OpenSkillIndex← back to search
claude-code

Claude-skill-registry atmos-functions

Cross-Component Dependencies with Atmos Functions

install
source · Clone the upstream repo
git clone https://github.com/majiayu000/claude-skill-registry
Claude Code · Install into ~/.claude/skills/
T=$(mktemp -d) && git clone --depth=1 https://github.com/majiayu000/claude-skill-registry "$T" && mkdir -p ~/.claude/skills && cp -r "$T/skills/data/atmos-functions" ~/.claude/skills/majiayu000-claude-skill-registry-atmos-functions && rm -rf "$T"
manifest: skills/data/atmos-functions/SKILL.md
source content

Cross-Component Dependencies with Atmos Functions

Preferred approach: Atmos Functions - Use Atmos template functions in stack YAML to pass values between components at plan/apply time. This avoids Terraform remote state lookups entirely.

!terraform.state
Syntax

Two-parameter form (current stack):

!terraform.state <component> <output>

Three-parameter form (specific stack):

!terraform.state <component> <stack> <output>

Examples

components:
  terraform:
    my-component:
      vars:
        # Current stack lookups (component output)
        vpc_id: !terraform.state vpc vpc_id
        subnet_ids: !terraform.state vpc private_subnet_ids

        # Cross-stack lookup (component stack output)
        grafana_role_arn: !terraform.state grafana core-use2-auto workspace_iam_role_arn

        # Nested output with YQ expression
        role_arn: !terraform.state iam-role/my-role plat-use2-dev .role.arn


!terraform.output
Alternative

There's also 

!terraform.output
which has the same syntax but different behavior:

# Same syntax as !terraform.state
vpc_id: !terraform.output vpc vpc_id

Key differences:

Aspect
!terraform.state
(preferred)
!terraform.output
How it worksReads directly from S3 state bucketRuns 
terraform init
+ 
terraform output
SpeedFast (direct S3 read)Slow (full Terraform initialization)
Use caseDefault choiceWhen state file format differs

Always prefer 

!terraform.state
- it's significantly faster because it reads the state file directly from S3 without running Terraform commands.

Best Practices

  • Use 
    !terraform.state
    over 
    !terraform.output
     - Direct S3 access is much faster than running Terraform.
  • Define lookups in catalog defaults - Put 
    !terraform.state
    expressions in 
    stacks/catalog/<component>/defaults.yaml
    rather than in stack files. The function automatically resolves based on the current stack context.
  • Use current-stack lookups when possible - Omit the stack parameter to look up components in the same stack, making configs more portable.
  • Cross-stack lookups for shared resources - Use the three-parameter form when referencing centralized resources (e.g., Grafana in core-auto from plat accounts).

Legacy Approach (Being Phased Out)

Some older components still use Cloud Posse's 

remote-state
Terraform module with 
remote-state.tf
files. This pattern is being phased out in favor of Atmos functions. The tfstate backend is in the core-auto account.

When you encounter 

remote-state.tf
files, prefer converting them to 
!terraform.state
expressions in the catalog.