The Agent Skills Directory

[SAFE]: The skill functions as a technical reference and documentation for a sandboxed scripting environment. It provides guidance on data transformation and analysis without introducing malicious instructions or safety bypasses.- [COMMAND_EXECUTION]: The documentation defines a query() function for SQL execution. It explicitly instructs the agent to use parameterized queries to prevent SQL injection, which is a key security best practice.- [EXTERNAL_DOWNLOADS]: The fetch() function is documented for performing HTTP requests. The examples use placeholder domains and generic headers, and the text includes instructions on how to safely handle and validate external API responses.- [PROMPT_INJECTION]: The skill includes operational rules that require the agent to seek user confirmation when data structures are ambiguous, preventing autonomous actions based on potentially untrusted or confusing input.

datadata-dql