The Agent Skills Directory

[SAFE]: The skill provides industry-standard security practices, including parameterized queries for SQL injection prevention, Zod/Pydantic schemas for input validation, and secure HTTP header configurations.
[PROMPT_INJECTION]: No attempts to override agent behavior, bypass safety guardrails, or extract system prompts were detected in the instructions or metadata.
[DATA_EXFILTRATION]: There are no commands or code patterns that attempt to access sensitive files (e.g., SSH keys, AWS credentials) or exfiltrate data to external domains.
[CREDENTIALS_UNSAFE]: No hardcoded API keys, tokens, or passwords are present. The skill correctly instructs users to manage secrets through environment variables.
[REMOTE_CODE_EXECUTION]: The skill does not download or execute remote scripts. All code references are static templates for application development using well-known, safe package registries.
[COMMAND_EXECUTION]: No dangerous shell commands or privilege escalation attempts were found. Terminal examples are limited to standard development workflows like 'npm run build' or 'docker build'.

fullstack-guardian