactive provider, shipped in v1.225.0)<!-- This is an auto-generated comment: release notes by coderabbit.ai -->
az login prompts.<details> <summary>fix(kubernetes): close gaps found field-testing Kustomize GitOps delivery @osterman (#2905)</summary>
Field-tested the Kustomize/git-delivery GitOps pipeline shipped in #2874 (real k3s cluster, real local git remotes, real fixtures — not mocks) and fixed the gaps found:
validate: "false" (a quoted YAML string, an easy typo) was silently ignored by atmos kubernetes validate/apply/deploy, leaving validation enabled with no warning. It now fails closed with a clear error.provision.targets.<name>.split had no type enforcement in the runtime-embedded JSON Schema (pkg/datafetcher/schema/atmos/manifest/1.0.json) — only in the docs-facing copy (stacks/stack-config/1.0.json) — so split: "yes" passed atmos validate stacks and was silently dropped at runtime, falling back to path-based auto-inference. The embedded schema is now synced, and the git target's parseConfig also fails closed as defense in depth (since kubernetes validate/apply/deploy don't go through the stack-config schema path).RemoveAll that replaces whatever currently exists at the managed path.git.repositories.<name>.uri — an already-cloned repo kept using its original remote forever. reconcile now syncs the local remote URL to the configured URI before fetching.[, ], (, ) with no spaces; rendered as plain markdown it was both mangled (brackets collide with link syntax) and hard-wrapped mid-token. It's now backtick-fenced as a code span, so it renders verbatim.atmos-git skill doc: it showed a nested signing: {mode: auto} instead of the real flat signing: auto string field — copying it verbatim fails to parse.atmos kubernetes validate --server fails on a manifest set that creates its own namespace and delivers into it in the same batch (inherent to Kubernetes server-side dry-run semantics — each object's dry-run is evaluated independently against already-persisted state), even though apply/deploy of the identical objects succeeds.Also includes an unrelated, incidental fix: bumped js-yaml/mermaid pnpm overrides in website/ to close 7 open Dependabot alerts (triggered by this repo's post-push security-remediate automation).
A /field-test pass is a hands-on DX pass that builds real fixtures and runs the actual CLI against them, specifically to catch "looks fine in review, breaks or misleads a real user" gaps that unit tests (which mostly exercise fake clients and hand-built inputs) don't cover. Every fix here was independently reproduced live before being fixed, and re-verified live after. The pass also confirmed several things work correctly as documented (directory/single-file delivery, the Kustomize metadata.name exemption, validate:false + --server interaction, offline validation gating delivery) — those aren't included here since nothing needed to change for them.
<!-- This is an auto-generated comment: release notes by coderabbit.ai -->
feat(ai): browsable, searchable Agent Skills Directory @osterman
<details> <summary>feat(ai): browsable, searchable Agent Skills Directory @osterman (#2881)</summary>
/ai/skills, replacing the hand-maintained (and drifted) skill table on the Agent Skills doc page.file-browser Docusaurus plugin with category grouping, a search box, configurable card icon/CTA label, and a "Copy as Markdown" button, reused for the new skills instance.SKILL.md and every nested reference file into one clipboard-ready document, so its full context can be grabbed without installing it.atmos ai skill list gains a --format flag (table/json/yaml/csv/tsv) and a Category column.SkillCount component that renders the live, build-time skill count inline in prose (homepage AISection, docs), so counts can't drift out of date again.Atmos AI → Skills category with Agent Skills, Skill Marketplace, and a link to the new directory) and cleans up a duplicate "Native CI" sidebar/doc link.SKILL.md files surfaced by the affected-file validator./blog/agent-skills-directory/ai/skills, /ai/agent-skills<!-- This is an auto-generated comment: release notes by coderabbit.ai -->
New Features
Bug Fixes
[] instead of null.Documentation
Style
Nothing published for this version
fix(kubernetes): single-file GitOps delivery and Kustomize metadata.name exemption @osterman
<details> <summary>fix(kubernetes): single-file GitOps delivery and Kustomize metadata.name exemption @osterman (#2874)</summary>
kubernetes.gitops.provision.targets.<name> (kind: git) now supports a split tri-state: split: false writes path as a single merged multi-document YAML file instead of always treating path as a directory of auto-named files; unset infers the mode from whether path's last segment looks like a manifest filename (.yaml/.yml/.json).metadata.name on Kustomize's own Kustomization/Component objects (matched against sigs.k8s.io/kustomize/api/types's own kind/version constants), since Kustomize's own schema and field-enforcement never require one.validate: false component-level flag opts a component out of both the apply/deploy structural auto-gate and the standalone atmos kubernetes validate command.split documented on kubernetes-deploy.mdx, and the Kustomize exemption / validate: false documented on kubernetes-validate.mdx.path was always treated as a directory, so configuring path: ".../kustomization.yaml" created a directory by that name containing an auto-generated file inside it, instead of the exact file Kustomize's remote-include mechanism requires.metadata.name unconditionally, forcing users to add a meaningless name to Kustomize Component/Kustomization objects just to satisfy Atmos, even though Kustomize's own tooling never requires one.!terraform.state) and committing it to a deployment repo as a proper kustomization.yaml for Argo CD/Flux to consume.<!-- This is an auto-generated comment: release notes by coderabbit.ai -->
New Features
split, with automatic mode selection based on the destination path.Kustomization and Component objects no longer require metadata.name.validate setting to bypass offline structural validation.Bug Fixes
Documentation
<details> <summary>fix(ci): Docker build image mirrors and concurrent output race @osterman (#2884)</summary>
cloudposse/github-action-docker-build-push from v3.0.0 to v3.1.0 in the release docker job (.github/workflows/build.yml).binfmt-image input to mirror.gcr.io/tonistiigi/binfmt:qemu-v7.0.0.LinePrefixWriter (pkg/io/line_prefix_writer.go), so concurrent Terraform node writers can't interleave a line mid-block.moby/buildkit) and QEMU binfmt images from public.ecr.aws.mirror.gcr.io/moby/buildkit), fixing that pull. The action's binfmt-image input still defaults to public.ecr.aws/eks-distro-build-tooling/binfmt-misc even at v3.1.0 (no upstream fix yet), so it's overridden here directly to the equivalent Google-mirrored tonistiigi/binfmt image, which publishes the same qemu-v7.0.0 tag. Verified live: both mirror.gcr.io/tonistiigi/binfmt:qemu-v7.0.0 and docker.io/tonistiigi/binfmt:qemu-v7.0.0 resolve to the same digest and pull successfully; mirror.gcr.io also falls through to Docker Hub origin on any cache miss, so it's never less reliable than a direct Docker Hub pull.TestExecuteTerraformConcurrentHooksUseNodeWriters (pkg/scheduler/adapters) with a real, reproducible race: LinePrefixWriter.writeLine acquired/released the shared output mutex per line, so a single Write() call that produced multiple lines (e.g. a hook's buffered \r-then-\n progress update) could have a different node's writer interleave a line in between, corrupting concurrent Terraform output. Reproduced with go test -race -count=200 before the fix (intermittent failures) and confirmed 200/200 clean after.</details>
Nothing published for this version
feat(toolchain): support Aqua github_archive package type @osterman
<details>
<summary>feat(toolchain): support Aqua github_archive package type @osterman (#2416)</summary>
github_archive package type to the Aqua-compatible toolchain registry parser and installer.https://github.com/{owner}/{repo}/archive/refs/tags/{version}.tar.gz, matching upstream aquaproj/aqua semantics.tar.gz (mirroring aqua's GetFormat()); asset, url, format, and format_overrides are intentionally ignored for this type.resetByPkgType so version overrides that switch a tool to github_archive clear stale asset/url.adr-tools example; cover validation, version-prefix handling, format-ignored semantics, URL pattern, and {{trimV .Version}}/path template idiom.github_archive for tools shipped as repository source archives (e.g., adr-tools, tfenv, tgswitch, and many single-script projects). Without this type, those entries failed with unsupported tool type: github_archive.<!-- This is an auto-generated comment: release notes by coderabbit.ai -->
New Features
github_archive packages, downloading GitHub tag archives.github_content packages, downloading individual repository files.Bug Fixes
Documentation
Nothing published for this version
Nothing published for this version
…vulnerability (GHSA-7p8r-x3mc-p8w7 / CVE-2026-18446), flagged by Dependabot after this branch was pushed.
<details> <summary>feat(auth): add azure/interactive browser authentication provider @aknysh (#2862)</summary>
azure/interactive provider kind: MSAL interactive browser authentication — authorization code + PKCE on a localhost redirect, the same flow az login uses (AcquireTokenInteractive). One command (atmos auth login) opens the browser, completes SSO, and mints Management/Graph/Key Vault tokens.interactiveProvider embeds deviceCodeProvider, reusing the MSAL client, silent token acquisition (refresh tokens make repeat logins silent — no browser), token fan-out, and the Azure CLI cache write-back with the correct guest home account ID from #2861. Spec shape is identical to azure/device-code.auth_method: interactive, and the MSAL cache account_source mirrors az's own labels (authorization_code for the browser flow, device_code otherwise).azure-interactive-auth), shipped roadmap milestone, PRD (docs/prd/azure-interactive-auth.md), and the fix doc for #2861 (docs/fixes/2026-08-03-azure-cli-cache-corruption-guest-users.md).acquireInteractive, checkInteractive) per the repo's DI convention, since the real interactive flow needs a live IdP and a browser; tests cover the success path, acquisition failure, and headless refusal against a sandboxed HOME.azure/device-code increasingly fails; azure/cli requires a pre-existing az login session (two commands); azure/oidc is CI-only. Azure users had no one-command human login equivalent to aws/iam-identity-center.azure/interactive"Interactive" is Microsoft's own term for this flow, not our invention:
AcquireTokenInteractive, and Microsoft's flow taxonomy divides authentication into "interactive and non-interactive flows" — where "interactive" specifically means the browser-based authorization-code sign-in a user completes, as opposed to device code, silent, or client-credential flows.aws/iam-identity-center opens a browser too, but the kind names the mechanism; aws/saml treats the browser as a driver: option; gcp/workload-identity-federation names the federation mechanism. azure/interactive follows the same rule using the platform's own vocabulary.azure/browser (self-explanatory but bakes UX into the kind name, contrary to the convention above) and a spec.flow: browser option on azure/device-code (avoids a new kind, but makes the kind name actively misleading when the flow isn't device code).Tested end to end in a real Entra tenant where the device code flow is blocked by a Microsoft-managed Conditional Access policy, with an operator who is a guest (B2B) user in that tenant. Starting from a fully wiped ~/.azure:
atmos auth login opened the default browser, SSO completed (no device code, no az login), and tokens were minted.atmos auth login again succeeded without opening the browser (same token expiry), confirming refresh-token persistence.atmos auth whoami — reported provider, identity, subscription principal, tenant, and expiry.az account show worked even though az login was never run, thanks to the Azure CLI-compatible cache write-back.account_source: authorization_code (matching what az itself records), carrying the true home account ID. This is the guest-user case that used to corrupt the az cache before #2861.Details in the PRD's Verification section (docs/prd/azure-interactive-auth.md).
