The Agent Skills Directory

[SAFE]: The skill provides architectural guidance and Infrastructure as Code (IaC) templates while explicitly instructing the AI to follow security best practices. This includes using managed secret storage (AWS Secrets Manager, Azure Key Vault), enforcing IAM least privilege (avoiding wildcard policies), and securing network configurations (avoiding default VPCs and open inbound rules).
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted user data regarding cloud requirements to generate architectural designs and code, creating a potential surface for indirect prompt injection.
Ingestion points: User-provided inputs describing cloud provider, workload, scale, and compliance requirements in SKILL.md.
Boundary markers: The instructions do not define specific delimiters to separate user input from the agent's internal logic.
Capability inventory: The skill is limited to generating text and code snippets (Terraform, CDK, Bicep); it lacks tools for autonomous file system modification, shell execution, or network operations.
Sanitization: There are no instructions for sanitizing, escaping, or validating user-supplied data before it is processed by the AI.
[NO_CODE]: The skill consists entirely of markdown documentation and configuration instructions. It does not include any executable scripts, scripts that fetch remote content, or binary files.

cloud-architect