The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The installation guide explicitly references downloading scripts and binary packages from external sources including get.x-cmd.com, github.com/x-cmd/release, and conda.prefix.dev. The documentation notes that these downloads are for installation purposes and identifies associated risk levels for each method.
[REMOTE_CODE_EXECUTION]: The skill facilitates the execution of remote code through the curl | sh installation pattern described in data/install.md. The documentation provides a high-risk warning for this method, recommending Homebrew as a safer alternative and requiring explicit user consent before an agent performs an automated installation.
[COMMAND_EXECUTION]: The skill's primary function involves executing shell commands to manage programming runtimes and CLI tools. It performs operations such as modifying environment paths and installing packages to the user-local directory ~/.x-cmd.root/.
[INDIRECT_PROMPT_INJECTION]: The skill processes package names and versions provided by users or other agents which are interpolated into shell commands for software installation and execution.
Ingestion points: Package names and versions passed to commands like x env use <pkg> and x env try <pkg> in SKILL.md.
Boundary markers: None explicitly defined for package name inputs.
Capability inventory: Executes shell commands, writes to user-local directories, and performs network downloads for binary installation.
Sanitization: The skill documentation does not explicitly detail input sanitization for package names provided at runtime.

x-env