Token导航 LogoToken导航TokenDH.com
研究检索敏感数据github未标认证来源可访问许可证需确认审计提醒

atmos-stacks大气堆栈

Agent Skill

atmos-stacks 用于查找、检索和筛选相关信息,适合在 Codex、Claude、Cursor、Gemini CLI 中需要根据关键词、任务场景或来源线索快速定位候选结果时使用。可结合来源仓库、安装命令和原始 README 继续核验具体用法。安装前建议确认权限范围、维护状态,以及是否会触发联网、命令执行或文件读写。

总安装

225

周安装

9

GitHub Stars

1,319

下载量

73
CodexClaudeCursorGemini CLI

安装说明

本站只整理中文说明和来源信息,不托管安装包,也不代用户安装。

GitHub

来源数

2

许可证

unknown

最后核验

2026-05-01

来源状态

来源可访问

安装方式

通过对话安装

复制提示词发给支持本地命令或 Skills 的 AI 助手,先确认命令和权限,再让它执行。

请帮我安装这个 Agent Skill:atmos-stacks(大气堆栈)
来源仓库:https://github.com/cloudposse/atmos
仓库路径:skills/atmos-stacks
安装命令:
npx skills add https://github.com/cloudposse/atmos --skill atmos-stacks
安装前请先检查当前环境是否支持对应 CLI,并向我确认将要执行的命令、安装目录、联网范围和文件读写权限;确认后再执行。

命令行安装

复制命令到本机终端执行。该命令会通过 npx skills 从第三方来源获取 Skill;本站只展示命令,不托管安装包,也不自动执行。

skills.shnpx skills
npx skills add https://github.com/cloudposse/atmos --skill atmos-stacks

简介

atmos-stacks 管理 YAML 格式的栈配置文件,分离基础设施代码与环境设置,支持跨环境复用组件。

  • 适用于需要按组织、租户、账号、区域、阶段组合定义部署目标的场景。
  • 基于 atmos.yaml 的发现规则加载栈清单,并通过深度合并生成最终配置。
  • 使用前应验证 included_paths 设置,确保不会意外包含敏感或无关目录。
  • 适用宿主包括 Codex、Claude、Cursor、Gemini CLI,接入前应确认版本、权限和运行环境要求。

SKILL.md

Atmos Stack Configuration

Stacks are YAML configuration files that define which components to deploy, with what settings, and how they relate to each other. They separate configuration from infrastructure code, enabling the same Terraform root modules to be deployed across many environments with different settings.

What Stacks Are

A stack manifest is a YAML file that declares components and their configuration for a specific combination of organization, tenant, account, region, and stage. Atmos discovers stack manifests based on included_paths and excluded_paths in atmos.yaml, then deep-merges all imported configurations to produce the final resolved state for each component.

Stacks are not Terraform workspaces, although Atmos derives workspace names from stack names. A single stack manifest can configure multiple components, and a single component can appear across many stacks with different variable values.

Stack Discovery

Atmos discovers stack manifests based on included_paths and excluded_paths configured in the stacks section of atmos.yaml. For the complete atmos.yaml configuration reference, see the atmos-config skill.

Stack Name Precedence

Atmos resolves the stack name using this priority (highest first):

  1. name field in the stack manifest (explicit override).
  2. name_template in atmos.yaml (Go template).
  3. name_pattern in atmos.yaml (token pattern).
  4. File basename (e.g., prod.yaml becomes prod).

Stack Manifest Structure

A stack manifest can contain the following top-level sections:

# Optional: explicit stack name override
name: "plat-ue2-prod"

# Import other configurations
import:
  - catalog/vpc/defaults
  - mixins/region/us-east-2
  - orgs/acme/plat/prod/_defaults

# Global-scope sections (apply to all components)
vars: {}
locals: {}
env: {}
settings: {}
hooks: {}
overrides: {}

# Component-type scope (apply to all components of that type)
terraform:
  vars: {}
  env: {}
  settings: {}
  hooks: {}
  backend_type: s3
  backend: {}
  providers: {}
  command: terraform
  overrides: {}

helmfile:
  vars: {}
  env: {}
  settings: {}
  hooks: {}
  command: helmfile
  overrides: {}

