x402-tools
An MCP server that lets an AI agent pay per call for live trust-signal and utility lookups over the x402 protocol.
list_products(free): fetch the live catalog of tools, prices, and schemas.package_trust_check($0.05 USDC): npm/PyPI install-safety analysis with registry metadata, install scripts, OSV advisories, typosquat risk, and a 0-100 trust score.repo_merge_lookup($0.05 USDC): predicts GitHub PR merge likelihood from a repo's AI/contribution policy and historical merge rates, 0-100 score.a11y_audit($0.08 USDC): WCAG 2.1/2.2 AA accessibility audit of a public page using headless Chromium and axe-core.schedule_solve($0.30 USDC): solves scheduling/roster problems with OR-Tools CP-SAT, assigning resources to slot demands with constraints.mx_rfc_validate($0.03 USDC): validates Mexican RFCs (person/company, embedded date, SAT check digit) and EFOS blacklist status.mx_clabe_validate($0.02 USDC): validates 18-digit Mexican CLABE bank codes including check digit and bank identification.mx_cfdi_verify($0.05 USDC): verifies Mexican CFDI invoices with SAT status, cancelability, and EFOS validation.Automatic x402 payments when a buyer private key is configured, with a spending guard (
X402_MAX_PRICE_USD) and challenge-only mode when no key is set.
Provides a tool to look up whether a GitHub repository is likely to merge an AI-authored or external pull request, using stated AI/contribution policy and historical merge rates to produce a 0-100 score.
Provides a tool to assess whether an npm package is safe to install, using registry metadata, install-script usage, OSV advisories, and typosquat risk to produce a 0-100 score.
Provides a tool to assess whether a PyPI package is safe to install, using registry metadata, install-script usage, OSV advisories, and typosquat risk to produce a 0-100 score.
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@x402-tools Is the npm package lodash safe to install?"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
x402-tools
An MCP server that lets an AI agent pay, per call, for a set of small trust-signal and utility lookups over the x402 protocol: package install-safety, GitHub PR-merge likelihood, WCAG accessibility audits, roster/scheduling solving, and Mexican RFC/CLABE/CFDI validation. The catalog and every price are read live from the seller API at call time — nothing here is hardcoded.
Built and operated by an AI agent, under a human owner
This tool was built by an AI coding agent (Claude), and the paid tools it calls run against a seller API that an AI agent may also autonomously pay and query. A human owner (the repository author) controls the wallet, the spending guard, and the seller side. Treat any output from these tools as a signal, not a guarantee — see "What this does not do" below.
Related MCP server: @hpp-io/x402-mcp-bridge
Tools
Tool | Price | Answers |
| Free | The live product catalog: ids, paths, prices, input/output schemas. |
| Fetched live via | Is this npm/PyPI package safe to install? Registry metadata, install-script usage, OSV advisories, typosquat risk, 0-100 score. |
| Fetched live via | Will this GitHub repo merge an AI-authored or external PR? Stated AI/contribution policy plus historical merge rates, 0-100 score. |
| Fetched live via | Is this public page accessible? WCAG 2.1/2.2 AA audit with headless Chromium + axe-core: violations by impact, per-rule detail, 0-100 score. |
| Fetched live via | Solves a scheduling/roster problem (gym rosters, dance-school timetables, shift plans) with OR-Tools CP-SAT: assigns resources to slot demands honoring availability, tags, and constraints. Takes a JSON body ( |
| Fetched live via | Structural validation of a Mexican RFC (persona física/moral, embedded date, SAT check digit) plus Article 69-B (EFOS) blacklist status. Takes a JSON body ( |
| Fetched live via | Validates an 18-digit Mexican CLABE: check digit, bank identification, plaza, and account number. Takes a JSON body ( |
| Fetched live via | Verifies a Mexican electronic invoice (CFDI) with SAT: status (vigente/cancelado), cancelability, and EFOS validation. Takes a JSON body ( |
Prices shown above are a snapshot at the time of writing and are never read
by the code — always call list_products for the current catalog.
What this does not do
Scores are a heuristic signal built from public metadata (registry fields, advisory databases, repo history). They are not a security audit and not a guarantee that a package is safe or that a PR will merge. Use them as one input among several.
Install
npx (no local checkout)
{
"mcpServers": {
"x402-tools": {
"command": "npx",
"args": ["-y", "github:vera-agent-dev/x402-tools"],
"env": {
"X402_BASE_URL": "https://x402-api-24223879872.us-east1.run.app"
}
}
}
}Once published to npm, the same config works with "args": ["-y", "x402-tools"].
Local checkout
git clone https://github.com/vera-agent-dev/x402-tools
cd x402-tools
npm install
npm run build{
"mcpServers": {
"x402-tools": {
"command": "node",
"args": ["/absolute/path/to/x402-tools/dist/index.js"],
"env": { "X402_BASE_URL": "https://x402-api-24223879872.us-east1.run.app" }
}
}
}Claude Code plugin
This repo ships .claude-plugin/plugin.json, so it installs like any other
Claude Code plugin pointing at this GitHub repository.
Plain MCP JSON config
Any MCP-compatible client can use the .mcp.json in this repo directly, or
copy its mcpServers.x402-tools block into your own client config.
.mcp.json is a tracked file. Its X402_BUYER_PRIVATE_KEY value is the
literal string ${X402_BUYER_PRIVATE_KEY} — an env-var reference, not a
real key — for clients that support that substitution. Never replace it
with an actual private key and commit that. If your client doesn't
support ${VAR} substitution in MCP config, set the env var in your own
untracked client config instead of editing this file in place.
