The Agent Skills Directory

[EXTERNAL_DOWNLOADS] (LOW): The skill utilizes nix-prefetch-url to download content from arbitrary URLs and uses the nix command to fetch packages from the Nix ecosystem. These are standard operations for a package manager.
[REMOTE_CODE_EXECUTION] (LOW): Commands like nix run and nix shell facilitate the execution of code downloaded from remote repositories. While this constitutes high-privilege capability, it is the primary intended use case for Nix and is generally considered acceptable in the context of development tools.
[COMMAND_EXECUTION] (LOW): The skill provides explicit methods for executing arbitrary commands within defined environments (e.g., nix shell --command).
[Indirect Prompt Injection] (LOW): The skill has a vulnerability surface for indirect prompt injection because it processes external data without specific sanitization.
Ingestion points: Untrusted data enters the context through nix eval --file ./default.nix and nix-prefetch-url.
Boundary markers: There are no explicit delimiters or instructions to ignore embedded commands within ingested files or remote content.
Capability inventory: The skill possesses powerful capabilities including nix run (code execution), nix eval (dynamic evaluation), and network access via curl.
Sanitization: No sanitization or validation logic is present to inspect the content of Nix expressions or fetched sources before processing.

nix