Repository navigation
0.7.0 (6/7): Ethereum Sepolia, the $1 spend cap, named settlement failures - #9
Merged
Merged
Conversation
The payment checks X402-06..10 now also pay on Ethereum Sepolia (eip155:11155111), through the same EVM adapter as Base Sepolia, with the payer key in EVM_PRIVATE_KEY. Circle's USDC there has EIP-3009 (read from the contract: authorizationState answers, EIP-712 domain "USDC" version "2"), and Permit2 and the x402 proxies are deployed, so a payer needs the token and no ETH. The SDK ships no default asset for Ethereum Sepolia, and its client's spend controls refuse any asset that is not a default. The adapter allows Circle's Sepolia USDC on that network only, with the cap the SDK puts on its defaults, $1 a payment; every other network is unchanged. Sepolia makes a block every 12 seconds, so the wait for a settled transaction is now per network: 10 blocks within 240 seconds there, 30 within 120 on Base Sepolia as before. verifySettlement takes the network to know which. X402-10's 5-second hold is unchanged: the SDK facilitator holds validBefore to the clock with a 6-second margin, and any block that could include the payment is later still. No public facilitator settles Ethereum Sepolia, so its fixture (port 3008) runs the official SDK's facilitator in process, paying the gas from EVM_FACILITATOR_PRIVATE_KEY; fixtures.sh starts it only when that key and EVM_PAYEE_ADDRESS are set, and run-all.sh runs it when it is up. BNB Smart Chain testnet stays read-only, and replaces Ethereum Sepolia as the tests' example of a network the payment checks refuse. Against `wasit serve` on Sepolia, with the funded payer: no-settle fails X402-06..10, wrong-settlement fails X402-06 after 10 blocks (131 s), wrong-network skips with nothing sent. The fixture passes X402-01..05. 235 core tests; 2 mutants of the allowance killed. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
…tions `wasit serve --network eip155:11155111` poses every mode on Ethereum Sepolia: Circle's Sepolia USDC with its EIP-712 domain, wrong-network asking for Ethereum mainnet (eip155:1), overprice one million USDC in 6 decimals, v1-challenge on v1's "sepolia". `wasit wallet --network eip155:11155111` takes it too, through the same profile. `wasit test` and the MCP tool name it among the networks the payment checks pay on. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
…e price The official x402 client pays only its default assets, plus the ones an adapter allows (Circle's USDC on Ethereum Sepolia), and at most $1 a payment, so that a paywall cannot charge a payer more than it meant to. Wasit keeps those limits: it pays automatically, and on stellar:pubnet with real money. A target priced above them used to get X402-06 ERROR (harness) with the SDK's own advice to change its configuration. Such a refusal is now ERROR (setup), with nothing sent and the negative checks skipped, saying which limit: the price in base units against the $1 cap, or the asset the client does not pay in. It is matched on the SDK's message (@x402/core 2.28); one it does not recognise passes through as before. Live against `wasit serve` on Base Sepolia priced at 2000000 units (2 USDC), with a payer holding 19.8: ERROR (setup) "asks for 2000000 base units ... more than the $1 a payment", and the paywall received no payment. 236 core tests; 3 mutants killed. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
A payment that verified but did not settle came back as "got 402." with no reason. The official SDK server does say why: it answers with a PAYMENT-RESPONSE whose success is false and whose errorReason names the facilitator's failure. Wasit read the reason only from a fresh challenge's `error` or the body, so it missed it. The refusal reason now reads PAYMENT-RESPONSE first, as "settlement failed: <code>" (a long RPC message after the code is dropped), and X402-06's hint for it says the facilitator did not settle a payment that verified. A PAYMENT-RESPONSE that claims success is not read as a failure. Found on Wasit's own Base Sepolia fixtures: a few runs failed X402-06 with "got 402.", and the fixture's facilitator log showed the public facilitator's settlement failing at eth_sendRawTransaction (invalid_exact_evm_transaction_failed). 238 core tests; 4 mutants killed. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
A refused settlement reached the client as a bare 402 before Wasit read PAYMENT-RESPONSE, and the fixture's log was the only place to see why. It now logs each verify refusal and settle failure with the facilitator's reason, which is how the intermittent invalid_exact_evm_transaction_failed on Base Sepolia was found. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
…nt failures CHECKS.md, the CLI, MCP and configuration guides, the CLI README, the agent skill and the site's stack strip name Ethereum Sepolia among the networks the payment checks pay on, with Circle's Sepolia USDC and the fixture's own facilitator key. CHECKS.md says the official client's limits ($1 a payment, its default assets plus Circle's Sepolia USDC) give no verdict, and that a failed settlement reads "settlement failed: <reason>"; its X402-08 row now says the public facilitator refuses a replay at settlement, as measured. The evidence: on Ethereum Sepolia the lying modes of `wasit serve` fail and the fixture passes 10/10, the settlement read back with viem (one Transfer, sent by the fixture's facilitator, payer at 0 ETH); a 2 USDC paywall gets ERROR (setup) and receives nothing; and on Base Sepolia the intermittent X402-06 failure is the public facilitator failing to broadcast its settlement (five of sixteen runs on 2026-10-06), which the payer-balance evidence now records as found. The wallet and serve-modes evidence names its commit. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
|
The latest updates on your projects. Learn more about Vercel for GitHub.
|
This branch was successfully deployed
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Part 6 of 7 of 0.7.0, which is on npm since 2026-10-08. The PRs follow the release branch's commit order and are merged with merge commits, so every commit keeps its hash.
What changes
eip155:11155111) in Circle's USDC.PAYMENT-RESPONSE.Commits
35b12a6feat(core): pay Ethereum Sepolia targets11672d4feat(cli): Ethereum Sepolia in wasit serve, wallet and the network options93e04c0feat(core): no verdict when the SDK client's spend controls refuse the price178b8a6fix(core): name a failed settlement's reason, read from PAYMENT-RESPONSEba2b6cctest(fixtures): the Permit2 fixture logs its facilitator's refusals888ca1ddocs: Ethereum Sepolia, the client's $1 spend cap, and named settlement failuresVerification
Unit tests and the clean install run in CI. The Ethereum Sepolia runs are in
docs/evidence/2026-10-06-ethereum-sepolia-verification-run.md(folded into the per-date file in PR 7/7).🤖 Generated with Claude Code