The Agent Skills Directory

[COMMAND_EXECUTION]: The skill performs project discovery and diagnostics using standard CLI tools such as vite, npm, find, grep, and lsof to inspect dependencies and configuration files.
[COMMAND_EXECUTION]: It validates project setups by executing the existing build pipeline through npm run build or vite build, which is necessary for its primary function as a build optimizer.
[COMMAND_EXECUTION]: The diagnostic playbook includes generating a temporary file (vite.config.minimal.js) and running a test build with it to troubleshoot plugin conflicts.
[DATA_EXFILTRATION]: The skill inspects environment variables using env | grep VITE_. This is standard practice in the Vite ecosystem to identify public variables intended for the client and does not involve transmitting data to external destinations.
[PROMPT_INJECTION]: The skill has an indirect prompt injection surface as it processes untrusted project data. Analysis of this surface follows:
Ingestion points: Reads package.json, vite.config.*, and application source files in the src/ directory.
Boundary markers: None explicitly implemented for script execution.
Capability inventory: Can execute shell commands (npm, vite, grep, find) and perform file-system writes (echo).
Sanitization: No specific content sanitization or escaping of project data was detected before processing.

vite-expert