AuthMethod/HomeAccountID plumbing)<!-- This is an auto-generated comment: release notes by coderabbit.ai -->
New Features
azure/interactive provider and browser sign-in prompts.Bug Fixes
Documentation
</details>
<details> <summary>refactor: consolidate file locking @osterman (#2856)</summary>
pkg/cache.FileLock, covering cloud configuration, Helm repositories, and workdir metadata.gofrs/flock acquisition and release logic.<!-- This is an auto-generated comment: release notes by coderabbit.ai -->
Bug Fixes
New Features
<details> <summary>docs: add succinct Gomplate datasource example @osterman (#2866)</summary>
file:// datasource and reference its values from a component's vars.<!-- This is an auto-generated comment: release notes by coderabbit.ai -->
atmos describe and viewing the resulting YAML output.!include functions.
<!-- end of auto-generated comment: release notes by coderabbit.ai -->
</details><details> <summary>feat(store): add atmos store CRUD CLI and type: store workflow step @osterman (#2858)</summary>
atmos store — a new CLI command group (set/get/delete/list) for raw CRUD access to any store backend configured under stores: in atmos.yaml (AWS SSM, AWS Secrets Manager, HashiCorp Vault, Azure Key Vault, GCP Secret Manager, Redis, Artifactory, 1Password, Keychain, GitHub Actions). Unlike atmos secret, no declaration is required — any key can be read, written, or deleted directly by name, optionally scoped to a stack and component.type: store workflow step that writes a value from a workflow, custom command, or hook — usable automatically as a hook too via the existing generic kind: step bridge, with no extra wiring.!store/!store.get YAML functions, so pipeline metadata (an image tag from a build step, a build number, a deployment marker) can be handed off to a completely different stack or component.pkg/store.Service facade, the cmd/store command family, and the store step handler, plus CLI/website docs, a changelog post, and a roadmap milestone.!store/!store.get); the only existing write path was a single Terraform-output-specific hook (kind: store), so anything else needing to be written into a store meant scripting around Atmos with a cloud CLI.<!-- This is an auto-generated comment: release notes by coderabbit.ai -->
New Features
atmos store commands to set, get, list, and delete values across configured backends.type: store workflow steps for writing templated values.Documentation
Tests
<details> <summary>Add tfmigrate support for Terraform components @osterman (#2534)</summary>
atmos terraform migrate plan, apply, and list for running user-authored tfmigrate migrations in Terraform component context.kind: tfmigrate lifecycle hooks with dynamic/static modes, toolchain resolution, Terraform/OpenTofu exec path wiring, and same-identity auth handling.tfmigrate history storage.tfmigrate history storage, so Atmos now documents and exposes the values users need without taking on history persistence in v1.<!-- This is an auto-generated comment: release notes by coderabbit.ai -->
atmos terraform migrate plan, apply, and list commands.tfmigrate lifecycle hooks with dynamic and explicit execution modes.<details> <summary>feat(vendor): native component updater PR workflow @osterman (#2756)</summary> <a href="https://atmos.tools/ci"><picture> <source media="(prefers-color-scheme: dark)" srcset="https://atmos.tools/img/atmos-ci-gradient.svg"> <source media="(prefers-color-scheme: light)" srcset="https://atmos.tools/img/atmos-ci-gradient-on-light.svg"> <img src="https://atmos.tools/img/atmos-ci-gradient-on-light.svg" alt="Atmos CI" height="32" align="right"> </picture></a>
Manually tested end-to-end as a real user — following --help/docs, in isolated sandboxes, and against a real repository (cloudposse/infra-live):
vendor.ci.pull_request config surface (title/body templates, labels, draft, reviewers, assignees) — #1701, #1702 (left as drafts, not merged).--pull-request created a pull request but never printed its URL in the default table output.atmos.yaml's vendor.update.*/vendor.ci.* config (groups, execution mode, PR title/labels/draft/reviewers) was silently ignored — read from the wrong viper instance instead of the parsed config.--all double-counted every component when a repo vendors exclusively via component.yaml and leaves an unused component type (e.g. packer) unconfigured.oci-artifacts coverage entry always claimed "complete" regardless of whether any OCI artifact existed in the project.atmos vendor verify/clean/--refresh-lock/--lock-enforcement (all three modes) and atmos sbom generate (CycloneDX vs. SPDX, NTIA mode, experimental gating, upload-outside-CI) in isolated sandboxes.ATMOS_PRO_GITHUB_TOKEN (the token github/sts mints) to the Component Updater's GitHub token precedence, so atmos auth exec --identity <github-sts-identity> -- atmos vendor update --pull-request gets a token that triggers downstream Actions workflows on the PR it opens — unlike the default GITHUB_TOKEN, which GitHub excludes from re-triggering workflows.<!-- This is an auto-generated comment: release notes by coderabbit.ai -->
atmos sbom generate with provenance/NTIA modes, SPDX/CycloneDX output, and optional CI artifact upload.atmos vendor update --pull-request workflows with component/group selection and deterministic PR publishing.atmos vendor verify, atmos vendor clean, and vendor.lock.yaml drift protection.</details>
<details> <summary>feat(workflow): support tags and labels selectors @zack-is-cool (#2857)</summary>
--tags and --labels on atmos workflow, forwarding them to nested type: atmos steps alongside optional --stack.go test ./internal/exec -run TestExecuteWorkflow_ForwardsCommandLineFilters -count=1go test ./pkg/workflow -run 'TestAppendAtmosStepFlags|TestControlCommandExecutorExecuteAtmos' -count=1go test ./cmd/workflow -run TestWorkflowSelectorFlags -count=1git diff --check<!-- This is an auto-generated comment: release notes by coderabbit.ai -->
New Features
--tags and --labels selectors to workflow commands.--stack while preserving workflow ordering and execution behavior.Documentation
<details> <summary>feat: add date-anchored default editions @osterman (#2762)</summary>
edition: defaults, the --edition override, and atmos describe/list edition commands.<!-- This is an auto-generated comment: release notes by coderabbit.ai -->
New Features
--edition, ATMOS_EDITION, or atmos.yaml.describe edition and list editions commands for inspecting default changes.--use-mocks.Bug Fixes
Documentation
<details> <summary>refactor(store): move backends into pkg/store/providers subpackage @osterman (#2575)</summary>
pkg/store into a new pkg/store/providers subpackage, keeping pkg/store as a pure interface/type boundary (the Store interfaces, StoreConfig/StoresConfig/StoreRegistry types, auth-config types, error sentinels, and generated mocks).pkg/store owns StoreRegistry, NewStoreRegistry, and a Register(type, factory) API; each backend registers its factory from an init(), so the type switch is replaced by a map lookup. pkg/config blank-imports pkg/store/providers so the built-in backends register at startup (database/sql driver pattern).pkg/store/providers exclusion to the provider-agnostic-auth depguard rule so its cloud-SDK imports are permitted (matching pkg/auth/providers).pkg/auth/providers / pkg/secrets/providers layout and making pkg/store a clean type/interface package.providers but was named after the StoreRegistry type it returned; pkg/store now owns both the registry type and its construction, and adding a backend is one self-contained file that registers itself.ErrStoreTypeNotFound, identity warnings are preserved. Builds clean, all affected tests pass, and golangci-lint --new-from-rev=origin/main reports zero issues.<!-- This is an auto-generated comment: release notes by coderabbit.ai -->
<details> <summary>feat(toolchain): add --format=plain/json to atmos toolchain get @osterman (#2845)</summary>
--format flag to atmos toolchain get with three modes: table (default, unchanged), plain (bare version string only), and json (structured output with tool/version/installed fields, or a full version list under --all).plain/json output through the data channel (stdout, pipeable) instead of the styled UI channel (stderr), via new printVersionsPlain/printVersionsJSON helpers in pkg/toolchain/get.go.--format=plain combined with --all with a new ErrToolchainPlainFormatWithAllFlag sentinel, since there's no single version to print in that case.2>&1 since it's written to stderr), e.g. atmos toolchain get vale-cli/vale 2>&1 | grep -oE '[0-9]+\.[0-9]+\.[0-9]+' | head -1.--format=plain collapses that to version=$(atmos toolchain get vale-cli/vale --format=plain), and --format=json gives scripts structured access to installed status without adding a new dependency.<!-- This is an auto-generated comment: release notes by coderabbit.ai -->
atmos toolchain get.--format/-f, environment-variable support, and shell completion.--all and plain output options.<details> <summary>feat: dependency-closure selection (--include-dependencies/--include-dependents) + scope --labels/--tags evaluation before filtering @osterman (#2807)</summary>
--all, --components, --query, -s, --tags, --labels, --affected) accepts two new depth-carrying flags on plan/apply/deploy/destroy/init:
--include-dependencies[=N] — also process everything the selection depends on (its prerequisites), N levels deep (bare flag = unlimited).--include-dependents[=N] — the reverse direction. Previously a bool wired only into --affected; now works with every selection and accepts a depth (true/false remain accepted for compatibility).destroy --include-dependencies warns that it also destroys shared prerequisites.atmos list components/stacks/instances preview the exact execution set a bulk run would touch.pkg/dependency.Filter gains per-direction depth bounds with a best-depth BFS; the scheduler adapter is restructured to seed-then-expand (tags/labels/query moved from post-filter to seed narrowing, per-node query skip suppressed when the seed already applied it).list dependencies is generalized into dependencies.ResolveScopedClosure and shared with the terraform bulk paths, so closure runs fully evaluate only the stacks the closure touches.metadata.tags/metadata.labels are selectors evaluated before auth/templating/YAML functions, so their values must be resolvable without authentication or process execution. Values using !terraform.state, !terraform.output, !store, !store.get, !secret, !aws.*, !emulator, !exec, !random, or template calls to atmos.Component/atmos.Store/atmos.GomplateDatasource/datasources are rejected with a by-design error and migration hint on any command that processes the manifest. Plain strings, simple templates, and local functions (!env, !git.*, !include) remain allowed.templates.settings.delimiters) via Go's template parser — no regex heuristics — and tags.SelectorUnresolved is now delimiter-aware too.--labels/--tags (original scope)--tags/--labels skips auth/template/YAML-function evaluation entirely, generalizing the existing -s/--stack early-skip. Now also threaded into list stacks/components/instances (previously they row-filtered after full evaluation).atmos list dependencies computes a lightweight dependency graph, derives the reachable closure, and only fully evaluates the stacks that closure touches — re-converging against the resolved graph so templated same-stack dependency targets still resolve.describe.settings.eager_evaluation remains the rollback switch forcing the old full-evaluation behavior.pkg/tags (purity + delimiter cases), pkg/dependency (depth/cycles/diamonds), pkg/scheduler/adapters (closure retains non-matching prereqs, destroy ordering, query suppression, depth merging), pkg/list/dependencies (scoped-evaluation convergence with a poisoned unrelated stack), and the list/cmd layers; regenerated help-text golden snapshots.website/blog/2026-07-27-include-dependencies-closure.mdx), and a roadmap entry.atmos terraform ... --all --labels=.../--tags=... and atmos list dependencies --stack <stack> fully evaluated every stack in the repo (templates, YAML functions, auth/backend) before consulting the selector, so an unrelated account's unreachable backend failed the command.docs/fixes/2026-06-22-describe-stacks-scope-and-cache-per-component-auth.md.docs/fixes/2026-07-25-scope-before-evaluate-labels-tags-list-dependencies.md for root-cause and verification detail of the scope-before-evaluate work (its Recommendations section is implemented by this PR's closure flags).<!-- This is an auto-generated comment: release notes by coderabbit.ai -->
