The Agent Skills Directory

[DATA_EXFILTRATION]: The skill's reference documentation (references/terraform_best_practices.md and references/provider_examples.md) includes Terraform HCL examples that utilize the file() function to reference sensitive local file paths, specifically ~/.ssh/id_rsa and ~/.ssh/id_rsa.pub. These are provided as common templates for configuring SSH-based remote provisioners and virtual machine authentication.
[COMMAND_EXECUTION]: Several shell scripts (scripts/run_ci_checks.sh, scripts/check_feature_version_consistency.sh) and a regression test suite (tests/test_feature_version_consistency.sh) are included in the repository. These scripts facilitate internal maintenance by using standard Unix utilities like awk, grep, and sed to verify documentation and template consistency.
[INDIRECT_PROMPT_INJECTION]: The skill operates by transforming user-supplied infrastructure requirements into generated HCL code, which is then passed to an external skill (devops-skills:terraform-validator) for syntax and security validation. This process involves a surface where untrusted user input could potentially influence the output processed by the downstream validator.
Ingestion points: User infrastructure requirements captured during Step 1 of the Core Workflow in SKILL.md.
Boundary markers: The skill implements a strict checklist and references internal best-practice guidelines to constrain code generation.
Capability inventory: The skill possesses the capabilities to write HCL files and invoke the devops-skills:terraform-validator tool.
Sanitization: The skill explicitly enforces security best practices, including encryption for RDS and S3, deletion protection for stateful resources, and the use of Terraform's sensitive and ephemeral features for secret handling.

terraform-generator