The Agent Skills Directory

[SAFE]: No malicious patterns or security risks were detected. The skill provides high-quality guidance for Infrastructure as Code (IaC) using Terraform, OpenTofu, and Docker Compose.
[DATA_EXPOSURE_AND_EXFILTRATION]: The skill strongly emphasizes secret management best practices. It explicitly warns against committing credentials and hardcoding secrets. Placeholders in template files (e.g., REPLACE_WITH_SECURE_SECRET) are used correctly for local development examples.
[EXTERNAL_DOWNLOADS]: The skill references several well-known and trusted external security tools and repositories, such as Gitleaks, Checkov, and Anton Babenko's Terraform pre-commit hooks. These downloads are performed via the pre-commit framework, a standard development practice for security linting.
[INDIRECT_PROMPT_INJECTION]: The skill handles user requirements for infrastructure generation. It mitigates injection risks by enforcing structured workflows (ASSESS -> DESIGN -> IMPLEMENT) and recommending strict validation schemas (Zod, JSON Schema) for environment variables and manifests.
[DYNAMIC_EXECUTION]: Code generation (HCL, YAML) is handled via templates and structured logic rather than unsafe evaluation. The skill includes specific anti-patterns for insecure HCL and Docker configurations.

scaffold