New Features
levels output for shortest dependency distances.key=value and key:value formats.Improvements
<details> <summary>test(config): regression test for atmos.d commands merge (#2570) @osterman (#2840)</summary>
TestMergeConfig_AtmosDCommandsMerging_TopLevelYamlFunction to pkg/config/config_import_test.go, reproducing the exact minimal repro from #2570: two .atmos.d/ files where the last-processed file has a top-level Atmos YAML function (!repo-root) directly on a command's own field.0b1182c6bb (PR #2677, released in v1.223.0), which strips the commands key from file content before preprocessAtmosYamlFunc runs so it can no longer overwrite the already-merged commands array.</details>
<details> <summary>fix(test): stop Windows toolchain-vanishing flake in acceptance tests @osterman (#2834)</summary>
TestPrintTelemetryDisclosureOnlyOnce (and its three sibling telemetry disclosure
tests) from deleting the shared <cache>/atmos root; they now isolate via a per-test
ATMOS_XDG_CACHE_HOME/XDG_CACHE_HOME redirect instead.internal/exec terraform/tofu version tests: RequireTerraformPath/
RequireTofuPath now resolve the binary path once and hand it back directly, instead of a
second independent exec.LookPath call at each call site.requireExecutablePath fail loudly (t.Fatalf) instead of silently returning an empty
path when ATMOS_TEST_SKIP_PRECONDITION_CHECKS=true and the binary isn't found.pkg/toolchain install/uninstall/list tests from the real shared XDG
toolchain cache directory (Toolchain.InstallPath now points at each test's own t.TempDir()).requireExecutablePath's retry window (2s → 15s) as a defensive backstop, and adds
permanent forensic instrumentation (executableLookupForensics) that dumps per-PATH-entry
toolchain directory state on a lookup failure.docs/fixes/ entry recording the root cause and investigation.executable file not found in %PATH% failure for tofu/terraform, always on a different
victim test in internal/exec each run.TestPrintTelemetryDisclosureOnlyOnce
was calling os.RemoveAll on the shared <cache>/atmos root — the same root the toolchain
install path lives under since #2579 — twice per run (setup + defer), deleting every
CI-provisioned tool (terraform, opentofu, helm, helmfile) out from under the other
concurrently-running go test package binaries.</details>
<details> <summary>ci: run required checks for GitHub merge queue @osterman (#2813)</summary>
<!-- This is an auto-generated comment: release notes by coderabbit.ai -->
<!-- end of auto-generated comment: release notes by coderabbit.ai --> </details>
<details> <summary>Support parent-scoped multi-file stacks @osterman (#2787)</summary>
metadata.inherits graph self-contained while canonicalizing equivalent imported duplicates by lexical parent path.docs/prd/top-level-stack-composition.md<!-- This is an auto-generated comment: release notes by coderabbit.ai -->
New Features
Bug Fixes
Documentation
<details> <summary>feat(stacks): support global-scope metadata defaults @osterman (#2808)</summary>
metadata fields (labels, tags, custom, enabled, locked, terraform_workspace_pattern) at the stack-manifest root, deep-merged into every component's own metadata as a stack-wide default.metadata.inherits base-component chain → the component's own local metadata: block, which always wins.component, inherits, type, name, terraform_workspace) remain component-only; setting one of these at global scope is now a hard validation error, both at runtime and via the manifest JSON Schema, instead of a silent no-op.metadata.mdx docs and adds a changelog post explaining the new global scope.metadata: block (e.g. in _defaults.yaml) was previously accepted by the schema but never applied — metadata.labels/metadata.tags/etc. set there silently did nothing, which is worse than an error, since users had no signal their config wasn't taking effect.metadata block into each component definition instead of declaring it once.<!-- This is an auto-generated comment: release notes by coderabbit.ai -->
New Features
metadata: defaults that are deep-merged into each component’s metadata.metadata.inherits chain (when enabled) → component-local metadata.Validation & Tests
metadata (only labels, tags, custom, enabled, locked, terraform_workspace_pattern).Documentation
<details> <summary>Clarify Atmos CI concurrency guidance @osterman (#2798)</summary>
<!-- This is an auto-generated comment: release notes by coderabbit.ai -->
concurrency behavior for Atmos/Terraform runs, including in-progress vs pending handling and pending-run eviction.cancel-in-progress: true can cancel an in-flight Terraform apply and potentially leave remote state locks requiring manual recovery.queue: max limits (up to 100 pending) and noted it can’t be combined with cancel-in-progress.<details> <summary>fix(output): render concurrent carriage-return updates safely @zack-is-cool (#2860)</summary>
go test ./pkg/scheduler/adapters ./pkg/io -count=1go test ./pkg/terraform/output -run '^TestSuppressSpinnersRestoresNestedScopes$' -count=1pre-commit run --files pkg/scheduler/adapters/terraform.go pkg/terraform/output/executor_utils.go pkg/terraform/output/spinner.go pkg/terraform/output/spinner_test.go<!-- This is an auto-generated comment: release notes by coderabbit.ai -->
Bug Fixes
Improvements
<details> <summary>fix(config): resolve git-root base_path for --config/--config-path @osterman (#2864)</summary>
pkg/config/load_config_args.go so that loading configuration via --config/--config-path also applies git-root discovery for an empty (or .) base_path, matching the plain auto-discovery flow in LoadConfig().TestLoadConfigFromCLIArgs_AppliesGitRootBasePath) that reproduces the bug and verifies base_path now resolves correctly.fast-uri transitive dependency (website) from 3.1.4 to 3.1.5 to remediate a high-severity host-confusion vulnerability (GHSA-7p8r-x3mc-p8w7 / CVE-2026-18446), flagged by Dependabot after this branch was pushed.loadConfigFromCLIArgs() never called applyGitRootBasePath(), unlike the main LoadConfig() auto-discovery path. As a result, a project with base_path: '' in atmos.yaml resolved correctly via plain auto-discovery but left base_path empty when the identical config was loaded via --config, breaking component/stack path resolution (e.g. atmos terraform test) with Error: failed to find import.fast-uri bump addresses a live Dependabot alert (a patch-level version bump, not blocked by .github/dependabot.yml's major-version ignore policy) surfaced automatically after pushing this branch.<!-- This is an auto-generated comment: release notes by coderabbit.ai -->
Bug Fixes
"." base path to the Git repository root.Tests
<details> <summary>fix(auth): prevent Azure CLI cache corruption for guest users @aknysh (#2861)</summary>
azure/cli provider — az's own cache is authoritative, and writing back what came from az is what corrupted it.AzureCredentials (cli / device_code / oidc) so the write-back can be gated per provider kind.azure/device-code provider (silent and interactive flows) and use it in both Azure CLI cache writers (UpdateAzureCLIFiles and the provider-level updateAzureCLICache), falling back to the previous {oid}.{tenant} derivation when unavailable.azure/subscription identity's field-by-field credential copy with a struct copy plus explicit overrides, and add a reflection-based regression test that fails if any future AzureCredentials field is dropped by the wrap.TestNewMSALCache's default-path case from the real ~/.azure.atmos auth login, the Azure CLI cache write-back created an MSAL Account entry with home_account_id derived as {oid}.{target-tenant} and hardcoded account_source: "device_code". For guest (B2B) users the home tenant differs from the target tenant, so az ended up with two Account entries for the same username and failed every subsequent command with Found multiple accounts with the same username (azure-cli#20168) — including az account get-access-token, which the azure/cli provider itself shells out to. In other words, one atmos auth login broke both az and the next atmos login for any guest user.az login → atmos auth login → az broken; after the fix, az stays healthy, the cache keeps exactly one Account entry, and the persisted credentials carry auth_method so the gate holds across credential caching.<!-- This is an auto-generated comment: release notes by coderabbit.ai -->
<!-- end of auto-generated comment: release notes by coderabbit.ai --> </details>
<details> <summary>fix(scaffold): resolve relative write-target directories consistently @osterman (#2855)</summary>
atmos scaffold generate so a relative target directory (e.g. the CLI's own default ./my-project) works, instead of rejecting every file with path traversal not allowed.validateWriteTarget in pkg/generator/engine/templating.go now resolves the write directory (realDir) through the same ResolveAndCleanBasePath helper already used for the target base (realBase), instead of a bare filepath.EvalSymlinks that stays relative for relative inputs.TestProcessFile_RelativeTargetPath, covering a relative targetPath end-to-end (previous tests only exercised absolute t.TempDir() targets, so this case was never caught).realBase was always absolutized before comparison, but realDir was resolved with a bare filepath.EvalSymlinks, which returns a relative path unchanged when given a relative input. Comparing an absolute path against a relative one never matched the containment check, so it fired as a false-positive path traversal on every write whenever the target directory was relative — including the command's own default target.filepath.Dir(fullPath) was already absolute in that case, masking the bug.<!-- This is an auto-generated comment: release notes by coderabbit.ai -->
Bug Fixes
./my-project.Tests
<!-- end of auto-generated comment: release notes by coderabbit.ai --> </details>
<details> <summary>fix(config): stop recommending deprecated stacks.name_pattern @osterman (#2842)</summary>
atmos aws eks update-kubeconfig and Spacelift stack-name generation now check stacks.name_template before falling back to the deprecated stacks.name_pattern, instead of only supporting the deprecated field.pkg/config, errors/errors.go, and pkg/helmfile/cluster.go that previously only pointed users at the deprecated fields now recommend name_template/cluster_name_template.aws eks update-kubeconfig command help/docs no longer teach the deprecated name_pattern/cluster_name_pattern fields.examples/, demo/, and non-backward-compat-test tests/fixtures scenarios from name_pattern to name_template (a handful of fixtures that specifically test the deprecated field, precedence, or backward compatibility were left untouched on purpose).name_template support and continued name_pattern backward compatibility for the two code paths that changed.stacks settings are not consistently applied") showed the profile-merge pipeline was already correct and consistent between describe config and list dependencies.stacks.name_pattern — deprecated in favor of stacks.name_template — was still treated as the primary/only stack-naming mechanism in a couple of code paths and in several error messages, which is what produced the reported inconsistency and general user confusion about which field to use.name_pattern continues to work unchanged for backward compatibility; only the recommended path and documentation change.<!-- This is an auto-generated comment: release notes by coderabbit.ai -->
<details> <summary>fix(toolchain): accept semver release-candidate strings in --use-version @osterman (#2841)</summary>
isValidSemver() in pkg/toolchain/version_spec.go so it accepts semver pre-release and
build-metadata suffixes (e.g. 1.225.0-rc.3, 1.2.3+build.5), by delegating to the
already-vendored Masterminds/semver/v3 library instead of a hand-rolled digit-only check.pkg/toolchain/version_spec_test.go (unit-level
isValidSemver/ParseVersionSpec cases) and pkg/version/reexec_test.go (end-to-end at the
--use-version entry point).docs/fixes/2026-07-31-use-version-release-candidate-semver.md.atmos --use-version=1.225.0-rc.3 failed with invalid version output format, even though
1.225.0-rc.3 is a spec-compliant semver string. The version parser split on . and required
every part to be pure digits, which rejects any release-candidate/pre-release version.<!-- This is an auto-generated comment: release notes by coderabbit.ai -->
