AI agent commerce

AI agent commerce: MCP server, API, SDKs, and ChatGPT.

Nexez Buyer is the ChatGPT app. Connect other agents to the ai.nexez/commerce MCP server at https://nexez.app/mcp, or use the Nexez API and SDKs.

Use Nexez Buyer in ChatGPT Connect another assistant

Developer install options

Ask Nexez Buyer

Find a highly rated photographer near Austin for under $500.
Compare agent-ready website design offers for price and delivery time.
Check whether my budget and timeline fit this seller's negotiation rules.

Nexez Buyer does not place orders, take payments, reserve inventory, or contact sellers.

Other assistants

Connect to the hosted MCP endpoint.

ChatGPT users open Nexez Buyer. Claude, Codex, and other MCP clients connect to the hosted server. The official MCP Registry record is ai.nexez/commerce, version 1.1.0. Its remote URL is https://nexez.app/mcp, transport streamable-http.

Assistants connect to this URL. Opening it in a browser is not the setup flow: the server accepts MCP requests, not a web page.

Setup

1In your assistant, add an MCP server that uses streamable HTTP.
2Set the server URL to https://nexez.app/mcp.
3Leave authentication empty. Public discovery does not need an API key.
4The official MCP Registry record for that URL is ai.nexez/commerce, version 1.1.0.
5Call nexez_search or nexez_directory to find listings, nexez_get_page to read one, and the validate tools only as dry runs.

https://nexez.app/mcp

Nexez Buyer

Live on ChatGPT · v0.1.0

Find your next offer. In ChatGPT.

Find public products and services from Nexez merchants, compare offers, and check whether your budget and timeline fit before taking the next step. No code or API keys needed to install.

Use Nexez Buyer in ChatGPT List your business

Start with one conversation

1. Open the listing. Follow the install steps in ChatGPT. Sign in if prompted.
2. Tell Nexez Buyer what you need. Include your budget, location, and timeline.
3. Compare and check fit. Review the available offers and any missing requirements.

Nexez Buyer does not place orders, take payments, reserve inventory, or contact sellers. Checking an offer does not guarantee availability or seller acceptance.

Install paths

Use tools, instructions, or both.

The OpenClaw plugin gives agents callable tools. The skill gives agents the discovery logic and safety rubric.

Native tool plugin

Nexez OpenClaw Plugin

Native tools for search, listing fetch, checkout validation, and negotiation handoff.

v0.2.2

Terminal

openclaw plugins install clawhub:@nexez/openclaw-nexez

Discovery skill

Nexez Agent Discovery

Instructions and rubrics for discovering, comparing, and safely acting on Nexez agent listings.

v0.1.2

Terminal

openclaw skills install nexez-agent-discovery

Agent workflow

From request to safe handoff.

Nexez is designed for agents that need to compare offers, explain why a seller fits, and only then request approval for checkout or negotiation.

Search by buyer intent

Use Nexez Buyer, the hosted MCP server, or the search API with natural buyer language.

Find a Chicago consultant for a one-week growth audit.

Read the listing manifest

Fetch agent.json or llms.txt for seller context, offers, FAQs, policies, and direct actions.

/{slug}/agent.json

Validate before acting

Dry-run checkout or negotiation so the agent can confirm the handoff before money or contact occurs.

POST /api/checkout { dryRun: true }

Hand off with approval

After user approval, open checkout, submit negotiation terms, or route to the seller action URL.

checkout, negotiate, book, or contact

Public machine surfaces

Everything points to one source of truth.

These endpoints stay on the public agent runtime so listings remain fast, crawlable, and separate from the authenticated app.

Developer docs

Hosted MCP

Registry ai.nexez/commerce version 1.1.0. Streamable HTTP, no API key for public tools.

https://nexez.app/mcp

Find matching listings and offer-level actions from a buyer request.

https://nexez.app/api/agent-search?q={query}

Index

List published listings, manifests, readiness, offers, and checkout URLs.

https://nexez.app/agent-pages.json

LLM guide

Plain-text global context for agents that read llms.txt first.

https://nexez.app/llms.txt

OpenAPI

Schemas and operation IDs for agent builders and action catalogs.

https://nexez.app/openapi.json

MCP catalog

Discovery catalog for listings exposing MCP-compatible resources.

https://nexez.app/.well-known/mcp.json

ARD catalog

Agentic Resource Discovery manifest so registries can index every Nexez MCP server by capability.

https://nexez.app/.well-known/ai-catalog.json

Capabilities

Compact manifest describing Nexez endpoints, safety posture, and distribution.

