1000 scored services for agent compatibility

Most APIs weren't built for agents. Now you know which ones were.

Rhumb measures execution reliability and access readiness for every service your agent might call. Best current fit: research, extraction, generation, and narrow enrichment. Neutral scores. Real failure mode data. Zero-signup discovery via MCP or REST API.

Search 1000 services →

Best fit today

Start with research, extraction, generation, and narrow enrichment. That is where Rhumb is strongest and most honest today.

Current execution boundary

Discovery is broader than execution today: Rhumb indexes 1,038 scored services and 415 capability definitions, but current governed execution is concentrated in 16 callable providers. Layer 3 is live but the public recipe catalog is still intentionally sparse.

1000
services scored
92
categories
20
scored dimensions
live
evidence data

Featured · Payments

Top services in payments

  1. 1
    8.8 L4
    Adyen L4 Native
    Aggregate 8.8 · Freshness Updated 2026-03-06T22:21:51.113+00:00 Assessed
  2. 2
    8.3 L4
    PayPal API L4 Native
    Aggregate 8.3 · Freshness Updated 2026-03-26T03:14:35.444+00:00 Assessed
  3. 3
    8.3 L4
    Paddle L4 Native
    Aggregate 8.3 · Freshness Updated 2026-03-16T06:18:54.293494+00:00 Assessed

Open full payments leaderboard

What is the AN Score?

The Agent-Native (AN) Score measures how well a developer API works for autonomous AI agents — not humans reading documentation, but machines making API calls without supervision.

Each service is evaluated across 20 dimensions on two axes: Execution (70% weight — reliability, error handling, schema stability, idempotency, latency) and Access Readiness (30% weight — signup friction, credential management, rate limits, documentation, sandbox availability).

Services are ranked into four tiers: L4 Native (8.0–10.0), L3 Ready (6.0–7.9), L2 Developing (4.0–5.9), and L1 Emerging (0.0–3.9).

Every score is published, transparent, and disputable. Rhumb surfaces active failure modes, evidence tiers, and freshness timestamps so agents can make informed routing decisions at runtime.

Research

Deep dives, not hot takes

Comparisons backed by AN Score data. Autopsies that show exactly where APIs break for agents. Guides that get you running in minutes.

Choose your path

Start with the mode that matches your job.

Quickstart guide → Browse for free, or pick the execution path that matches how your agent gets authorized.

🔍

Browse

Free

Explore scores, comparisons, and failure modes first. No account, no key, no payment rail required.

Start with discovery →
🔑

API key

Managed

Use Rhumb-managed execution when you want routing, failover, and billing under one key.

Get dashboard access →
⚡

x402

Zero signup

Use pay-per-call USDC when the agent should pay directly and the payment should act as authorization.

Learn the payment flow →
🧷

BYOK

Zero markup

Bring your own provider credentials when you need direct control, enterprise boundaries, or existing vendor accounts.

See credential modes →
Zero-signup agent payments

Agents with x402 can execute capabilities and pay per call with USDC when zero-signup, request-level payment authorization is the point. For repeat traffic, the default path is still API key or wallet-prefund.

Learn about x402 →

MCP

Query Rhumb from your agents

Add Rhumb to any MCP-compatible agent. AN Scores at decision time. Perfect for agent routing logic, tool-selection prompts, and automated dependency evaluation.

Explore all categories →
terminal
$ npx rhumb-mcp # starts MCP server
✓ Rhumb MCP server running
  tools: find_services, get_score, get_alternatives, get_failure_modes
→ find_services ("payment processing")
1. stripe 8.1 L4 Native
2. lemon-squeezy 6.6 L3 Ready
3. square 6.3 L3 Ready