New Features
--use-version now accepts release-candidate versions such as 1.225.0-rc.3, including versions with build metadata.latest.Bug Fixes
Documentation
<details> <summary>fix(dag): stop concurrent map crash in bulk terraform commands @osterman (#2831)</summary>
fatal error: concurrent map iteration and map write crash in DAG-scheduled bulk terraform commands (terraform <cmd> --all/--affected/--query) at higher --max-concurrency.ProcessComponentConfig now shallow-clones the component section before any downstream code mutates it, so concurrent workers never write into the map tree owned by the shared FindStacksMap cache.imports, and terraform_workspace_pattern/terraform_workspace_template, from cache-owned maps in place).internal/exec/process_stacks_shared_cache_test.go) that fail pre-fix both deterministically and under -race.brace-expansion pnpm.overrides (website) to 1.1.18/2.1.4, patching CVE-2026-14257 / GHSA-mh99-v99m-4gvg (high-severity DoS via unbounded expansion length), reported by Dependabot alert #261.FindStacksMap caches processed stack config and returns it by reference on cache hits, shared across all goroutines within a process. ProcessStacks and mergeGlobalAuthConfig write top-level keys into that shared component section, while findComponentInStacks has every DAG worker iterate every stack's component section (not just its own) looking for a match — so one worker's write races with another worker's read/iteration of the same cached map, crashing exactly as reported.ProcessStacks call in the same process even outside the crash path.brace-expansion bump addresses an open, high-severity Dependabot alert; deferring it risks a DoS crash if attacker-influenced input reaches an affected glob/brace-pattern code path in the docs site tooling.<!-- This is an auto-generated comment: release notes by coderabbit.ai -->
Bug Fixes
describe output to include referenced imports consistently.Documentation
<details> <summary>Support explicit CI git checkout and bundle Docker CLI @osterman (#2812)</summary>
--ci and ATMOS_CI controls for no-argument CI checkout, including explicit opt-out behavior.cmd/root.go for CI git-clone bootstrap detection with Cobra-identity + the existing cmd/git flag-handler infrastructure (resolveCICloneMode), removing ~140 lines of bespoke parsing.atmos git clone's no-arg CI checkout to resolve the branch from the CI provider's parsed short name instead of the raw ref, which previously failed real branch/PR checkouts.actions/checkout (docs/prd/git-ops.md) never actually worked for a real branch or PR.<!-- This is an auto-generated comment: release notes by coderabbit.ai -->
New Features
--ci support for atmos git clone, configurable with ATMOS_CI.Bug Fixes
Documentation
<!-- end of auto-generated comment: release notes by coderabbit.ai --> </details>
<details> <summary>fix(config): stop silently dropping malformed atmos.d/.atmos.d files @osterman (#2837)</summary>
atmos.d/ and .atmos.d/ config files that fail to parse now hard-fail LoadConfig with the offending file path and YAML line number, instead of being silently swallowed at debug log level.atmos.yaml was found, its co-located atmos.d/.atmos.d is checked) and the zero-config fallback path (no atmos.yaml anywhere, opportunistic git-root .atmos.d check).atmos version/--version, --help, atmos config validate/atmos validate config/atmos validate schema config, and CI git-clone bootstrap are unaffected — those commands already continue past config-init errors by design.atmos.d/ and .atmos.d/, the exact sort-order repro from the issue (good/broken/good files), and full LoadConfig integration tests for both call paths.docs/fixes/2026-07-30-atmos-d-malformed-yaml-silent-drop.md documenting the fix.atmos.d//.atmos.d/ caused Atmos to exit 0 with no visible error, and because the merge loop bails on the first bad file, every file sorting after the broken one silently never loaded either — whether a setting applied depended on its filename's sort position relative to an unrelated broken file.atmos.yaml and malformed profile configs (#2825) already hard-fail. This closes the one remaining silent source, reusing the exact error already produced by the existing merge code (only the swallow points needed to change).</details>
<details> <summary>Speed up container discovery and harden emulator listings @osterman (#2828)</summary>
<!-- This is an auto-generated comment: release notes by coderabbit.ai -->
New Features
Bug Fixes
<details> <summary>fix: don't process Go templates in Terraform source code @thejrose1984 (#2830)</summary>
component_info section from Go template rendering in both the
ProcessStacks and describe stacks pipelinescomponent_info stays in the template context, so {{ .component_info.component_path }}
and friends keep working in stack manifeststerraform-source-go-templates fixture whose Terraform
description contains {{project}}terraform-config-inspect and
stores the result — including every variable and output description — in
component_info.terraform_config. That section is part of the component section Atmos
serializes and renders as a Go template.description that legitimately contains double curly braces, such as the GCP resource
name format projects/{{project}}/locations/{{location}}/services/{{name}}, aborted every
Atmos command with template: templates-all-atmos-sections:156: function "project" not defined.<!-- This is an auto-generated comment: release notes by coderabbit.ai -->
Bug Fixes
component_info content is no longer incorrectly processed as Go templates.{{...}} and }} are preserved while other stack templates continue to render normally.Documentation
component_info is protected from template rendering but remains available for use in other templates.<!-- end of auto-generated comment: release notes by coderabbit.ai --> </details>
<details> <summary>fix(auth): never cache ambient provider credentials in the keyring @thejrose1984 (#2819)</summary>
Stops the auth manager persisting credentials to the keyring for providers that re-resolve their principal from the environment on every authentication (gcp/adc, gcp/workload-identity-federation, azure/cli, azure/oidc, github/oidc). Such chains are never served from, nor written to, the keyring, and stale entries are purged so poisoned keyrings self-heal.
gcp/adc is documented as stateless, but its short-lived token was cached and
replayed, so after gcloud auth application-default login switched accounts
Atmos kept authenticating as the previous principal. atmos/pro is deliberately
excluded: it mints a single-use token, so cached reuse is load-bearing.
Closes #2695 Suggested release label: patch
<!-- This is an auto-generated comment: release notes by coderabbit.ai -->
auth logout, auth logout provider, and auth logout --all now clear ambient-related keyring entries even without --keychain.--dry-run logout previews to match real cleanup behavior.<details> <summary>fix: close yq/merge concurrency races, route yq logs through Atmos logger @osterman (#2826)</summary>
go-logging) diagnostics through the Atmos logger via a new internal/yq package, so they inherit Atmos's formatting, configured log destination, and secret masking instead of writing straight to stderr, unformatted and unmasked.pkg/utils and pkg/yaml, closing a cross-package data race left over after #2822.pkg/merge.MergeContext.WithFile, where concurrent per-stack-file goroutines sharing a parent import chain could write into the same backing-array slot at the same time.brace-expansion dependency (via pnpm.overrides) to patched versions, closing Dependabot alert #261.go-logging's SetBackend wraps any plain Backend in an unsynchronized, map-based type unless the backend itself implements Leveled. #2822's mutex-based fix for #2821 only covered pkg/utils, leaving pkg/yaml/edit.go free to mutate the same global state independently — confirmed with go test -race.MergeContext.WithFile race surfaced incidentally while validating the yq fix under -race (TestExecuteHelmfile_ComponentNotFound), and turned out to be a genuine, separate concurrency bug worth fixing here rather than leaving in place.brace-expansion's expand() bounded the number of results but not their length, letting a small attacker-controlled input crash the Node process with an uncatchable out-of-memory error (CVE-2026-14257). The fix stays within the pinned major lines, so it isn't blocked by dependabot.yml's major-version-bump policy.<!-- This is an auto-generated comment: release notes by coderabbit.ai -->
<details> <summary>fix(config): validate invalid configuration @osterman (#2825)</summary>
atmos config validate must diagnose invalid Atmos configuration instead of being blocked by the same decoding failure it is intended to report.<!-- This is an auto-generated comment: release notes by coderabbit.ai -->
<details> <summary>Add env-step export controls to workflows and hooks @osterman (#2814)</summary>
type: env steps while preserving template assignment.<!-- This is an auto-generated comment: release notes by coderabbit.ai -->
export option for workflow and task env steps to control whether values propagate to later child-process environments (default: true).env propagation and precedence so later steps can reliably resolve {{ .env.NAME }} and so exports are isolated across retries and ordered hooks.export: false keeps values template-only (not set for later subprocess environments).env step exports.<details> <summary>fix(container): apply Buildx driver and cache configuration @osterman (#2815)</summary>
<!-- This is an auto-generated comment: release notes by coderabbit.ai -->
New Features
cache-from/cache-to.Bug Fixes
buildx-compatible engine (unless using bake).Tests / CI
<details>
<summary>fix(utils): initialize yq globals once instead of per evaluation @arcaven (#2822)</summary>
Concurrent stack file processing can end an Atmos run with fatal error: concurrent map writes. That is a runtime fatal error rather than a panic, so the global handler added in #2334 cannot intercept it and no retry inside Atmos can either. go test -race flags the cause on main from a 20 line test, and the fix keeps the hot path read-only.
configureYqLogger no longer calls logging.SetLevel on every EvaluateYqExpression and EvaluateYqExpressionWithType call. The default level is installed in init(), and later calls rewrite it only when the wanted level is not already installed, under a mutex.sync.Once rather than being lazily assigned by every Evaluate call.TestEvaluateYqExpression_ConcurrentCallsAreRaceFree, which fails under -race without either change.Both globals were written on every evaluation, from the per stack file goroutines that processYAMLConfigFileWithContextInternal spawns:
logging.SetLevel writes an unsynchronized map inside go-logging, and yq reads that same map from every decoder through Logger.Debugf. The Go runtime checks concurrent map access, so this is the one that reaches users as a crash.yqlib.InitExpressionParser assigns the exported global yqlib.ExpressionParser behind a plain nil check. That write is a pointer, so the runtime does not catch it and it corrupts quietly instead.On unmodified main at fc4960a the new test reports Found 2 data race(s). With the change, pkg/utils, pkg/yaml/... and internal/exec all pass, the last two unchanged here but sharing the same yq globals.
Two notes for reviewers:
init() is what removes the write entirely for a non-Trace run. --logs-level Trace still performs one write on the first transition, which could in principle race a yq read already in flight. Hoisting the call to configuration load would close that as well; I did not want to reach into pkg/config uninvited.pkg/yaml/edit.go:71 calls logging.SetLevel on every evaluateWithOptions, so it writes the same map. I have not measured whether that path runs concurrently, so I left it alone rather than guess. Happy to fold it in here, or to file it separately, whichever you prefer.Cost: yqlib.InitExpressionParser measures about 0.4ms once, and the sync.Once keeps that off commands which never evaluate an expression.
<!-- This is an auto-generated comment: release notes by coderabbit.ai -->
Bug Fixes
Tests
<!-- end of auto-generated comment: release notes by coderabbit.ai --> </details>
Nothing published for this version
Nothing published for this version
Nothing published for this version
…vulnerability (GHSA-7p8r-x3mc-p8w7 / CVE-2026-18446), flagged by Dependabot after this branch was pushed.
