The Agent Skills Directory

[COMMAND_EXECUTION]: The skill establishes persistence on macOS by instructing the creation of a launchd configuration file in '~/Library/LaunchAgents/'. This allows the server to maintain its state and execute automatically across user sessions.
[COMMAND_EXECUTION]: The skill requires the user to compile a Swift binary from source code provided in the 'service-management.md' file. This process introduces a risk of arbitrary code execution during the installation phase, as the compiled binary is then used to launch the primary server.
[REMOTE_CODE_EXECUTION]: The troubleshooting section contains a command pattern where data from a local network endpoint is piped directly to the Python interpreter ('curl ... | python3'). Although intended for JSON formatting via '-m json.tool', the pattern of passing network-retrieved content to a language interpreter is a common vector for remote code execution.
[PROMPT_INJECTION]: The skill includes misleading claims stating that macOS launchd policy requires a compiled Swift binary instead of a script. This misinformation may be used to justify unnecessary compilation and execution of custom code on the host system.

server