The Agent Skills Directory

[CREDENTIALS_UNSAFE]: The skill explicitly instructs against hardcoding secrets or committing them to version control. It promotes the use of sst secret set for encrypted storage in S3 and runtime access via resource linking. Examples use benign placeholders like pk_live_xxx and sk_live_xxx.
[COMMAND_EXECUTION]: The skill uses standard CLI commands for the SST framework (sst dev, sst deploy, sst remove). These are well-known tools for AWS infrastructure management. The GitHub Actions examples follow industry standards using official actions and OIDC authentication.
[REMOTE_CODE_EXECUTION]: No suspicious remote code execution patterns were found. The 'live Lambda dev' mode described is a core feature of the SST framework that proxies invocations to a local machine for faster development, rather than an exploit.
[DATA_EXFILTRATION]: The skill focuses on secure resource management within AWS. There are no patterns that suggest exfiltration of sensitive data to unauthorized external domains. Resource linking ensures that IAM permissions are scoped only to the linked components.
[PROMPT_INJECTION]: The instructions and critical requirements use clear, instructional language to guide the agent toward safe infrastructure patterns. No attempts to bypass safety filters or override system prompts were detected.
[OBFUSCATION]: All code examples and instructions are in plain, readable TypeScript and Markdown. No Base64, zero-width characters, or other obfuscation techniques were used.

infra-iac-sst