<details> <summary>feat(store): add atmos store CRUD CLI and type: store workflow step @osterman (#2858)</summary>
atmos store — a new CLI command group (set/get/delete/list) for raw CRUD access to any store backend configured under stores: in atmos.yaml (AWS SSM, AWS Secrets Manager, HashiCorp Vault, Azure Key Vault, GCP Secret Manager, Redis, Artifactory, 1Password, Keychain, GitHub Actions). Unlike atmos secret, no declaration is required — any key can be read, written, or deleted directly by name, optionally scoped to a stack and component.type: store workflow step that writes a value from a workflow, custom command, or hook — usable automatically as a hook too via the existing generic kind: step bridge, with no extra wiring.!store/!store.get YAML functions, so pipeline metadata (an image tag from a build step, a build number, a deployment marker) can be handed off to a completely different stack or component.pkg/store.Service facade, the cmd/store command family, and the store step handler, plus CLI/website docs, a changelog post, and a roadmap milestone.!store/!store.get); the only existing write path was a single Terraform-output-specific hook (kind: store), so anything else needing to be written into a store meant scripting around Atmos with a cloud CLI.<!-- This is an auto-generated comment: release notes by coderabbit.ai -->
New Features
atmos store commands to set, get, list, and delete values across configured backends.type: store workflow steps for writing templated values.Documentation
Tests
<details> <summary>Add tfmigrate support for Terraform components @osterman (#2534)</summary>
atmos terraform migrate plan, apply, and list for running user-authored tfmigrate migrations in Terraform component context.kind: tfmigrate lifecycle hooks with dynamic/static modes, toolchain resolution, Terraform/OpenTofu exec path wiring, and same-identity auth handling.tfmigrate history storage.tfmigrate history storage, so Atmos now documents and exposes the values users need without taking on history persistence in v1.<!-- This is an auto-generated comment: release notes by coderabbit.ai -->
atmos terraform migrate plan, apply, and list commands.tfmigrate lifecycle hooks with dynamic and explicit execution modes.<details> <summary>feat(vendor): native component updater PR workflow @osterman (#2756)</summary> <a href="https://atmos.tools/ci"><picture> <source media="(prefers-color-scheme: dark)" srcset="https://atmos.tools/img/atmos-ci-gradient.svg"> <source media="(prefers-color-scheme: light)" srcset="https://atmos.tools/img/atmos-ci-gradient-on-light.svg"> <img src="https://atmos.tools/img/atmos-ci-gradient-on-light.svg" alt="Atmos CI" height="32" align="right"> </picture></a>
Manually tested end-to-end as a real user — following --help/docs, in isolated sandboxes, and against a real repository (cloudposse/infra-live):
vendor.ci.pull_request config surface (title/body templates, labels, draft, reviewers, assignees) — #1701, #1702 (left as drafts, not merged).--pull-request created a pull request but never printed its URL in the default table output.atmos.yaml's vendor.update.*/vendor.ci.* config (groups, execution mode, PR title/labels/draft/reviewers) was silently ignored — read from the wrong viper instance instead of the parsed config.--all double-counted every component when a repo vendors exclusively via component.yaml and leaves an unused component type (e.g. packer) unconfigured.oci-artifacts coverage entry always claimed "complete" regardless of whether any OCI artifact existed in the project.atmos vendor verify/clean/--refresh-lock/--lock-enforcement (all three modes) and atmos sbom generate (CycloneDX vs. SPDX, NTIA mode, experimental gating, upload-outside-CI) in isolated sandboxes.ATMOS_PRO_GITHUB_TOKEN (the token github/sts mints) to the Component Updater's GitHub token precedence, so atmos auth exec --identity <github-sts-identity> -- atmos vendor update --pull-request gets a token that triggers downstream Actions workflows on the PR it opens — unlike the default GITHUB_TOKEN, which GitHub excludes from re-triggering workflows.<!-- This is an auto-generated comment: release notes by coderabbit.ai -->
atmos sbom generate with provenance/NTIA modes, SPDX/CycloneDX output, and optional CI artifact upload.atmos vendor update --pull-request workflows with component/group selection and deterministic PR publishing.atmos vendor verify, atmos vendor clean, and vendor.lock.yaml drift protection.</details>
<details> <summary>feat(workflow): support tags and labels selectors @zack-is-cool (#2857)</summary>
--tags and --labels on atmos workflow, forwarding them to nested type: atmos steps alongside optional --stack.go test ./internal/exec -run TestExecuteWorkflow_ForwardsCommandLineFilters -count=1go test ./pkg/workflow -run 'TestAppendAtmosStepFlags|TestControlCommandExecutorExecuteAtmos' -count=1go test ./cmd/workflow -run TestWorkflowSelectorFlags -count=1git diff --check<!-- This is an auto-generated comment: release notes by coderabbit.ai -->
New Features
--tags and --labels selectors to workflow commands.--stack while preserving workflow ordering and execution behavior.Documentation
<details> <summary>fix(config): resolve git-root base_path for --config/--config-path @osterman (#2864)</summary>
pkg/config/load_config_args.go so that loading configuration via --config/--config-path also applies git-root discovery for an empty (or .) base_path, matching the plain auto-discovery flow in LoadConfig().TestLoadConfigFromCLIArgs_AppliesGitRootBasePath) that reproduces the bug and verifies base_path now resolves correctly.fast-uri transitive dependency (website) from 3.1.4 to 3.1.5 to remediate a high-severity host-confusion vulnerability (GHSA-7p8r-x3mc-p8w7 / CVE-2026-18446), flagged by Dependabot after this branch was pushed.loadConfigFromCLIArgs() never called applyGitRootBasePath(), unlike the main LoadConfig() auto-discovery path. As a result, a project with base_path: '' in atmos.yaml resolved correctly via plain auto-discovery but left base_path empty when the identical config was loaded via --config, breaking component/stack path resolution (e.g. atmos terraform test) with Error: failed to find import.fast-uri bump addresses a live Dependabot alert (a patch-level version bump, not blocked by .github/dependabot.yml's major-version ignore policy) surfaced automatically after pushing this branch.<!-- This is an auto-generated comment: release notes by coderabbit.ai -->
Bug Fixes
"." base path to the Git repository root.Tests
<details> <summary>fix(auth): prevent Azure CLI cache corruption for guest users @aknysh (#2861)</summary>
azure/cli provider — az's own cache is authoritative, and writing back what came from az is what corrupted it.AzureCredentials (cli / device_code / oidc) so the write-back can be gated per provider kind.azure/device-code provider (silent and interactive flows) and use it in both Azure CLI cache writers (UpdateAzureCLIFiles and the provider-level updateAzureCLICache), falling back to the previous {oid}.{tenant} derivation when unavailable.azure/subscription identity's field-by-field credential copy with a struct copy plus explicit overrides, and add a reflection-based regression test that fails if any future AzureCredentials field is dropped by the wrap.TestNewMSALCache's default-path case from the real ~/.azure.atmos auth login, the Azure CLI cache write-back created an MSAL Account entry with home_account_id derived as {oid}.{target-tenant} and hardcoded account_source: "device_code". For guest (B2B) users the home tenant differs from the target tenant, so az ended up with two Account entries for the same username and failed every subsequent command with Found multiple accounts with the same username (azure-cli#20168) — including az account get-access-token, which the azure/cli provider itself shells out to. In other words, one atmos auth login broke both az and the next atmos login for any guest user.az login → atmos auth login → az broken; after the fix, az stays healthy, the cache keeps exactly one Account entry, and the persisted credentials carry auth_method so the gate holds across credential caching.<!-- This is an auto-generated comment: release notes by coderabbit.ai -->
<!-- end of auto-generated comment: release notes by coderabbit.ai --> </details>
Nothing published for this version
Nothing published for this version
Nothing published for this version
Nothing published for this version
feat: add date-anchored default editions @osterman
<details> <summary>feat: add date-anchored default editions @osterman (#2762)</summary>
edition: defaults, the --edition override, and atmos describe/list edition commands.<!-- This is an auto-generated comment: release notes by coderabbit.ai -->
New Features
--edition, ATMOS_EDITION, or atmos.yaml.describe edition and list editions commands for inspecting default changes.--use-mocks.Bug Fixes
Documentation
<details> <summary>refactor(store): move backends into pkg/store/providers subpackage @osterman (#2575)</summary>
pkg/store into a new pkg/store/providers subpackage, keeping pkg/store as a pure interface/type boundary (the Store interfaces, StoreConfig/StoresConfig/StoreRegistry types, auth-config types, error sentinels, and generated mocks).pkg/store owns StoreRegistry, NewStoreRegistry, and a Register(type, factory) API; each backend registers its factory from an init(), so the type switch is replaced by a map lookup. pkg/config blank-imports pkg/store/providers so the built-in backends register at startup (database/sql driver pattern).pkg/store/providers exclusion to the provider-agnostic-auth depguard rule so its cloud-SDK imports are permitted (matching pkg/auth/providers).pkg/auth/providers / pkg/secrets/providers layout and making pkg/store a clean type/interface package.providers but was named after the StoreRegistry type it returned; pkg/store now owns both the registry type and its construction, and adding a backend is one self-contained file that registers itself.ErrStoreTypeNotFound, identity warnings are preserved. Builds clean, all affected tests pass, and golangci-lint --new-from-rev=origin/main reports zero issues.<!-- This is an auto-generated comment: release notes by coderabbit.ai -->
<details> <summary>feat(toolchain): add --format=plain/json to atmos toolchain get @osterman (#2845)</summary>
--format flag to atmos toolchain get with three modes: table (default, unchanged), plain (bare version string only), and json (structured output with tool/version/installed fields, or a full version list under --all).plain/json output through the data channel (stdout, pipeable) instead of the styled UI channel (stderr), via new printVersionsPlain/printVersionsJSON helpers in pkg/toolchain/get.go.--format=plain combined with --all with a new ErrToolchainPlainFormatWithAllFlag sentinel, since there's no single version to print in that case.2>&1 since it's written to stderr), e.g. atmos toolchain get vale-cli/vale 2>&1 | grep -oE '[0-9]+\.[0-9]+\.[0-9]+' | head -1.--format=plain collapses that to version=$(atmos toolchain get vale-cli/vale --format=plain), and --format=json gives scripts structured access to installed status without adding a new dependency.<!-- This is an auto-generated comment: release notes by coderabbit.ai -->
atmos toolchain get.--format/-f, environment-variable support, and shell completion.--all and plain output options.<details> <summary>fix(scaffold): resolve relative write-target directories consistently @osterman (#2855)</summary>
atmos scaffold generate so a relative target directory (e.g. the CLI's own default ./my-project) works, instead of rejecting every file with path traversal not allowed.validateWriteTarget in pkg/generator/engine/templating.go now resolves the write directory (realDir) through the same ResolveAndCleanBasePath helper already used for the target base (realBase), instead of a bare filepath.EvalSymlinks that stays relative for relative inputs.TestProcessFile_RelativeTargetPath, covering a relative targetPath end-to-end (previous tests only exercised absolute t.TempDir() targets, so this case was never caught).realBase was always absolutized before comparison, but realDir was resolved with a bare filepath.EvalSymlinks, which returns a relative path unchanged when given a relative input. Comparing an absolute path against a relative one never matched the containment check, so it fired as a false-positive path traversal on every write whenever the target directory was relative — including the command's own default target.filepath.Dir(fullPath) was already absolute in that case, masking the bug.<!-- This is an auto-generated comment: release notes by coderabbit.ai -->
Bug Fixes
./my-project.Tests
<!-- end of auto-generated comment: release notes by coderabbit.ai --> </details>
Nothing published for this version
Nothing published for this version
Nothing published for this version
Nothing published for this version
fix(config): stop recommending deprecated stacks.name_pattern @osterman
<details> <summary>feat: dependency-closure selection (--include-dependencies/--include-dependents) + scope --labels/--tags evaluation before filtering @osterman (#2807)</summary>
--all, --components, --query, -s, --tags, --labels, --affected) accepts two new depth-carrying flags on plan/apply/deploy/destroy/init:
