The Agent Skills Directory

[COMMAND_EXECUTION]: The skill enables execution of arbitrary tools and commands through nix run and nix shell using the standard nixpkgs repository. This is the primary intended functionality for reproducible environment management.
[DYNAMIC_EXECUTION]: The skill facilitates runtime evaluation of Nix code using nix eval. This is used for inspecting attributes, debugging local .nix files, and querying built-in functions, which is routine for the Nix ecosystem.
[INDIRECT_PROMPT_INJECTION]: The skill processes user-supplied expressions and package names at runtime.
Ingestion points: Shell command arguments and Nix expressions passed to the nix CLI (SKILL.md).
Boundary markers: None provided.
Capability inventory: Subprocess execution via nix, local file reading via nix eval --file, and potential network access via utilities like curl when run inside a Nix shell.
Sanitization: Relies on the native Nix sandboxing mechanisms and the agent's execution environment constraints.

nix