The Agent Skills Directory

[NO_CODE]: The skill references a script scripts/watch-files.mjs which is not included in the provided file list.
[COMMAND_EXECUTION]: The documentation describes a mechanism that uses child_process.execSync to run the claude CLI by interpolating a {{file}} placeholder. This pattern is vulnerable to command injection if a file is created or renamed with shell metacharacters (e.g., ; rm -rf /).
[PROMPT_INJECTION]: The skill implements an automated file watcher that passes file content to an AI model, creating a surface for indirect prompt injection.
Ingestion points: Any file added or modified within the directory monitored by the watcher.
Boundary markers: No boundary markers or 'ignore' instructions are mentioned to prevent the model from obeying instructions found inside the watched files.
Capability inventory: The skill executes commands via child_process.execSync and interacts with the claude CLI.
Sanitization: The documentation does not specify any sanitization, escaping, or validation of the file paths or content before execution.

file-watcher