Vercel Deploy

What it does

This skill enables an AI agent to deploy projects directly to Vercel. It automates the deployment process, handling framework detection, packaging, and uploading your project. By default, deployments are created as preview environments; production deployments require explicit user instruction.

When to use it

• You need to quickly deploy a frontend UI or user experience prototype to Vercel for testing and feedback.
• You want to automate the deployment process of an existing Vercel project.
• You’re working on a project that requires frequent deployments during development.
• You need to deploy from a tarball archive.

Key capabilities

• Deploys projects to Vercel as preview environments by default.
• Supports production deployments upon explicit user request.
• Handles framework detection, packaging, and deployment automatically.
• Provides previewUrl and claimUrl in JSON output for deployed projects.
• Offers a fallback deployment script for situations where CLI authentication fails.

Example prompts

• "Deploy this project to Vercel." (Deploys the current directory)
• "Vercel deploy /path/to/my/project -y" (Deploys a specific project path)
• “Can you do a production deployment?” (Initiates a production deployment)

Tips & gotchas

• Timeout: The deployment process can take several minutes. Always use a 10-minute (600,000ms) timeout for the vercel deploy command.
• Network Permissions: Deployments may require escalated network access if sandbox networking blocks outbound requests. Be prepared to rerun the deployment with sandbox_permissions=require_escalated.
• Authentication: If you encounter "No existing credentials found" errors, use the fallback bash script method for deployment.