# Component definitions
components:
  terraform:
    vpc:
      metadata: {}
      vars: {}
      env: {}
      settings: {}
      hooks: {}
      backend_type: s3
      backend: {}
      providers: {}
      command: terraform
      auth: {}
  helmfile:
    echo-server:
      vars: {}
      env: {}
      settings: {}

Configuration Sections Reference

vars

Variables passed as inputs to Terraform, Helmfile, or Packer components. Defined at global, component-type, or component level. Deep-merged across all levels with component-level values taking precedence.

vars:
  environment: prod
  region: us-east-1
  tags:
    Environment: Production
    ManagedBy: Atmos

Maps are recursively merged; lists are replaced (not appended).

locals

File-scoped temporary variables for reducing repetition within a single YAML file. Locals do NOT inherit across file imports. They can reference each other using {{.locals.name}} syntax with automatic dependency resolution.

locals:
  namespace: acme
  name_prefix: "{{ .locals.namespace }}-{{ .vars.stage }}"

components:
  terraform:
    vpc:
      vars:
        name: "{{ .locals.name_prefix }}-vpc"

Locals can also access settings, vars, and env defined in the same file using {{.settings.key}}, {{.vars.key}}, and {{.env.KEY}}.

env

Environment variables set when executing components. Simple key-value pairs merged shallowly across levels.

env:
  AWS_PROFILE: acme-prod
  TF_IN_AUTOMATION: "true"

settings

Integration metadata and configuration not passed to Terraform. Used for Spacelift, Atlantis, validation, and other Atmos integrations.

settings:
  spacelift:
    workspace_enabled: true
    autodeploy: false
  depends_on:
    - vpc

metadata

Component-only section that controls Atmos behavior for that component. Cannot be used at global or component-type level.

components:
  terraform:
    vpc:
      metadata:
        component: vpc           # Terraform root module path
        inherits:
          - vpc/defaults         # Inheritance chain
        type: abstract           # abstract or real (default)
        enabled: true            # Enable/disable component
        locked: false            # Prevent modifications
        terraform_workspace: "custom-ws"
        custom:
          owner: platform-team

hooks

Lifecycle event handlers that execute actions at specific points (e.g., after terraform apply).

hooks:
  store-outputs:
    events:
      - after-terraform-apply
    command: store
    name: prod/ssm
    outputs:
      vpc_id: .vpc_id

command

Override the executable for a component type or specific component. Useful for OpenTofu, custom wrappers, or version-pinned binaries.

terraform:
  command: tofu

backend

Terraform backend configuration. Atmos generates backend.tf.json automatically.

terraform:
  backend_type: s3
  backend:
    s3:
      bucket: acme-ue1-root-tfstate
      region: us-east-1
      encrypt: true
      use_lockfile: true

providers

Terraform provider configuration. Atmos generates providers_override.tf.json automatically.

terraform:
  providers:
    aws:
      region: us-east-1
      assume_role:
        role_arn: "arn:aws:iam::{{ .vars.account_id }}:role/TerraformRole"

auth

Authentication configuration for cloud providers. Primarily defined in atmos.yaml but can be referenced at component level.

components:
  terraform:
    vpc:
      auth:
        identity: prod-admin

overrides

Scoped overrides that apply only to components defined in the current manifest and its imports (not to all components in the top-level stack). This is different from regular vars/env/settings which affect all components.

overrides:
  env:
    TEST_ENV_VAR: "overridden-value"
  vars:
    custom_tag: override
  settings:
    spacelift:
      autodeploy: true

Deep-Merge Behavior and Override Precedence

Atmos deep-merges configuration from multiple levels. The precedence order (lowest to highest priority):

  1. Global scope (vars:, env:, settings:)
  2. Component-type scope (terraform.vars:, helmfile.env:)
  3. Base component defaults (via metadata.inherits, in list order)
  4. Component-level scope (components.terraform.<name>.vars:)
  5. Overrides (overrides:, terraform.overrides:)

For maps, keys are recursively merged with higher-priority values overriding lower-priority ones. For lists, the entire list at higher priority replaces the lower-priority list (lists are not appended).

