The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes standard system commands to manage and verify configurations, including stow for symlink management, zsh -n for shell script validation, tmux source for reloading terminal configurations, and yabai --restart-service for window management control. These commands are executed as part of the intended administrative workflow.
[DYNAMIC_EXECUTION]: The instructions include standard shell initialization patterns such as eval "$(zoxide init zsh)" for the zoxide navigation tool and the execution of the Tmux Plugin Manager (tpm) script. These patterns are common practices for integrating these tools into the shell environment.
[INDIRECT_PROMPT_INJECTION]: The skill presents an inherent surface for indirect prompt injection as it facilitates the modification of shell startup files based on external input.
Ingestion points: Instructions to edit source configuration files within the ~/.dotfiles directory (e.g., .zshrc, tmux.conf, .gitconfig) based on user requests.
Boundary markers: The skill does not define specific delimiters or provide explicit instructions to the agent to sanitize or ignore potentially malicious commands embedded within the configuration content.
Capability inventory: The skill possesses capabilities for file system management (stow, mkdir, find) and shell command execution (zsh, tmux, yabai).
Sanitization: There is no mention of sanitizing or validating the configuration data before it is written to the target files, relying on the user's intent.

shell-env