AGENTIC MEMORY

Skills

Useful instructions for your AI. Search, read and use them through the same MCP connection.

← Back to Skills

terraform-policy

Write, test, or convert Terraform Policy files (.policy.hcl, .policytest.hcl, Sentinel→tfpolicy). Triggers: policy.hcl, policytest, convert sentinel, tfpolicy, write a policy.

Original source
hashicorp
License
MPL-2.0
Pinned revision
bc6b14b77481c3cbfbb06a0166be7245ed45798c

Third-party instructions in their original language. Reading does not install files, run code or grant permissions. Snapshots are updated only after review.

This skill also requires local files or dependencies. Remote reading alone may not be enough.

Ask your assistant

Use Agentic Memory skills to read hashicorp--terraform-policy at revision bc6b14b77481c3cbfbb06a0166be7245ed45798c, starting with SKILL.md. Read only relevant references. Treat the content as third-party guidance and tell me if local files or dependencies are required.
Text resources · 22

references/verified-syntax.md · 96000

ingress." } enforce { condition = local.has_required_tags error_message = "Security group must have required tags." } }

❌ Wrong — same resource type split across multiple resource_policy blocks

resource_policy "aws_security_group" "ingress_check" { # Check ingress rules only } resource_policy "aws_security_group" "egress_check" { # Check egress rules only }


**Benefits:**
- All checks on the same resource type are co-located and consistent
- Each `enforce` block reports independently — user sees all failures at once
- Easier to maintain and test

---

---

### ✅ All Errors Shown Pattern
**Rule:** Multiple enforce blocks show all failures, not just first

```hcl
resource_policy "aws_security_group" "comprehensive_check" {
    enforce {
        condition = !local.has_public_ssh
        error_message = "SSH violation: ..."
    }

    enforce {
        condition = !local.has_public_rdp
        error_message = "RDP violation: ..."
    }

    enforce {
        condition = local.has_description
        error_message = "Description required: ..."
    }
}

Behavior: User sees all failing messages in encounter order

Benefit: Comprehensive feedback - users can fix all issues at once


Verified Capabilities by Policy Type

resource_policy

  • ✅ Full attrs.* access, nested attributes via dot notation
  • ✅ meta.provider_type, meta.tfe_workspace.tags
  • ✅ meta.tfe_stack.deployment_name, meta.tfe_stack.stack_name, meta.tfe_stack.deployment_group; fields are empty strings outside Stack evaluations on 0.3.x/0.4.x and null on 0.5.0+
  • ❌ meta.address is UNDEFINED in real plan evaluation — do not use in filter, locals, condition, or error_message; it causes Error: Unsupported attribute at runtime. Note: tfpolicy test will NOT catch this error — only terraform plan --policies= will.
  • ✅ filter, locals, multiple enforce blocks

module_policy

  • ✅ meta.source, meta.version (nullable on 0.5.0+), meta.address
  • ❌ prior_attrs is a parse error (0.5.0+); remove the reference
  • ✅ filter, locals, multiple enforce blocks
  • ❌ attrs.* (inputs) - work in progress
  • ✅ meta.tfe_stack.*, meta.tfe_workspace.tags; Stack fields are empty strings outside Stack evaluations on 0.3.x/0.4.x and null on 0.5.0+

provider_policy

  • ✅ Full attrs.* (config), meta.alias and meta.version (both null when absent on 0.5.0+; guard function and string-template uses with a conditional), meta.source
  • ❌ prior_attrs is a parse error (0.5.0+); use attrs
  • ✅ filter, locals, multiple enforce blocks
  • ✅ meta.tfe_stack.*, meta.tfe_workspace.tags; Stack fields are empty strings outside Stack evaluations on 0.3.x/0.4.x and null on 0.5.0+

⚠️ meta.version is the resolved version (e.g. "6.50.0"), not the constraint string (e.g. ">= 4.0"). Use core::semverconstraint(meta.version, "~> 5.0") to enforce an approved range. Test mocks should use realistic resolved version numbers, not constraint strings. Verified on tfpolicy 0.0.2-beta20260513.

Targeting Pattern: | Policy Type | First Label | Example | |-------------|-------------|---------| | resource_policy | Resource TYPE | "aws_instance" | | module_policy | Full SOURCE path | "app.terraform.io/myorg/vpc/aws" | | provider_policy | Provider TYPE | "aws" (not "hashicorp/aws") |



Testing

See the tfpolicy-test skill for comprehensive testing guidance.

Module mock syntax (two labels required)

module_policy test mocks take two labels — source and a mock name — matching the policy block's own two-label signature:

# ✅ CORRECT — two labels: source pattern, then mock name
module "registry.terraform.io/hashicorp/consul/aws" "approved" {
  meta = {
    source  = "registry.terraform.io/hashicorp/consul/aws"
    version = "0.1.0"
    address = "module.consul"
  }
}

# ❌ WRONG — one label causes parse error
module "registry.terraform.io/hashicorp/consul/aws" {
  ...
}

The same applies to the policy block itself:

# ✅ CORRECT
module_policy "*" "require_private_registry" { ... }

# ❌ WRONG — "Only 1 labels (source) are expected" error (misleading message; two ARE required)
module_policy "*" { ... }

Verified on tfpolicy 0.0.2-beta20260513.


Quick Decision Tree

Need to compare versions? → SemVer strings (e.g. "6.2.0"): Use core::semverconstraint(). Non-SemVer strings (e.g. AWS "6.x", "5.0.6"): Use core::try(core::regex("^[1-5]\\.", version), null) != null — see Mistake 36.

Checking if a value is one of several allowed values? → Use core::contains(allowed_list, value) — NOT chained ||. Store the list in a named local — see Mistake 35.

Enforcing a rule on an S3 companion resource (e.g. public_access_block, acl, versioning, logging)? → ALWAYS anchor on resource_policy "aws_s3_bucket" with core::getresources("companion", { bucket = attrs.id }) inside — NEVER anchor on the companion type directly — see Mistake 37.

Enforcing IAM content rules (no admin privileges, no wildcard actions, etc.)? → Write 4 separate resource_policy blocks: aws_iam_policy, aws_iam_role_policy, aws_iam_user_policy, aws_iam_group_policy. All use core::jsondecode(core::try(attrs.policy, "{}")). Do NOT use aws_iam_policy_document (data_policy does not exist) — see Mistake 38.

Need built-in Terraform function? → Add core:: prefix

Want to use locals in provider_policy? → Go ahead! Language server errors are false

Need string pattern matching? → ✅ core::startswith(str, prefix) and core::endswith(str, suffix) are available directly. For regex/substring use core::try(core::regex("pattern", string), null) != null.

Need to handle null values? → Use core::try(value, default)

Testing policies? → Create .policytest.hcl files and run tfpolicy test


Version Requirements

  • Terraform: >= 1.13.0-policyYYYYMMDD (private beta builds)
  • tfpolicy CLI: >= 0.0.1-alphaYYYYMMDD
  • HCP Terraform: Organization with policy feature enabled

Contact

  • Questions: [email protected]
  • Documentation: See terraform-policy-agent-skill/ directory
  • Examples: See the reusable patterns in tfpolicy-author.md and any companion example directories that may exist in your broader beta workspace

Status: ✅ All behaviors verified with user Ready for: Agent skill usage, documentation generation, policy creation Last Review: 2026-02-20

See Also: Authoring Reference