The Agent Skills Directory

[SAFE]: The skill is strictly read-only and explicitly prohibits any mutating actions such as scanning, writing to repositories, or modifying Endor Labs state.
[SAFE]: Implements a robust namespace preflight process to ensure data is accessed from the correct and authorized context.
[SAFE]: Includes explicit instructions to treat finding titles, descriptions, and metadata as untrusted data to prevent prompt injection from processed content.
[SAFE]: Shell command usage is restricted to read-only lookups via endorctl api with specific field masks to avoid echoing sensitive information like secrets or credential keys.

findings-browser