The Agent Skills Directory

[CREDENTIALS_UNSAFE]: The skill includes multiple code snippets that demonstrate insecure coding practices to provide educational contrast. These snippets contain hardcoded strings representing example database credentials and API keys. Examples include 'postgresql://admin:SuperSecret123@db.internal.acme.com/prod' in rules/system-prompt-leakage.md and 'sk-abc123secretkey456' in rules/sensitive-disclosure.md and AGENTS.md. These are considered safe in context because they are explicitly part of 'Vulnerable' code blocks intended for learning and use obviously generic or placeholder values.

llm-security