The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill operates by reading and modifying local project files (e.g., package.json, vite.config.ts, and CI workflow YAML), which could potentially contain malicious instructions designed to influence the agent's behavior.
Ingestion points: Processes repository configuration files such as package.json, vite.config.ts, and .github/workflows/ during setup, migration, and maintenance tasks (documented in SKILL.md and references/bootstrap.md).
Boundary markers: There are no explicit instructions or delimiters used to isolate ingested data from the agent's instruction context.
Capability inventory: The skill is authorized to execute various shell commands including pnpm exec vp, pnpm install, and pnpm dlx across multiple reference files.
Sanitization: No specific sanitization or validation of the content within ingested files is described before it is used in command line arguments or logic.
[REMOTE_CODE_EXECUTION]: The skill provides instructions for running pnpm dlx to execute the migration tool (vp migrate) from an external package registry (documented in references/bootstrap.md). While this is a standard developer workflow for tool migration, it involves fetching and executing code at runtime.
[COMMAND_EXECUTION]: The skill frequently utilizes pnpm exec vp and other package manager commands to perform its primary functions, such as testing, building, and linting. This is typical behavior for a development-focused AI skill.

vite-plus