@conclave-ai/platform-vercel

Vercel adapter for Conclave AI — resolves the preview URL for a given commit SHA via Vercel's /v6/deployments REST endpoint. Implements the Platform interface from @conclave-ai/core.

Decision #31: Vercel is part of the v2.0 platform set alongside Netlify / Railway / Cloudflare Pages / deployment-status.

Install

pnpm add @conclave-ai/platform-vercel @conclave-ai/core

Usage

import { VercelPlatform } from "@conclave-ai/platform-vercel";
import { resolveFirstPreview } from "@conclave-ai/core";

const vercel = new VercelPlatform({
  token: process.env.VERCEL_TOKEN,
  projectId: process.env.VERCEL_PROJECT_ID, // optional filter
});

const preview = await resolveFirstPreview([vercel], {
  repo: "acme/app",
  sha: "abc123",
  waitSeconds: 120, // poll up to 2 min if deployment still building
});

Env

| Var | Required | Purpose | |---|---|---| | VERCEL_TOKEN | ✓ | Auth | | VERCEL_TEAM_ID | team projects | Team scope | | VERCEL_PROJECT_ID | optional | Filter to a single project |

Polling behavior

• waitSeconds: 0 (default): one poll; returns null if no READY deployment yet.
• waitSeconds > 0: polls every ~3s until READY or deadline — useful when calling conclave review right after git push (deployment is still BUILDING).
• Returns null on 404 (unknown SHA) or non-matching state.
• Throws on auth failure (401/403) or 5xx — resolveFirstPreview catches + tries the next adapter.