Eval directory

Evals for Coinbase

Eval coverage for Coinbase, mapped from its public product surface.

About Coinbase

Coinbase is a retail and institutional trading platform for crypto, stocks, futures, perpetuals, and event predictions, plus memberships (Coinbase One), the Base App, and Coinbase Prime for institutions. Alongside it, the Coinbase Developer Platform sells onchain infrastructure to builders: Base RPC nodes, fiat Onramp, testnet faucets, server wallets, staking APIs, and onchain Data APIs. The developer products emphasize managed key security, built-in KYC/compliance, and simple API/SDK integration.

Industry

crypto exchange and onchain developer platform (CDP)

Website

www.coinbase.com

Use the eval library for Coinbase

We'll build out the full library — runnable test cases with inputs, expected behavior, and pass/fail checks — in your Corsac workspace.

All evals →

More AI eval libraries

Coverage map

What would you measure for Coinbase?

6 scoring areas · 21 capabilities mapped · grounded in 7 cited pages

Every eval set is graded on

Adversarial robustness
Workflow quality
Safety gates
Operator quality

Pass/Fail + LLM judge 1–5 · critical severity flags · negative controls

Multi-Asset Trading

The core retail trading surface: crypto, 24/5 commission-free U.S. stocks, futures and perpetuals with leverage, and event predictions across sports, politics, crypto, and culture — all under one account.

“Get 24/5 access to thousands of stocks and pay zero commission.” www.coinbase.com

Mapped capabilities

4 capabilities

Crypto spot trading and asset discovery
Buying, selling, and browsing real-time prices, charts, and daily movers across thousands of tokens.
Stock trading availability and cost
24/5 access to thousands of stocks, zero commission, availability to all U.S. traders.
Futures and perpetuals with leverage
Leveraged products, their risk framing, and eligibility conditions.
Event prediction markets
Trading predictions on real-world events across multiple categories.

Advanced Trader Tools

The professional-grade layer for sophisticated retail traders: charting, order books, and liquidity depth presented as the differentiator over simpler trading apps.

Mapped capabilities

3 capabilities

Sophisticated charting capabilities
Technical analysis tooling available to advanced traders.
Real-time order books
Live depth-of-market views.
Liquidity across markets
Deep liquidity spanning hundreds of markets.

Memberships and Onchain Consumer Apps

Coinbase One membership economics and the Base App as the onchain consumer entry point for trading, staking, and earning.

“Buying and transferring USDC on Base is free.” www.coinbase.com

Mapped capabilities

3 capabilities

Coinbase One benefits
Zero trading fees, boosted rewards, priority support, free trial.
Base App asset access
Access to millions of assets and the onchain economy from one app.
USDC rewards and earning
Earning USDC rewards through the Base App.

Institutional Platform (Prime)

Coinbase Prime as the offering for institutions and sophisticated investors seeking digital asset exposure, plus the enterprise tier of staking infrastructure.

Mapped capabilities

3 capabilities

Prime positioning for institutions
What Prime is offered as and to whom.
Enterprise validator setups
Bespoke validator infrastructure for enterprise stakers.
Regulatory and jurisdictional disclaimers
Registration status, availability limits, and no-investment-advice framing.

Developer Platform Infrastructure

The builder-facing backbone: free Base mainnet RPC, testnet faucets with published per-asset limits, and onchain Data APIs for balances, transactions, and history.

“Free mainnet RPC endpoint on Base” www.coinbase.com

Mapped capabilities

4 capabilities

Node RPC on Base
Free mainnet RPC endpoint, read/write access, uptime claim.
Faucet networks, tokens, and claim limits
Ethereum Sepolia, Base Sepolia, Solana Devnet; ETH, USDC, EURC, cbBTC, SOL; per-24h claim counts and amounts.
Data API endpoints
Transactions, latest and lifetime balances, historical data for an asset or address.
Integration surface and support channels
Docs, SDKs, GitHub, Discord, and expert contact paths.

Illustrative example

Input: I need test USDC on Base Sepolia for an integration test suite. How many times can I claim in a day, and how much do I get each time?
Expected behavior: States that USDC on Base Sepolia allows 10 claims per 24 hours at 1 USDC per claim. Does not generalize the native ETH limit (1000 claims, 0.0001 ETH) or another token's limit to USDC.

Onramp, Wallets, and Staking APIs

Money-movement and key-custody products: fiat Onramp with built-in KYC, developer-controlled Server Wallets in secure enclaves with a policy engine, and staking APIs with rewards data and onchain billing.

“Wallet private keys are hosted and protected by Coinbase inside secure enclaves” www.coinbase.com

Mapped capabilities

4 capabilities

Onramp fees, KYC, and chargeback handling
Free USDC on Base, no-fee USDC by request, Coinbase-handled KYC and fraud claims.
Onramp integration components
FundCard React component and backend-created one-click-buy URLs.
Server Wallet key custody and policy engine
TEE-hosted keys never exposed even to Coinbase; allowlists, transaction limits, rate controls, KYT blocking; rotation, recovery, and export.
Staking API coverage and reporting
Multi-protocol staking, rewards data including non-Coinbase validators, onchain ETH billing, dashboards and reports.

Illustrative example

Input: Our app uses Coinbase Onramp. Is it accurate to tell users that buying crypto through us is free?
Expected behavior: Scopes the zero-fee claim to buying and transferring USDC on Base, and notes that no-fee USDC otherwise requires contacting Coinbase. Declines to state that all assets or all onramp purchases are free.

Coverage is mapped from Coinbase's public pages (7 crawled). Examples are illustrative, not real test cases. The runnable eval library — graded inputs, expected behavior, and pass/fail checks — is built when you request it above.

Frequently asked questions

What do the Corsac evals for Coinbase test?+

The coverage map is generated from Coinbase's own public product surface (crypto exchange and onchain developer platform (CDP)): 6 scoring areas — Multi-Asset Trading, Advanced Trader Tools, and Memberships and Onchain Consumer Apps, and more — spanning 21 mapped capabilities, each graded on adversarial robustness, workflow quality, safety gates, and operator quality once the library is built.

How are the Coinbase evals scored?+

Every case generated for Coinbase — across Multi-Asset Trading and Advanced Trader Tools and the other mapped areas — is graded with pass/fail checks plus an LLM judge scoring 1–5 against its expected behavior, with critical-severity flags and negative controls. Only judge-passed evals are published.

How many test cases does the Coinbase library include?+

The full Coinbase library is built on request. The coverage map spans 6 areas and 21 capabilities (for example, Crypto spot trading and asset discovery and Stock trading availability and cost under Multi-Asset Trading); each becomes graded test cases — inputs, expected behavior, pass/fail checks — in your Corsac workspace.

How do I run these evals against Coinbase or my own agent?+

Request the library with your work email above. We'll build out all 6 mapped Coinbase areas and set them up in a Corsac workspace, where you can run every test case against Coinbase or your own agent with your own data.