The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes arbitrary Nx generators based on user-provided or inferred names and options using nx generate <generator-name> <options> --no-interactive. This allows for a wide range of file system and configuration changes.
[COMMAND_EXECUTION]: Uses node -e to execute inline JavaScript for path resolution (node -e "console.log(require.resolve(...))"), which is a form of dynamic code execution via the terminal.
[COMMAND_EXECUTION]: Instructs the agent to run verification commands such as nx lint, nx test, and nx build, which execute scripts defined in the repository's configuration files.
[REMOTE_CODE_EXECUTION]: While focused on Nx, the use of npx nx list or nx generate can trigger the installation and execution of npm packages if the specified plugin is not locally available.
[PROMPT_INJECTION]: Indirect surface. The skill processes external data (generator schemas, repository source code, and user input) and maps it directly to execution flags. Evidence Chain:
Ingestion points: User requests for generators, generator schema files, and local repository source code (SKILL.md).
Boundary markers: Absent; there are no instructions to ignore embedded commands within the data being processed.
Capability inventory: Shell command execution (nx), file system modification, and dynamic Node.js execution (SKILL.md).
Sanitization: Absent; input is mapped directly to command-line arguments without escaping or validation logic.

nx-generate