On 2026-11-06, Scope (YC P26) founder Anand Paj (@anandPa94) released the first hosted-database agent signup benchmark. Agents start with no account, provision the named DB, write a record, and hand back credentials across 20 Claude Code/Codex/Cursor configs and 1,035 sessions. Official leaderboard: Prisma Postgres ~88.3%, Upstash ~63.3%, Neon/Nile ~43.3%; CAPTCHAs stopped 373 sessions; 7 of 18 products had zero successes.
Key Takeaways
- ✓Source: Scope founder Anand Paj (@anandPa94) X post 2026-11-06 + official board tryscope.com/leaderboard/databases (vendor/evaluator)
- ✓Scale: 18 hosted DBs × 20 agent configs = 1,035 sessions; only ~16.6% reached a first working result overall
- ✓Leaders: Prisma Postgres ~88.3% (53/58), Upstash ~63.3%, Neon/Nile ~43.3%, ClickHouse Cloud ~38.3%
- ✓Friction: CAPTCHAs stopped 373 sessions, bot blocks 105; 7/18 products had zero successes; 50 of 172 wins needed no account
- ✓Takeaway: for agent-facing devtools, signup→write→hand-back-credentials is now part of the conversion funnel

Key Decision Metrics at a Glance
Turn your technical choice into a development budget
Compare 40 dev plans & simulate token costs vs $20/mo subscriptions
Project Links & Resources
Direct AccessIn-Depth Technical Analysis
Core Background & Industry Pain Points
Coding agents are often asked to open a DB, write data, and return credentials, yet most hosted databases still assume a human browser. On 2026-11-06, Scope (YC P26, @tryscope_app) founder Anand Paj (@anandPa94) published the first hosted-database agent signup benchmark with methodology and session transcripts. Evaluator/founder first-person release—not a joint vendor announcement.
Architecture Highlights & Internals
Fixed task: start with no account, provision the named product, write a nonce record, and hand back externally verifiable credentials; an account or API key alone does not count. Twenty model×harness configs spanning Claude Code, Codex, and Cursor ran from about 2026-11-02 to 11-06 for 1,035 sessions. Success is the average per-config success share on each product.
Authoritative Benchmarks & Measured Scores
Official board (evaluator numbers): Prisma Postgres ~88.3% (53/58), Upstash ~63.3%, Neon and Nile ~43.3%, ClickHouse Cloud ~38.3%; Turso ~5.0%, Cloudflare D1/MongoDB Atlas ~3.3%, Supabase ~1.7%. Seven of 18 products had zero successes. Overall ~16.6% reached a first working result; CAPTCHAs stopped 373 sessions and bot blocks 105. Fifty of 172 wins used a working DB without creating an account (e.g. Prisma npx create-db).
Developer Hands-on Guide
For agent-facing products, test no-CAPTCHA flows, machine-readable docs, and try-before-signup. On Prisma, npx create-db@latest --json yields a temporary Postgres. Full ranks, failure modes, and transcripts are on the official board—cite as Scope evaluator results, not a vendor self-score.
Benchmark side-by-side against alternatives, or calculate monthly token cost vs subscription break-even.
Discussion & Comments
0Sign in to join the discussion
Connect with AI developers to exchange benchmark insights.