The Agent Skills Directory

[SAFE]: The skill implements a robust multi-agent orchestration framework where each component (scanner, extractor, stripper, injector, compressor, and critic) has a single, well-defined focus area. The instructions are transparent and consistent with the primary purpose of linguistic refinement.
[SAFE]: Supporting TypeScript (Bun) scripts used for artifact management and indexing include security best practices, such as verifying that operations occur within safe directories and explicitly refusing to follow symlinks.
[PROMPT_INJECTION]: The skill ingests user-provided text for humanization, creating a surface for indirect prompt injection. However, the risk is mitigated by the highly constrained nature of the sub-agent tasks, which are limited to editing and linguistic analysis rather than general-purpose execution. This surface is necessary for the skill's primary function and shows no evidence of exploitation.

humanmaxxing