The Agent Skills Directory

[SAFE]: No malicious patterns or security vulnerabilities were detected in the skill instructions or metadata.- [SAFE]: The skill includes explicit safety boundaries, specifically instructing the agent not to misrepresent authorship, bypass academic/compliance reviews, or promise detector-proof output.- [SAFE]: External references and documentation links point to well-known and reputable sources (such as OpenAI, Stanford HAI, and Digital.gov) to provide context on the limitations of AI detection, rather than to download executable code.- [SAFE]: The skill uses a limited set of tools (Read, Grep, Edit) appropriate for its stated purpose of prose editing and documentation improvement.- [SAFE]: Although the skill processes untrusted input (PR descriptions, documentation, and user text), it focuses on linguistic quality and rhythm rather than executing code or performing sensitive operations, minimizing the risk associated with indirect prompt injection.

writing-humanizer