The Agent Skills Directory

[CREDENTIALS_UNSAFE]: The skill references sensitive file paths (e.g., ~/.aws/credentials, service-account-key.json) and configuration keys (e.g., client_secret) within documentation blocks to illustrate standard cloud provider authentication methods. It also includes an anti-example containing a dummy password to explicitly warn against hardcoding secrets.
[EXTERNAL_DOWNLOADS]: The documentation mentions community-standard tools and plugins for infrastructure linting and pre-commit checks, such as tflint-ruleset-aws and pre-commit-terraform. These are referenced as part of a recommended developer environment and originate from well-known software repositories.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to process user requirements into infrastructure code, which presents a surface for instructions embedded in external data.
Ingestion points: Infrastructure requirements and existing HCL code provided by the user.
Boundary markers: The workflow mandates terraform validate and tflint checks as explicit boundaries before execution.
Capability inventory: The skill involves resource creation via terraform apply and state manipulation via terraform state commands.
Sanitization: The skill emphasizes input validation blocks, plan review phases, and linter enforcement to mitigate risks from malformed or malicious input.
[DYNAMIC_EXECUTION]: The skill provides documentation for executing HCL via Terraform and mentions testing frameworks like Terratest (Go-based) and OPA (Open Policy Agent) for policy enforcement, which are standard components of modern infrastructure pipelines.

terraform-engineer