The Agent Skills Directory

[Dependency Management]: The skill relies on the awslabs.redshift-mcp-server package from a well-known organization. This is a standard and expected integration for providing Redshift connectivity through the Model Context Protocol (MCP).
[Indirect Prompt Injection Surface]: Because the skill processes query results and error messages from a database, it has an indirect prompt injection surface. The skill mitigates this through 'Core Rule 13,' which requires explicit HTML escaping of all untrusted values before they are included in reports, and 'Core Rule 10,' which ensures the agent waits for user confirmation before proceeding with data collection.
[Privilege Management and Least Privilege]: The skill operates under a strictly read-only model for database interactions, ensuring it cannot perform destructive operations like dropping tables or modifying data. It also explicitly forbids requesting user credentials, relying instead on pre-configured IAM roles provided by the platform.
[Operational Oversight]: Instructions mandate interactive execution by default, ensuring that all tool calls and data collection steps are visible to the user in the chat session, allowing for real-time monitoring and intervention.

redshift-support-specialist