If x402 is “payments for the agent economy”, then agent-friendly API design and evaluation is the quality layer that makes it real. We want a world where APIs are not just available, but measurable, comparable, and dependable for agents in production!

There’s more that matters for repeat Agent Experience (AX), and we’re actively exploring them for the next iteration: • Self-recovery: can the agent fix a failed call and try again successfully? • Rate limits: does the API clearly tell the agent when to retry? • Read vs write clarity: is it clear which calls only read data and which modify state? • End-to-end actions: can one call complete a real user task?

Metric #3: Production Readiness Can an agent rely on this API over time in a real workflow? We evaluate whether: • Calls succeed consistently under real agent behavior. • Latency stays within usable bounds • Cost remains predictable and controllable

Metric #2: Output Usefulness Does a successful response actually advance the task? We split this into: • Output correctness: is the information right? • Output helpfulness: does it actually move the task forward? Stable and predictable response schemas matter a lot here.

Metric #1: Input Friendliness Can a cold-start agent construct a valid call? We evaluate whether: • Required vs. optional params are explicit. • Naming is consistent across docs and schema. • Examples are complete and executable.

Our philosophy is straightforward. An API is agent-ready only if an agent can: a. Call it successfully. b. Understand the response. c. Recover when things go wrong. Otherwise, it’s just a demo endpoint.

That’s the gap OpenMid Evaluator is built for. We pass each eligible x402 API endpoint to our evaluation pipeline daily (soon for 8004 agents as well), compute quality score for each and display all the evaluation results per API.

Today we’re launching OpenMid Evaluator (in beta): evaluator.openmid.xyz x402 makes payments native to HTTP — it unlocks a massive API surface: • Agents can discover an endpoint, pay per request, and just use it. • No subscriptions. No accounts. Pure pay-per-use. ⚠️ But how many of these APIs actually work for agents, end to end?

3️⃣ Ready to integrate. Our official docs for x402 merchants are now live: openmid.xyz/docs If you have questions or want feedback, join our Telegram and talk to us directly: t.me/+Y3BDOa08O59lZmVh

2️⃣ Sponsored transactions on Base, no setup required We are sponsoring Base Mainnet transaction facilitation for all x402 merchants, so you can test real onchain payments without setting up infra or paying upfront costs. Try it here: facilitator.openmid.xyz

1️⃣ Register your agent during payment, not as a separate step. We shipped an optimized x402 v2 extension that lets you register an agent to the ERC-8004 registry as part of the x402 payment lifecycle. All you need is the erc-8004 extension plus a 7702 auth. No extra registry calls, no custom glue code.

🚀 Big update for x402 developers today at OpenMid. OpenMid Facilitator is now live on Base Mainnet! You can now register an agent to the ERC-8004 registry directly inside the x402 payment flow, and run everything for free with sponsored facilitation. More details below 🧵

8004 Facilitator live today on Base Sepolia: facilitator-testnet.openmid.… Code: github.com/open-mid/8004-fac…

OpenMid introduces two 8004-native extensions to x402: • /register — onchain identity (8004 identity registry, now open-sourced) • /feedback — onchain reputation (releasing soon with the upcoming 8004 update) It also supports both x402 v1 & v2.

Today we are launching OpenMid — the first 8004-native facilitator for x402, built to connect identity, reputation, and payments for the agent economy. As unveiled at Trustless Agent Day @EFDevcon by @davidecrapis and @codywang999. Devconnect session replay below 👇

Hello world!