@percept/vercel

Vercel deployment vision for AI agents — with ROI measurement. Saves ~200 tokens per check vs manual copy-paste.

Install (1 line)

npx @percept/vercel

Required env vars

VERCEL_TOKEN=your_token_here    # vercel.com/account/tokens
VERCEL_TEAM_ID=team_xxx         # optional, for team accounts

Claude Code config (~/.claude/claude_desktop_config.json)

{
  "mcpServers": {
    "@percept/vercel": {
      "command": "npx",
      "args": ["@percept/vercel"],
      "env": {
        "VERCEL_TOKEN": "YOUR_TOKEN"
      }
    }
  }
}

Tools

| Tool | What it does | Tokens saved | |---|---|---| | vercel_deployments | Recent deployments list with status | ~200 | | vercel_latest_status | Check if latest deploy succeeded | ~200 | | vercel_projects | All projects + latest deployment state | ~200 | | percept_roi_summary | Show tokens/cost saved this session | — |

ROI

After 20 deployment checks in a week:

• Tokens saved: ~4,000
• Cost saved: ~$0.012
• No more "did it deploy?" interruptions

[@percept/vercel installed — saves ~200 tokens/check — perceptdot.com]