The Agent Skills Directory

[SAFE]: The skill documents performance benchmarks and strategy selection for the Redactive library, focusing on providing technical guidance to users.
[SAFE]: Includes explicit security best practices, such as recommending 'safe mode' over 'codegen mode' when dealing with untrusted or user-supplied policies or strict Content Security Policy (CSP) environments.
[COMMAND_EXECUTION]: Mentions standard development commands for running benchmarks (e.g., pnpm benchmark), which is standard practice for software development and performance analysis.
[SAFE]: Describes an architectural feature using new Function for performance optimization in 'codegen' mode, but correctly identifies this as a trust-sensitive feature and provides a non-dynamic 'safe' alternative for general use.

redactive-performance