The Agent Skills Directory

[COMMAND_EXECUTION]: The skill mandates the execution of several local scripts (e.g., ./scripts/analyze-codebase.py, ./scripts/detect-hardcoded-values.sh, ./scripts/generate-migration-report.py, and ./scripts/init-nextjs-shadcn.sh) that are not provided in the skill package for security analysis, posing a risk of unverified filesystem and process operations.
[EXTERNAL_DOWNLOADS]: Downloads and executes code via npx from well-known sources like create-next-app and shadcn/ui. While these sources are established, the automation of remote code execution is a relevant security factor for environments requiring strict control.
[REMOTE_CODE_EXECUTION]: Initializes an MCP server using npx shadcn@latest mcp init, which involves setting up a persistent execution bridge between the agent and local environment for documentation and component discovery.
[PROMPT_INJECTION]: The skill exhibits an attack surface for indirect prompt injection by ingesting and analyzing untrusted external codebases.
Ingestion points: Reads files and directory structures from user-specified codebase paths using the analyze-codebase.py script.
Boundary markers: No specific delimiters or "ignore instructions" warnings are used to isolate external content from the migration logic.
Capability inventory: The skill possesses broad capabilities including executing local scripts (Python/Bash), writing files (project creation), and performing network operations (npm/npx).
Sanitization: The skill does not describe any sanitization, validation, or filtering of the content read from the external codebase before it is processed by the agent.

nextjs-shadcn-builder