The Agent Skills Directory

[COMMAND_EXECUTION]: The script scripts/shadcn_add.py uses the subprocess module to execute npx shadcn add commands. This is used to programmatically install UI components into the project. The script avoids using shell=True and constructs the command as a list, which mitigates shell injection risks.
[INDIRECT_PROMPT_INJECTION]: The skill's scripts ingest data from project-level configuration files which represents a potential attack surface if those files are controlled by an attacker:
Ingestion points: scripts/shadcn_add.py reads package.json to determine the shadcn version and components.json to identify installation paths.
Boundary markers: Absent for these configuration files, as they are expected to be valid JSON.
Capability inventory: The skill can execute shell commands via npx and write configuration files to the local file system.
Sanitization: The _get_shadcn_version function sanitizes the version string from package.json by splitting on whitespace. The tailwind_config_gen.py script uses a strict regular expression (_VALID_PLUGIN_NAME) to validate plugin names before including them in a require() call in the generated configuration, preventing arbitrary code injection into the project's build configuration.

ui-styling