The Agent Skills Directory

[External Content Retrieval]: The skill fetches writing guidelines from an external source located at https://raw.githubusercontent.com/vercel-labs/writing-guidelines/main/command.md. This is a standard mechanism for ensuring the agent uses the most current rules. Because the repository belongs to a trusted organization, this retrieval is considered safe and appropriate for the skill's functionality.
[Indirect Prompt Injection Surface]: The skill processes instructions fetched from a remote URL to audit user files. This pattern is a known surface for indirect prompt injection, as the agent adopts rules provided in the external content. However, as the source is managed by the skill's authoring organization, the risk is effectively managed.
[File Access]: The skill reads local files specified by the user to perform the writing audit. This access is limited to the scope defined by the user and is necessary for the skill to provide the requested feedback.

writing-guidelines