--include-dependencies[=N] — also process everything the selection depends on (its prerequisites), N levels deep (bare flag = unlimited).--include-dependents[=N] — the reverse direction. Previously a bool wired only into --affected; now works with every selection and accepts a depth (true/false remain accepted for compatibility).destroy --include-dependencies warns that it also destroys shared prerequisites.atmos list components/stacks/instances preview the exact execution set a bulk run would touch.pkg/dependency.Filter gains per-direction depth bounds with a best-depth BFS; the scheduler adapter is restructured to seed-then-expand (tags/labels/query moved from post-filter to seed narrowing, per-node query skip suppressed when the seed already applied it).list dependencies is generalized into dependencies.ResolveScopedClosure and shared with the terraform bulk paths, so closure runs fully evaluate only the stacks the closure touches.metadata.tags/metadata.labels are selectors evaluated before auth/templating/YAML functions, so their values must be resolvable without authentication or process execution. Values using !terraform.state, !terraform.output, !store, !store.get, !secret, !aws.*, !emulator, !exec, !random, or template calls to atmos.Component/atmos.Store/atmos.GomplateDatasource/datasources are rejected with a by-design error and migration hint on any command that processes the manifest. Plain strings, simple templates, and local functions (!env, !git.*, !include) remain allowed.templates.settings.delimiters) via Go's template parser — no regex heuristics — and tags.SelectorUnresolved is now delimiter-aware too.--labels/--tags (original scope)--tags/--labels skips auth/template/YAML-function evaluation entirely, generalizing the existing -s/--stack early-skip. Now also threaded into list stacks/components/instances (previously they row-filtered after full evaluation).atmos list dependencies computes a lightweight dependency graph, derives the reachable closure, and only fully evaluates the stacks that closure touches — re-converging against the resolved graph so templated same-stack dependency targets still resolve.describe.settings.eager_evaluation remains the rollback switch forcing the old full-evaluation behavior.pkg/tags (purity + delimiter cases), pkg/dependency (depth/cycles/diamonds), pkg/scheduler/adapters (closure retains non-matching prereqs, destroy ordering, query suppression, depth merging), pkg/list/dependencies (scoped-evaluation convergence with a poisoned unrelated stack), and the list/cmd layers; regenerated help-text golden snapshots.website/blog/2026-07-27-include-dependencies-closure.mdx), and a roadmap entry.atmos terraform ... --all --labels=.../--tags=... and atmos list dependencies --stack <stack> fully evaluated every stack in the repo (templates, YAML functions, auth/backend) before consulting the selector, so an unrelated account's unreachable backend failed the command.docs/fixes/2026-06-22-describe-stacks-scope-and-cache-per-component-auth.md.docs/fixes/2026-07-25-scope-before-evaluate-labels-tags-list-dependencies.md for root-cause and verification detail of the scope-before-evaluate work (its Recommendations section is implemented by this PR's closure flags).🤖 Generated with Claude Code
<!-- This is an auto-generated comment: release notes by coderabbit.ai -->
New Features
levels output for shortest dependency distances.key=value and key:value formats.Improvements
<details> <summary>fix(config): stop recommending deprecated stacks.name_pattern @osterman (#2842)</summary>
atmos aws eks update-kubeconfig and Spacelift stack-name generation now check stacks.name_template before falling back to the deprecated stacks.name_pattern, instead of only supporting the deprecated field.pkg/config, errors/errors.go, and pkg/helmfile/cluster.go that previously only pointed users at the deprecated fields now recommend name_template/cluster_name_template.aws eks update-kubeconfig command help/docs no longer teach the deprecated name_pattern/cluster_name_pattern fields.examples/, demo/, and non-backward-compat-test tests/fixtures scenarios from name_pattern to name_template (a handful of fixtures that specifically test the deprecated field, precedence, or backward compatibility were left untouched on purpose).name_template support and continued name_pattern backward compatibility for the two code paths that changed.stacks settings are not consistently applied") showed the profile-merge pipeline was already correct and consistent between describe config and list dependencies.stacks.name_pattern — deprecated in favor of stacks.name_template — was still treated as the primary/only stack-naming mechanism in a couple of code paths and in several error messages, which is what produced the reported inconsistency and general user confusion about which field to use.name_pattern continues to work unchanged for backward compatibility; only the recommended path and documentation change.<!-- This is an auto-generated comment: release notes by coderabbit.ai -->
Nothing published for this version
Nothing published for this version
Bump the brace-expansion pnpm.overrides (website) to 1.1.18/2.1.4, patching CVE-2026-14257 / GHSA-mh99-v99m-4gvg (high-severity DoS via unbounded expa…
<details> <summary>test(config): regression test for atmos.d commands merge (#2570) @osterman (#2840)</summary>
TestMergeConfig_AtmosDCommandsMerging_TopLevelYamlFunction to pkg/config/config_import_test.go, reproducing the exact minimal repro from #2570: two .atmos.d/ files where the last-processed file has a top-level Atmos YAML function (!repo-root) directly on a command's own field.0b1182c6bb (PR #2677, released in v1.223.0), which strips the commands key from file content before preprocessAtmosYamlFunc runs so it can no longer overwrite the already-merged commands array.</details>
<details> <summary>fix(toolchain): accept semver release-candidate strings in --use-version @osterman (#2841)</summary>
isValidSemver() in pkg/toolchain/version_spec.go so it accepts semver pre-release and
build-metadata suffixes (e.g. 1.225.0-rc.3, 1.2.3+build.5), by delegating to the
already-vendored Masterminds/semver/v3 library instead of a hand-rolled digit-only check.pkg/toolchain/version_spec_test.go (unit-level
isValidSemver/ParseVersionSpec cases) and pkg/version/reexec_test.go (end-to-end at the
--use-version entry point).docs/fixes/2026-07-31-use-version-release-candidate-semver.md.atmos --use-version=1.225.0-rc.3 failed with invalid version output format, even though
1.225.0-rc.3 is a spec-compliant semver string. The version parser split on . and required
every part to be pure digits, which rejects any release-candidate/pre-release version.<!-- This is an auto-generated comment: release notes by coderabbit.ai -->
New Features
--use-version now accepts release-candidate versions such as 1.225.0-rc.3, including versions with build metadata.latest.Bug Fixes
Documentation
<details> <summary>fix(dag): stop concurrent map crash in bulk terraform commands @osterman (#2831)</summary>
fatal error: concurrent map iteration and map write crash in DAG-scheduled bulk terraform commands (terraform <cmd> --all/--affected/--query) at higher --max-concurrency.ProcessComponentConfig now shallow-clones the component section before any downstream code mutates it, so concurrent workers never write into the map tree owned by the shared FindStacksMap cache.imports, and terraform_workspace_pattern/terraform_workspace_template, from cache-owned maps in place).internal/exec/process_stacks_shared_cache_test.go) that fail pre-fix both deterministically and under -race.brace-expansion pnpm.overrides (website) to 1.1.18/2.1.4, patching CVE-2026-14257 / GHSA-mh99-v99m-4gvg (high-severity DoS via unbounded expansion length), reported by Dependabot alert #261.FindStacksMap caches processed stack config and returns it by reference on cache hits, shared across all goroutines within a process. ProcessStacks and mergeGlobalAuthConfig write top-level keys into that shared component section, while findComponentInStacks has every DAG worker iterate every stack's component section (not just its own) looking for a match — so one worker's write races with another worker's read/iteration of the same cached map, crashing exactly as reported.ProcessStacks call in the same process even outside the crash path.brace-expansion bump addresses an open, high-severity Dependabot alert; deferring it risks a DoS crash if attacker-influenced input reaches an affected glob/brace-pattern code path in the docs site tooling.<!-- This is an auto-generated comment: release notes by coderabbit.ai -->
Bug Fixes
describe output to include referenced imports consistently.Documentation
<details> <summary>Support explicit CI git checkout and bundle Docker CLI @osterman (#2812)</summary>
--ci and ATMOS_CI controls for no-argument CI checkout, including explicit opt-out behavior.cmd/root.go for CI git-clone bootstrap detection with Cobra-identity + the existing cmd/git flag-handler infrastructure (resolveCICloneMode), removing ~140 lines of bespoke parsing.atmos git clone's no-arg CI checkout to resolve the branch from the CI provider's parsed short name instead of the raw ref, which previously failed real branch/PR checkouts.actions/checkout (docs/prd/git-ops.md) never actually worked for a real branch or PR.<!-- This is an auto-generated comment: release notes by coderabbit.ai -->
New Features
--ci support for atmos git clone, configurable with ATMOS_CI.Bug Fixes
Documentation
<!-- end of auto-generated comment: release notes by coderabbit.ai --> </details>
Nothing published for this version
Nothing published for this version
Nothing published for this version
fix(test): stop Windows toolchain-vanishing flake in acceptance tests @osterman
<details> <summary>fix(test): stop Windows toolchain-vanishing flake in acceptance tests @osterman (#2834)</summary>
TestPrintTelemetryDisclosureOnlyOnce (and its three sibling telemetry disclosure
tests) from deleting the shared <cache>/atmos root; they now isolate via a per-test
ATMOS_XDG_CACHE_HOME/XDG_CACHE_HOME redirect instead.internal/exec terraform/tofu version tests: RequireTerraformPath/
RequireTofuPath now resolve the binary path once and hand it back directly, instead of a
second independent exec.LookPath call at each call site.requireExecutablePath fail loudly (t.Fatalf) instead of silently returning an empty
path when ATMOS_TEST_SKIP_PRECONDITION_CHECKS=true and the binary isn't found.pkg/toolchain install/uninstall/list tests from the real shared XDG
toolchain cache directory (Toolchain.InstallPath now points at each test's own t.TempDir()).requireExecutablePath's retry window (2s → 15s) as a defensive backstop, and adds
permanent forensic instrumentation (executableLookupForensics) that dumps per-PATH-entry
toolchain directory state on a lookup failure.docs/fixes/ entry recording the root cause and investigation.executable file not found in %PATH% failure for tofu/terraform, always on a different
victim test in internal/exec each run.TestPrintTelemetryDisclosureOnlyOnce
was calling os.RemoveAll on the shared <cache>/atmos root — the same root the toolchain
install path lives under since #2579 — twice per run (setup + defer), deleting every
CI-provisioned tool (terraform, opentofu, helm, helmfile) out from under the other
concurrently-running go test package binaries.</details>
<details> <summary>fix(config): stop silently dropping malformed atmos.d/.atmos.d files @osterman (#2837)</summary>
atmos.d/ and .atmos.d/ config files that fail to parse now hard-fail LoadConfig with the offending file path and YAML line number, instead of being silently swallowed at debug log level.atmos.yaml was found, its co-located atmos.d/.atmos.d is checked) and the zero-config fallback path (no atmos.yaml anywhere, opportunistic git-root .atmos.d check).atmos version/--version, --help, atmos config validate/atmos validate config/atmos validate schema config, and CI git-clone bootstrap are unaffected — those commands already continue past config-init errors by design.atmos.d/ and .atmos.d/, the exact sort-order repro from the issue (good/broken/good files), and full LoadConfig integration tests for both call paths.docs/fixes/2026-07-30-atmos-d-malformed-yaml-silent-drop.md documenting the fix.atmos.d//.atmos.d/ caused Atmos to exit 0 with no visible error, and because the merge loop bails on the first bad file, every file sorting after the broken one silently never loaded either — whether a setting applied depended on its filename's sort position relative to an unrelated broken file.atmos.yaml and malformed profile configs (#2825) already hard-fail. This closes the one remaining silent source, reusing the exact error already produced by the existing merge code (only the swallow points needed to change).</details>
<details> <summary>Speed up container discovery and harden emulator listings @osterman (#2828)</summary>
<!-- This is an auto-generated comment: release notes by coderabbit.ai -->
New Features
Bug Fixes
<details> <summary>fix: don't process Go templates in Terraform source code @thejrose1984 (#2830)</summary>
component_info section from Go template rendering in both the
ProcessStacks and describe stacks pipelinescomponent_info stays in the template context, so {{ .component_info.component_path }}
and friends keep working in stack manifeststerraform-source-go-templates fixture whose Terraform
description contains {{project}}terraform-config-inspect and
stores the result — including every variable and output description — in
component_info.terraform_config. That section is part of the component section Atmos
serializes and renders as a Go template.description that legitimately contains double curly braces, such as the GCP resource
name format projects/{{project}}/locations/{{location}}/services/{{name}}, aborted every
Atmos command with template: templates-all-atmos-sections:156: function "project" not defined.<!-- This is an auto-generated comment: release notes by coderabbit.ai -->
Bug Fixes
component_info content is no longer incorrectly processed as Go templates.{{...}} and }} are preserved while other stack templates continue to render normally.Documentation
component_info is protected from template rendering but remains available for use in other templates.<!-- end of auto-generated comment: release notes by coderabbit.ai --> </details>
<details> <summary>fix(auth): never cache ambient provider credentials in the keyring @thejrose1984 (#2819)</summary>
Stops the auth manager persisting credentials to the keyring for providers that re-resolve their principal from the environment on every authentication (gcp/adc, gcp/workload-identity-federation, azure/cli, azure/oidc, github/oidc). Such chains are never served from, nor written to, the keyring, and stale entries are purged so poisoned keyrings self-heal.
