The Agent Skills Directory

[PERSISTENCE]: The skill's primary function is to modify the user's shell configuration file (~/.zshrc). By adding functions and aliases to this file, the skill ensures that specified commands are automatically loaded and available in every new terminal session, which is a standard persistence mechanism.
[COMMAND_EXECUTION]: The scripts/add_alias.py script accepts arbitrary shell commands via the --cmd argument and writes them directly into the shell configuration. In the default function-based mode, the script performs no sanitization or escaping of the command string, relying entirely on the agent or user to provide safe input. This allows for the injection of arbitrary shell code into the user's environment.
[DATA_EXPOSURE]: The skill reads from and writes to ~/.zshrc. Shell configuration files often contain sensitive information, such as environment variables, API keys, or path configurations. The skill has full read/write access to this sensitive file.
[INDIRECT_PROMPT_INJECTION]: The skill presents an attack surface for indirect injection as it processes untrusted user input (trigger names and commands) and interpolates them into executable shell scripts.
Ingestion points: User-provided arguments for --name, --cmd, and --comment in scripts/add_alias.py.
Boundary markers: The script uses sentinel markers (# >>> lov-zsh-alias >>>) to delimit its blocks in the RC file, but these do not prevent the shell from executing malicious content within the block.
Capability inventory: The skill can write arbitrary content to the filesystem (~/.zshrc) and execute shell commands through the aliases it creates.
Sanitization: The script performs minimal backslash and double-quote escaping for the alias style but provides no sanitization for the default function style.

lov-zsh-alias