The Agent Skills Directory

[SAFE]: The skill body consists of markdown documentation intended to guide developers in creating consistent and predictable agent behaviors.
[EXTERNAL_DOWNLOADS]: References a public GitHub repository (mattpocock/skills) for attribution and reference purposes. This is a well-known source for developer tools and does not involve automated downloads or execution.
[DATA_EXPOSURE_AND_EXFILTRATION]: Mentions file paths within the agent's own installation directory (e.g., ${HOME}/.claude-octopus/plugin/). These references are instructional for plugin management and do not involve unauthorized access to user secrets or system credentials.
[COMMAND_EXECUTION]: References shell commands such as make sync and testing scripts in a purely instructional context for developer verification, with no instructions for the agent to execute them autonomously or maliciously.

skill-authoring