gcp/adc is documented as stateless, but its short-lived token was cached and
replayed, so after gcloud auth application-default login switched accounts
Atmos kept authenticating as the previous principal. atmos/pro is deliberately
excluded: it mints a single-use token, so cached reuse is load-bearing.
Closes #2695 Suggested release label: patch
<!-- This is an auto-generated comment: release notes by coderabbit.ai -->
auth logout, auth logout provider, and auth logout --all now clear ambient-related keyring entries even without --keychain.--dry-run logout previews to match real cleanup behavior.Nothing published for this version
Nothing published for this version
Nothing published for this version
…process with an uncatchable out-of-memory error (CVE-2026-14257). The fix stays within the pinned major lines, so it isn't blocked by dependabot.yml's…
<details> <summary>ci: run required checks for GitHub merge queue @osterman (#2813)</summary>
<!-- This is an auto-generated comment: release notes by coderabbit.ai -->
<!-- end of auto-generated comment: release notes by coderabbit.ai --> </details>
<details> <summary>fix: close yq/merge concurrency races, route yq logs through Atmos logger @osterman (#2826)</summary>
go-logging) diagnostics through the Atmos logger via a new internal/yq package, so they inherit Atmos's formatting, configured log destination, and secret masking instead of writing straight to stderr, unformatted and unmasked.pkg/utils and pkg/yaml, closing a cross-package data race left over after #2822.pkg/merge.MergeContext.WithFile, where concurrent per-stack-file goroutines sharing a parent import chain could write into the same backing-array slot at the same time.brace-expansion dependency (via pnpm.overrides) to patched versions, closing Dependabot alert #261.go-logging's SetBackend wraps any plain Backend in an unsynchronized, map-based type unless the backend itself implements Leveled. #2822's mutex-based fix for #2821 only covered pkg/utils, leaving pkg/yaml/edit.go free to mutate the same global state independently — confirmed with go test -race.MergeContext.WithFile race surfaced incidentally while validating the yq fix under -race (TestExecuteHelmfile_ComponentNotFound), and turned out to be a genuine, separate concurrency bug worth fixing here rather than leaving in place.brace-expansion's expand() bounded the number of results but not their length, letting a small attacker-controlled input crash the Node process with an uncatchable out-of-memory error (CVE-2026-14257). The fix stays within the pinned major lines, so it isn't blocked by dependabot.yml's major-version-bump policy.<!-- This is an auto-generated comment: release notes by coderabbit.ai -->
<details> <summary>fix(config): validate invalid configuration @osterman (#2825)</summary>
atmos config validate must diagnose invalid Atmos configuration instead of being blocked by the same decoding failure it is intended to report.<!-- This is an auto-generated comment: release notes by coderabbit.ai -->
<details> <summary>Add env-step export controls to workflows and hooks @osterman (#2814)</summary>
type: env steps while preserving template assignment.<!-- This is an auto-generated comment: release notes by coderabbit.ai -->
export option for workflow and task env steps to control whether values propagate to later child-process environments (default: true).env propagation and precedence so later steps can reliably resolve {{ .env.NAME }} and so exports are isolated across retries and ordered hooks.export: false keeps values template-only (not set for later subprocess environments).env step exports.<details> <summary>fix(container): apply Buildx driver and cache configuration @osterman (#2815)</summary>
<!-- This is an auto-generated comment: release notes by coderabbit.ai -->
New Features
cache-from/cache-to.Bug Fixes
buildx-compatible engine (unless using bake).Tests / CI
<details>
<summary>fix(utils): initialize yq globals once instead of per evaluation @arcaven (#2822)</summary>
Concurrent stack file processing can end an Atmos run with fatal error: concurrent map writes. That is a runtime fatal error rather than a panic, so the global handler added in #2334 cannot intercept it and no retry inside Atmos can either. go test -race flags the cause on main from a 20 line test, and the fix keeps the hot path read-only.
configureYqLogger no longer calls logging.SetLevel on every EvaluateYqExpression and EvaluateYqExpressionWithType call. The default level is installed in init(), and later calls rewrite it only when the wanted level is not already installed, under a mutex.sync.Once rather than being lazily assigned by every Evaluate call.TestEvaluateYqExpression_ConcurrentCallsAreRaceFree, which fails under -race without either change.Both globals were written on every evaluation, from the per stack file goroutines that processYAMLConfigFileWithContextInternal spawns:
logging.SetLevel writes an unsynchronized map inside go-logging, and yq reads that same map from every decoder through Logger.Debugf. The Go runtime checks concurrent map access, so this is the one that reaches users as a crash.yqlib.InitExpressionParser assigns the exported global yqlib.ExpressionParser behind a plain nil check. That write is a pointer, so the runtime does not catch it and it corrupts quietly instead.On unmodified main at fc4960a the new test reports Found 2 data race(s). With the change, pkg/utils, pkg/yaml/... and internal/exec all pass, the last two unchanged here but sharing the same yq globals.
Two notes for reviewers:
init() is what removes the write entirely for a non-Trace run. --logs-level Trace still performs one write on the first transition, which could in principle race a yq read already in flight. Hoisting the call to configuration load would close that as well; I did not want to reach into pkg/config uninvited.pkg/yaml/edit.go:71 calls logging.SetLevel on every evaluateWithOptions, so it writes the same map. I have not measured whether that path runs concurrently, so I left it alone rather than guess. Happy to fold it in here, or to file it separately, whichever you prefer.Cost: yqlib.InitExpressionParser measures about 0.4ms once, and the sync.Once keeps that off commands which never evaluate an expression.
<!-- This is an auto-generated comment: release notes by coderabbit.ai -->
Bug Fixes
Tests
<!-- end of auto-generated comment: release notes by coderabbit.ai --> </details>
Nothing published for this version
Nothing published for this version
Nothing published for this version
Nothing published for this version
Support parent-scoped multi-file stacks @osterman
<details> <summary>Support parent-scoped multi-file stacks @osterman (#2787)</summary>
metadata.inherits graph self-contained while canonicalizing equivalent imported duplicates by lexical parent path.docs/prd/top-level-stack-composition.md<!-- This is an auto-generated comment: release notes by coderabbit.ai -->
New Features
Bug Fixes
Documentation
<details> <summary>feat(stacks): support global-scope metadata defaults @osterman (#2808)</summary>
metadata fields (labels, tags, custom, enabled, locked, terraform_workspace_pattern) at the stack-manifest root, deep-merged into every component's own metadata as a stack-wide default.metadata.inherits base-component chain → the component's own local metadata: block, which always wins.component, inherits, type, name, terraform_workspace) remain component-only; setting one of these at global scope is now a hard validation error, both at runtime and via the manifest JSON Schema, instead of a silent no-op.metadata.mdx docs and adds a changelog post explaining the new global scope.metadata: block (e.g. in _defaults.yaml) was previously accepted by the schema but never applied — metadata.labels/metadata.tags/etc. set there silently did nothing, which is worse than an error, since users had no signal their config wasn't taking effect.metadata block into each component definition instead of declaring it once.<!-- This is an auto-generated comment: release notes by coderabbit.ai -->
New Features
metadata: defaults that are deep-merged into each component’s metadata.metadata.inherits chain (when enabled) → component-local metadata.Validation & Tests
metadata (only labels, tags, custom, enabled, locked, terraform_workspace_pattern).Documentation
<details> <summary>Clarify Atmos CI concurrency guidance @osterman (#2798)</summary>
<!-- This is an auto-generated comment: release notes by coderabbit.ai -->
concurrency behavior for Atmos/Terraform runs, including in-progress vs pending handling and pending-run eviction.cancel-in-progress: true can cancel an in-flight Terraform apply and potentially leave remote state locks requiring manual recovery.queue: max limits (up to 100 pending) and noted it can’t be combined with cancel-in-progress.Nothing published for this version
Nothing published for this version
…keeps this PR mergeable without a separate open vulnerability.
