The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill defines a workflow for processing user-specified cloud management tasks, which serves as a potential ingestion point for untrusted data. Although the instructions advise clarifying goals, the lack of explicit sanitization or strong boundary markers presents a surface where malicious instructions embedded in user tasks could influence tool behavior, particularly when used with agents that have file and command execution capabilities like 'claude-code'.
[METADATA_POISONING]: The skill's instructions and resources section reference multiple external files and assets, such as 'references/tagging-standards.md' and 'assets/cost-analysis-template.xlsx', which are not included in the file list defined in the skill's metadata. This inconsistency between the documentation and the actual package contents is deceptive to users and automated agents.
[SAFE]: The provided Terraform and HCL code snippets for resource tagging, lifecycle management, and budget monitoring are standard industry patterns for cost optimization and do not contain any malicious execution, data exfiltration patterns, or obfuscation.

cost-optimization