The Agent Skills Directory

[EXTERNAL_DOWNLOADS] (MEDIUM): The skill utilizes npx shadcn@latest add <name> --yes to install UI components. This command downloads and executes a script from the npm registry. Because it uses the @latest tag and the --yes flag, it bypasses version pinning and manual verification, allowing for the execution of external code. While standard for this tech stack, it represents an unverified dependency risk.
[COMMAND_EXECUTION] (LOW): The skill executes various system and development commands including npm run build, npm run lint, git diff, and ls. These are necessary for the skill's primary purpose but involve interaction with the underlying shell.
[PROMPT_INJECTION] (LOW): The skill is vulnerable to indirect prompt injection as it is instructed to ingest data from untrusted or external files.
Ingestion points: The agent reads features/INDEX.md and external feature specification files provided via [feature-spec-path]. It also invites the user to provide 'inspiration URLs'.
Boundary markers: Absent. There are no delimiters or instructions to ignore embedded commands within these external files.
Capability inventory: The skill has the ability to write to the file system (src/), execute shell commands (npm, npx), and interact with git.
Sanitization: Absent. The agent processes the contents of these files directly into its task context without validation.

frontend