<details> <summary>fix(ci): publish Docker images for prereleases @osterman (#2792)</summary>
<!-- This is an auto-generated comment: release notes by coderabbit.ai -->
<!-- end of auto-generated comment: release notes by coderabbit.ai --> </details>
<details> <summary>fix(hooks): resolve component path from metadata.component target @osterman (#2802)</summary>
infracost, trivy, checkov, kics, tflint, kind: command, kind: git, kind: step/kind: steps) now resolve $ATMOS_COMPONENT_PATH — and the hook subprocess's own working directory — from the component's resolved metadata.component target or provisioned workdir, instead of the stack-facing component/alias name.hooks.ComponentPath(), extended to cover every provisionable component type (terraform/helmfile/packer/ansible/kubernetes/helm).kind: git hooks with no repository: now use the resolved component directory (not the repo root) as their Git workdir, so commit.paths are component-relative like other hook kinds.kind: step/kind: steps hooks default working_directory to the resolved component path when not explicitly set — except type: atmos steps, which re-invoke the atmos binary itself and must keep inheriting the ambient working directory so the nested run resolves the project's atmos.yaml/stacks, not the target component's own directory.source: and hooks configured now has its source provisioned before those hooks fire (before.terraform.* is a "run before Terraform" event, not a "run before provisioning" event) — except before.terraform.init, which is the source provisioner's own lifecycle event and must keep running first. This only triggers for components that actually have hooks; every terraform subcommand already provisions its own component independently in RunE, so gating on hooks.HasHooks() avoids a second, racing provisioning attempt for components with no hooks at all.when: always after-hooks that must still fire and report the failure.vendorToTarget created the target directory before downloading, so a failed download left an empty component directory — enough for path resolution to treat the component as existing — and a partial copy failure would leave a non-empty one that needsProvisioning treats as fully provisioned, silently skipping re-provisioning on the next run. The provisioner now removes the directory only when this attempt created it; a pre-existing directory is left untouched (covered by positive/negative unit tests in pkg/provisioner/source).--all/--affected/--components/--query/--tags/--labels) no longer fail with "component is required": the metadata-resolution step is skipped for the global before/after hook (no single component is resolved yet at that point) and is best-effort everywhere else, so an intentionally invalid stack/component test scenario is rejected by Terraform's own validation with its normal message, not by hook-context preparation with a different one.docs/fixes/2026-07-24-hook-component-path-metadata-component.md) and updated the custom-hooks, git-ops, and hooks-step-types PRDs to document all of the above.TestComponentPathFor_Fallbacks) surfaced by this change: a hand-rolled filepath.Separator-prefixed "absolute" path isn't actually absolute per Go's filepath.IsAbs on Windows, so u.GetComponentPath's Abs() call legitimately expanded it with the current drive letter, breaking the test's literal expectation. Rooted the fixture paths in a genuine t.TempDir() instead.docs/prd/git-ops.md that failed editorconfig-checker's indent_size=2 rule once this PR made the file "affected" by CI's affected-file validation.github.com/google/cel-go v0.26.0 → v0.29.0 (GHSA-gcjh-h69q-9w9g), and the website's transitive postcss pin bumped from ^8.5.10 to ^8.5.18 (resolves to 8.5.23, GHSA-r28c-9q8g-f849). NOTICE and website/pnpm-lock.yaml regenerated accordingly.metadata.component: nat-gateway on a component named e.g. nat-gateway-alias), hooks pointed at a directory that doesn't exist on disk (.../nat-gateway-alias instead of .../nat-gateway). For kind: infracost this failed silently — it just logged "Could not autodetect any projects" and reported $0.00, looking like a normal (if boring) result instead of a broken cost-estimation run. atmos describe component already resolved metadata.component correctly; hooks just weren't sharing that resolution.type: atmos steps, before.terraform.init's own provisioner event, and when: always fallback hooks on early failures — each of which needed its own narrow fix to keep working the way it did before this PR, without losing the original fix.postcss (arbitrary .map file disclosure) and a medium-severity private-field-exposure issue in cel-go, both flagged by Dependabot on this branch; folding them in here keeps this PR mergeable without a separate open vulnerability.metadata.component path resolution, fully subsumed by this PRsource.uri code path: #2364 / #2371, #2134 / #2137, #2309 (Bug 2), #2684 — this PR covers the plain/local metadata.component branch they didn't reach<!-- This is an auto-generated comment: release notes by coderabbit.ai -->
ATMOS_COMPONENT_PATH now matches that resolved location.source: are provisioned (best-effort) before applicable before.terraform.* hooks; provisioning failures are non-blocking.working_directory to the component directory while preserving explicit values and ambient behavior for Atmos steps.source: provisioning, and working-directory defaults.
<!-- end of auto-generated comment: release notes by coderabbit.ai --></details>
<details> <summary>fix(install): self-elevate install.sh via sudo on Linux @osterman (#2806)</summary>
maybe_sudo helper in website/static/install.sh and routed every root-requiring command in the apt/yum/apk (install_via_cloudsmith) path through it, so the script self-elevates via sudo when not already root, and fails with a clear actionable message (pointing at the rootless binary install) if no sudo is available..github/workflows/install-sh-smoke.yml that exercises the package install method on ubuntu-latest, to catch regressions of this class of bug going forward.website/docs/quick-start/install-atmos.mdx: a stray Unicode box-drawing character (│) had silently replaced the literal pipe (|) in the Debian/Ubuntu, RedHat/CentOS, and Alpine manual install snippets, breaking them if copy-pasted; also added sudo to the Debian/Ubuntu tab's commands to match the RedHat tab's existing convention.curl -fsSL https://atmos.tools/install.sh | bash — the install command shown on the homepage — fails on a plain (non-root) Ubuntu 24.04 shell with Permission denied writing to /usr/share/keyrings/ and /etc/apt/sources.list.d/, since the apt/yum/apk install path never elevated privileges.maybe_sudo is never invoked from those functions.<!-- This is an auto-generated comment: release notes by coderabbit.ai -->
sudo when available, with clearer guidance when neither root nor sudo exists.atmos executable via command -v (and normalizing its path).sudo usage and version selection).install.sh smoke workflow to run an additional package-mode install and verify atmos is present in PATH on ubuntu-latest.
<!-- end of auto-generated comment: release notes by coderabbit.ai -->
</details><details> <summary>fix: restore auth for evaluated list commands @osterman (#2801)</summary>
--identity=false.atmos.Component() forwards an inherited context to an auth-less nested target.<!-- This is an auto-generated comment: release notes by coderabbit.ai -->
describe stacks and list so credentials are prepared only when template processing / YAML-function evaluation is enabled, or when identity is explicitly selected (while respecting the disabled identity sentinel).AuthContext inheritance for nested template/YAML evaluations.<details> <summary>feat(auth): improve AWS user webflow and console access @osterman (#2786)</summary>
aws/user identities.atmos auth login --webflow to force a fresh browser OAuth2/PKCE sign-in for a direct aws/user identity without reading, deleting, or overwriting configured IAM access keys.webflow_enabled: false and clear validation errors for provider mode or non-aws/user identities.IAM-user identities can obtain temporary STS credentials through Atmos browser authentication, but they could not open the AWS console and could not explicitly choose browser authentication when saved IAM credentials existed. This change makes both flows available while keeping long-lived credentials intact.
actionlint, and repository pre-commit checks.<!-- This is an auto-generated comment: release notes by coderabbit.ai -->
New Features
--webflow to force a fresh browser login for AWS IAM user identities.Bug Fixes
Documentation
<!-- end of auto-generated comment: release notes by coderabbit.ai --> </details>
<details> <summary>fix(secrets): store SSM scalar secrets without JSON quotes @osterman (#2797)</summary>
secret: true AWS SSM stores without JSON quoting.SecureString value rather than a JSON-wrapped scalar.<!-- This is an auto-generated comment: release notes by coderabbit.ai -->
Bug Fixes
null text) without unwanted formatting or type changes.GetKey).Documentation
prefix must start with /.secret: true read/write behavior and related decoding ambiguity.Tests
terraform and tofu.
<!-- end of auto-generated comment: release notes by coderabbit.ai -->
</details><details> <summary>fix(terraform): propagate provider cache to output lookups @osterman (#2791)</summary>
!terraform.output and atmos.Component() Terraform runsterraform init calls that share a provider cacheThe normal Terraform command pipeline synthesized TF_PLUGIN_CACHE_DIR, but the internal output executor created a separate subprocess environment without that automatic cache configuration. Each lookup therefore initialized providers in its component workdir unless CI injected the cache manually.
go test ./...scripts/run-custom-golangci-lint.shgo-fumpt, Go build, custom lint)<!-- This is an auto-generated comment: release notes by coderabbit.ai -->
<details> <summary>fix(ci): serialize verifier lifecycle and color errors @osterman (#2794)</summary>
NO_COLOR behavior.cosign while another worker was executing it, causing Linux ETXTBSY failures.CI signal even though the rest of Atmos recognizes CI as ANSI-capable.<!-- This is an auto-generated comment: release notes by coderabbit.ai -->
NO_COLOR.<details> <summary>Add deprecation validation warnings @osterman (#2793)</summary>
<!-- This is an auto-generated comment: release notes by coderabbit.ai -->
depends_on replacement guidance.depends_on handling for legacy manifest layouts.<details> <summary>fix(yaml): decode JSON after whitespace-trimmed function tags @osterman (#2780)</summary>
[ and { as valid boundaries after YAML function tags.!template list and map output.docs/fixes.!template[...] or !template{...}, which strict tag matching left as literal strings.<!-- This is an auto-generated comment: release notes by coderabbit.ai -->
!template parsing so whitespace-trimmed templates that render JSON arrays/objects decode correctly.!template boundary handling around whitespace, JSON delimiters, end-of-input, and zero-argument tags, preventing accidental literal-string output.Nothing published for this version
Nothing published for this version
Nothing published for this version
fix(secrets): store SSM scalar secrets without JSON quotes @osterman
<details> <summary>fix(secrets): store SSM scalar secrets without JSON quotes @osterman (#2797)</summary>
secret: true AWS SSM stores without JSON quoting.SecureString value rather than a JSON-wrapped scalar.<!-- This is an auto-generated comment: release notes by coderabbit.ai -->
Bug Fixes
null text) without unwanted formatting or type changes.GetKey).Documentation
prefix must start with /.secret: true read/write behavior and related decoding ambiguity.Tests
terraform and tofu.
<!-- end of auto-generated comment: release notes by coderabbit.ai -->
</details><details> <summary>fix(terraform): propagate provider cache to output lookups @osterman (#2791)</summary>
!terraform.output and atmos.Component() Terraform runsterraform init calls that share a provider cacheThe normal Terraform command pipeline synthesized TF_PLUGIN_CACHE_DIR, but the internal output executor created a separate subprocess environment without that automatic cache configuration. Each lookup therefore initialized providers in its component workdir unless CI injected the cache manually.
go test ./...scripts/run-custom-golangci-lint.shgo-fumpt, Go build, custom lint)<!-- This is an auto-generated comment: release notes by coderabbit.ai -->
<details> <summary>fix(ci): serialize verifier lifecycle and color errors @osterman (#2794)</summary>
NO_COLOR behavior.cosign while another worker was executing it, causing Linux ETXTBSY failures.CI signal even though the rest of Atmos recognizes CI as ANSI-capable.<!-- This is an auto-generated comment: release notes by coderabbit.ai -->
NO_COLOR.Nothing published for this version
One column per quarter.