The Agent Skills Directory

[PROMPT_INJECTION]: No evidence of prompt injection or instructions intended to override agent safety protocols was found.
[DATA_EXFILTRATION]: The skill does not attempt to access sensitive file paths or perform network operations to unauthorized domains. It demonstrates legitimate database and file operations for educational purposes.
[CREDENTIALS_UNSAFE]: No hardcoded credentials, API keys, or secrets are present. The examples use generic placeholders like dsn or query without embedding actual sensitive data.
[OBFUSCATION]: The content is provided in clear text with no evidence of Base64 encoding, zero-width characters, or homoglyph attacks.
[REMOTE_CODE_EXECUTION]: No remote code execution patterns, such as piping curl to bash, were detected.
[EXTERNAL_DOWNLOADS]: The skill mentions well-known, reputable Python libraries (psycopg, asyncpg, structlog) for database connectivity and logging, which are standard for the described tasks.
[DYNAMIC_CONTEXT_INJECTION]: No use of dynamic context injection syntax (!command) was found in the documentation.

python-resource-management