The _defaults.yaml Pattern

A common convention is to use _defaults.yaml files at each level of the directory hierarchy:

stacks/
  orgs/
    acme/
      _defaults.yaml            # Organization-wide defaults
      plat/
        _defaults.yaml          # Tenant defaults
        dev/
          _defaults.yaml        # Stage defaults
          us-east-2.yaml        # Top-level stack (deployable)
          us-west-2.yaml
        prod/
          _defaults.yaml
          us-east-2.yaml
          us-west-2.yaml

The underscore prefix ensures these files sort to the top of directory listings. They are excluded from stack discovery via excluded_paths and must be explicitly imported. Atmos has no special handling for _defaults.yaml -- it is purely a naming convention.

Each level imports the parent _defaults.yaml and adds its own defaults:

# stacks/orgs/acme/plat/prod/_defaults.yaml
import:
  - orgs/acme/plat/_defaults

vars:
  stage: prod
  tags:
    Environment: Production

Describing Stacks

Use atmos describe stacks to view the fully resolved configuration after all imports, inheritance, and overrides:

# View all stacks
atmos describe stacks

# Filter by stack
atmos describe stacks --stack plat-ue2-prod

# Filter by component and section
atmos describe stacks --components vpc --sections vars

# Output as JSON
atmos describe stacks --format json | jq '.["plat-ue2-prod"]'

Use atmos describe component for a single component:

atmos describe component vpc -s plat-ue2-prod

YAML Functions

Atmos provides YAML functions for dynamic value resolution at runtime:

  • !terraform.output <component> <output> -- Read Terraform outputs from another component.
  • !terraform.state <component> <path> -- Access Terraform state values.
  • !store <store-name> <component> <key> -- Read from external key-value stores (SSM, Vault, etc.).
  • !env <VAR_NAME> -- Read environment variables with optional defaults.
  • !exec <command> -- Execute shell commands and use output.
  • !include <path> -- Load content from external files.

Common Patterns and Best Practices

  1. Organize by org/tenant/stage/region: Structure stacks hierarchically so defaults cascade naturally through imports.
  2. Use catalog for component defaults: Place reusable component configurations in stacks/catalog/ and import them.
  3. Keep inheritance shallow: Limit to 2-3 levels of metadata.inherits to maintain readability.
  4. Use _defaults.yaml at every level: Define shared vars, env, and settings at the appropriate organizational level.
  5. Exclude non-deployable files: Configure excluded_paths to prevent catalog, mixin, and defaults files from being treated as top-level stacks.
  6. Prefer name_template over name_pattern: The Go template approach is more flexible and is the recommended method.
  7. Use atmos describe stacks liberally: Always verify the resolved configuration before applying changes.

References

适合场景

01

用户想查找某类 Agent Skill 时

02

需要根据任务场景推荐可安装能力包时

03

需要对比不同来源的安装命令和来源信息时

能力概览

能力 1

按任务关键词查找相关 Skills

能力 2

展示可复制的安装命令

能力 3

保留来源站点、仓库和原始说明,方便继续核验

能力 4

展示第三方安全扫描或审计结果

安装后应在对应宿主中按原始 README 的触发条件使用;具体调用方式请以来源页面和 README 为准。

平台分布

Codex

38.01%
按下载量换算28

Claude

31.11%
按下载量换算23

Cursor

17.49%
按下载量换算13

Gemini CLI

8.88%
按下载量换算6

安全审计

Gen Agent Trust Hub

通过

Socket

通过

Snyk

可疑

权限和风险

敏感数据

该 Skill 可能接触密钥、Token、环境变量或敏感配置,应进入高风险复核队列,默认不自动发布。

安装前确认

本站仅展示第三方公开信息,不托管安装包,不提供自动安装或运行环境。安装前应自行审查源码、依赖和命令行为。来源安全扫描存在 warning/failed 结果,不能写成本站确认安全。当前只有一个来源,正式发布前建议补源仓库或其他目录站核验。

来源信息

继续浏览同类 Skills