Vercel Blob

Overview

Vercel Blob has no native bucket concept, so the adapter maps each storagesdk bucket to a pathname prefix (<bucket>/<key>). Multiple logical buckets can coexist in one Vercel Blob store; snapshots and forks land at sibling prefixes.

Configuration

import { Storage } from '@storagesdk/core';
import { vercel } from '@storagesdk/adapters/vercel';

const storage = new Storage({
  adapter: vercel({
    bucket: 'agent-runs',
    token: process.env.BLOB_READ_WRITE_TOKEN,
  }),
});

vercel({
  bucket: string;
  token?: string;          // defaults to BLOB_READ_WRITE_TOKEN env var
  access?: 'public' | 'private';  // default 'public'
});

Notes

• Peer dependency: @vercel/blob.
• token defaults to process.env.BLOB_READ_WRITE_TOKEN (set automatically on Vercel runtimes); pass explicitly when running elsewhere.
• Per-object metadata on upload() is silently dropped — Vercel Blob has no user-metadata field. Use a sidecar object if you need it.
• minSize on uploadUrl() is silently dropped (Vercel signed URLs cap with maximumSizeInBytes; no lower bound).
• access: 'public' (the default) serves through Vercel’s CDN — brief stale reads possible. Use access: 'private' when snapshot/fork manifest reads need strong consistency.

Compatibility