https://nexez.app/.well-known/nexez.json

agent-access.json

public discovery contract

{
  "schema_version": "nexez.agent-access.v1",
  "docs_url": "https://nexez.ai/agents",
  "runtime_base_url": "https://nexez.app",
  "chatgpt_app": {
    "name": "Nexez Buyer",
    "version": "0.1.0",
    "url": "https://chatgpt.com/plugins/plugin_asdk_app_6a9759b229008191bdb36577a8601b81"
  },
  "mcp": {
    "registry": "ai.nexez/commerce",
    "version": "1.1.0",
    "endpoint": "https://nexez.app/mcp",
    "transport": "streamable-http"
  },
  "openclaw": {
    "plugin": {
      "name": "@nexez/openclaw-nexez",
      "version": "0.2.2",
      "install": "openclaw plugins install clawhub:@nexez/openclaw-nexez"
    },
    "skill": {
      "slug": "nexez-agent-discovery",
      "version": "0.1.2",
      "install": "openclaw skills install nexez-agent-discovery"
    }
  },
  "sdks": {
    "typescript": {
      "name": "@nexez/agent-sdk",
      "version": "0.3.1",
      "status": "published",
      "install": "npm install @nexez/agent-sdk",
      "package": "https://www.npmjs.com/package/@nexez/agent-sdk",
      "source": "sdk/typescript"
    },
    "python": {
      "name": "nexez-agent-sdk",
      "module": "nexez_agent_sdk",
      "version": "0.3.1",
      "status": "published",
      "install": "python -m pip install nexez-agent-sdk",
      "package": "https://pypi.org/project/nexez-agent-sdk/",
      "source": "https://pypi.org/project/nexez-agent-sdk/"
    }
  },
  "examples": "https://github.com/nexez-ai/nexez-agent-examples/tree/main",
  "endpoints": {
    "search": "https://nexez.app/api/agent-search?q={query}",
    "index": "https://nexez.app/agent-pages.json",
    "llms": "https://nexez.app/llms.txt",
    "openapi": "https://nexez.app/openapi.json",
    "mcp": "https://nexez.app/mcp",
    "mcp_catalog": "https://nexez.app/.well-known/mcp.json",
    "ard_catalog": "https://nexez.app/.well-known/ai-catalog.json"
  }
}

SDKs

Helpers for agent builders.

The TypeScript SDK is live on npm, and the Python SDK is live on PyPI for agent runtimes that prefer scripts, workers, notebooks, or server automation.

Published npm

Nexez Agent SDK · v0.3.1

npm

npm install @nexez/agent-sdk

Typed client helpers for agent search, manifest fetch, checkout dry-run, and negotiation handoff.

Published PyPI

Nexez Python Agent SDK · v0.3.1

PyPI

python -m pip install nexez-agent-sdk

Dependency-free Python helpers for agent search, manifest fetch, checkout dry-run, and negotiation handoff.

agent-sdk.ts

search · manifest · dry-run

import { createNexezClient } from '@nexez/agent-sdk'

const nexez = createNexezClient()

const matches = await nexez.search('book a strategy session next week', {
  location: 'Chicago, IL',
  limit: 5,
})

const first = matches.results[0]
const page = await nexez.getAgentPage(first.page.slug)

await nexez.validateCheckout({
  slug: first.page.slug,
  offer: first.offer?.key ?? 'services-0',
  query: 'Buyer wants a strategy session next week.',
  buyerAgent: 'buyer-agent',
})

Agent examples

Copy the full buyer flow.

The examples show agents how to search, fetch structured context, validate handoffs, stop for buyer approval, submit negotiation terms, and poll status. The public source is the nexez-ai examples repository.

Buyer approval

examples/agents/python/buyer_approval.py

Build a consent card with seller, offer, terms, risk notes, dry-run result, and explicit approval copy.

Find and validate

examples/agents/python/find_and_validate.py

Search by buyer intent, fetch the listing manifest, then dry-run checkout or negotiation before any action.

Location shortlist

examples/agents/python/location_shortlist.py

Rank nearby or remote matches, compare actionability, validate the top offer, and stop for buyer approval.

Submit negotiation

examples/agents/python/submit_negotiation.py

Prepare terms, validate rules, stop for buyer approval, submit, then poll the returned status URL.

TypeScript parity

examples/agents/typescript

The same agent workflows are mirrored for builders using the published TypeScript SDK.

agent-sdk.py

search · manifest · negotiation dry-run

from nexez_agent_sdk import create_client

nexez = create_client(buyer_agent="buyer-agent")

matches = nexez.search(
    "book a strategy session next week",
    location="Chicago, IL",
    limit=5,
)

first = matches["results"][0]
page = nexez.get_agent_page(first["page"]["slug"])

nexez.validate_negotiation(
    slug=page["page"]["slug"],
    offer=first.get("offer", {}).get("key", "services-0"),
    query="Buyer wants a strategy session next week.",
    budget="USD 2100",
)

Safety posture

Open for discovery. Careful for actions.

Public discovery requires no secret.

Use dry-run validation before checkout or negotiation.

Do not infer private seller rules from public constraints.

Ask for buyer approval before spending money or contacting a seller.

For buyer agents

Search by outcome, location, budget signal, or service type. Nexez returns listings with structured offers, trust signals, and direct next actions.

For seller listings

Public listings remain on the agent runtime, while the marketing site and dashboard stay on their own domains. That split protects crawlability and signed-in workflows.

For registries

The ARD catalog at /.well-known/ai-catalog.json lists every Nexez MCP server, its capabilities, and sample queries. A discovery service can index the network from that single file.

Questions about Nexez Buyer

What is Nexez Buyer?

Nexez Buyer is a ChatGPT app that searches public Nexez listings, compares offers, and checks whether a budget and timeline fit. It does not place orders, take payments, reserve inventory, or contact sellers.

How do I open Nexez Buyer in ChatGPT?

Use the Nexez Buyer button on this page. It opens the ChatGPT app listing. Sign in if ChatGPT asks you to, then describe what you need, including budget, location, and timeline.

Does Nexez Buyer place an order?

No. Nexez Buyer searches and compares public offers, and it can dry-run a check of price, requirements, or proposed terms. That check does not charge a payment method, create an order, reserve inventory, submit terms, or contact a seller.

How do other assistants connect?

Add a streamable HTTP MCP server pointed at https://nexez.app/mcp. No API key is required for the public tools. The official MCP Registry record is ai.nexez/commerce, version 1.1.0.

Where is the developer install path?

This page lists the OpenClaw plugin, the OpenClaw skill, the TypeScript and Python SDKs, and the public agent examples. Developer docs, buyer approval, and the platform docs are linked below.

Related Nexez pages

Ready for the agent web.

Publish a seller listing and let agents discover the offer through the clean runtime.

Create a listing