The Agent Skills Directory

[SAFE]: The skill implements robust security boundaries, explicitly instructing the agent never to request, store, or log sensitive information such as database passwords or access tokens. It favors interactive prompts and environment variables to handle credentials securely.- [SAFE]: Extensive human-in-the-loop requirements are defined for all destructive, schema-mutating, or secret-producing operations, ensuring the user maintains control over critical project changes.- [SAFE]: The provided Python script (scripts/prepare_import.py) is a local utility for data conversion that performs no network operations and includes checks to prevent accidental file overwriting. It uses standard library modules for processing CSV, JSON, and delimited files safely.- [SAFE]: Operational instructions are scoped to vendor-provided tools (CLI and MCP) and official endpoints (polygres.com), with clear guidelines on handling timeouts and errors without bypassing security protocols.

polygres-cli