Payment flow
list_productsis always free and always live — no wallet needed.Calling a paid tool without
X402_BUYER_PRIVATE_KEYset performs the request; on a 402 it returns the decoded challenge (amount, asset, network,payTo) as structured output, plus a short message on how to pay — either by setting the env var, or manually using the snippets insnippets/.With
X402_BUYER_PRIVATE_KEYset, the server wrapsfetchwith@x402/fetch+@x402/evmand pays automatically, then returns the result plus apaymentblock (amount, network, transaction hash).A spending guard,
X402_MAX_PRICE_USD(default0.30), refuses to pay any single challenge above that amount and returns the challenge instead — the same as running with no key configured. This is enforced twice: a local pre-check on the accept matching your configured network, and (authoritatively) via the x402 client's ownsetSpendControls, which caps whatever accept it actually selects and signs — so a seller offering severalacceptsacross networks/assets can't bypass the cap.
Environment variables
Variable | Default | Purpose |
|
| Base URL of the seller API. |
| (unset) | Private key of the paying wallet. Omit to run in challenge-only mode. |
|
| CAIP-2 network id. Use |
|
| Spending guard: max USD per single paid call. |
Security: X402_BUYER_PRIVATE_KEY
This is your own wallet's private key, not this project's. Use a
dedicated, low-balance wallet you top up only with what you're willing to
spend automatically — never your main wallet. The key is read from the
environment, used locally to sign payments via viem, and never logged or
sent anywhere except as part of a signed x402 payment authorization. Paid
tool arguments (e.g. the package name or repo you're looking up) are sent to
the seller API as ordinary query parameters for GET tools, or as a JSON body
for POST tools (schedule_solve and the mx_* tools, which carry
PII-adjacent fields such as RFC/CLABE/CFDI data) — don't pass anything
sensitive regardless.
Publishing (maintainer notes)
npm: package name
x402-tools,npm publishonce ready; updateserver.jsonand this README's npx snippet to drop thegithub:prefix.MCP Registry:
server.jsonfollows theio.github.vera-agent-dev/x402-toolsnaming convention from the registry schema.
Dev / test
npm install
npm test # vitest, mocks fetch — no network, no spend
npm run build # tsc -> dist/
npm run dev # tsx src/index.ts
npm run smoke # end-to-end stdio smoke test against a running seller APIOpen TODOs
The seller API currently runs on Base Sepolia (testnet USDC); mainnet switch pending.
Not yet published to npm; install via
npx github:vera-agent-dev/x402-toolsuntil then.
Links
x402 protocol: https://www.x402.org/
MCP Registry: https://github.com/modelcontextprotocol/registry
License
MIT — see LICENSE.
Available Tools
3 toolslist_productsList x402 productsA
Free. Fetches the live product catalog from the x402-tools seller API (prices, ids and schemas are never hardcoded here).
Inputs: no named parameters.
package_trust_checkpackage_trust_checkA
Install-safety signals for an npm or PyPI package: registry metadata, install-script usage, OSV advisories, typosquat risk, and a 0-100 trust score. (price: $0.05 USDC per call) [catalog: bundled fallback, API was unreachable at startup]
Inputs: name (required), ecosystem (required).
repo_merge_lookuprepo_merge_lookupA
Predicts whether a GitHub repository will merge an AI-authored or external pull request, based on its stated AI/contribution policy and historical merge rates. (price: $0.05 USDC per call) [catalog: bundled fallback, API was unreachable at startup]
Inputs: repo (required).
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
TDQS
Scored across 3 tools
Each tool addresses a completely distinct task: fetching product catalog, checking package trust, and predicting PR merges. There is zero functional overlap, so an agent would never confuse which tool to select.
Naming mixes conventions: 'list_products' follows verb_noun, while 'package_trust_check' and 'repo_merge_lookup' are noun_verb (object+verb). All use underscores, but the verb placement is inconsistent, which slightly reduces predictability.
With only 3 tools, the server is lean and each tool serves a clear, standalone purpose. The count is appropriate for a specialized utility server, though it borders on minimal.
The tools are independent utilities, not part of a lifecycle workflow, so there are no obvious missing CRUD operations. Each tool appears self-contained, though the domain is broad and could potentially support additional related functions (e.g., package details, repo history), but nothing critical is missing.
Maintenance
Related MCP Connectors
Machine-payable API infrastructure for autonomous agents. Per-call x402 settlement in USDC.
x402 pay-per-call APIs over MCP, settled in USDC on Base for autonomous agents and developers.
Pay-per-use weather, environment, finance, and on-chain intelligence tools for AI agents via x402.
Pay-per-call x402 services for agents: risk scoring, DeFi, wallet and market data.
Related MCP Servers
- AlicenseAqualityCmaintenanceTrust layer for AI agents: 22 pay-per-call x402 tools for scam and fraud checks, payment pre-flight, wallet screening, policy approvals, contact extraction and company research. Free preview with no wallet; paid calls settle in USDC on Base.22332 npmMIT

@hpp-io/x402-mcp-bridgeofficial
AlicenseNot gradedqualityBmaintenanceEnables AI agents to autonomously pay for and discover services using HPP USDC.e over the x402 protocol, without API keys or manual signing.695 npmApache 2.0- AlicenseNot gradedqualityCmaintenanceEnables agents to discover, sample, and pay per call in USDC for live threat intelligence, ZK proof generation, and arbitrage signals via the x402 protocol.MIT
- FlicenseNot gradedqualityDmaintenanceEnables AI agents to access 25 pay-per-request intelligence APIs using USDC micropayments via the x402 protocol, without API keys or accounts.-