dev.nexbid/discovery MCP Server

Search and discover advertiser products through an open marketplace for AI agents. The repository description frames this as an open infrastructure for agent-native commerce, focusing on standardized, auditable, and interoperable protocols for AI agent-driven commerce.

🛠️ Key Features

Open marketplace approach for advertiser product discovery by AI agents
Protocol Commerce initiative scope: standardized, auditable, interoperable protocols
Includes protocol specification materials and developer artifacts

🚀 Use Cases

Discover advertiser products for agent workflows
Support agent-driven commerce where discovery relies on open protocol specifications

⚡ Developer Benefits

Specification, SDKs, and technical manifesto are included (as described)
AdCP v0.1.0 specification provides JSON schemas and examples

⚠️ Limitations

Material provided describes discovery commerce and specification content; no runtime behaviors or API/tool details beyond toolCount: 19 are stated

Protocol Commerce

The open infrastructure for agent-native commerce.

Protocol Commerce is an open initiative to create standardized, auditable, and interoperable protocols for AI agent-driven commerce. This repository contains the specification, SDKs, and technical manifesto.

Directory	Description
`manifesto/`	Why open protocols matter for commerce — principles, landscape analysis, and rationale
`adcp-spec/`	AdCP v0.1.0 — Agentic Discovery Commerce Protocol specification, JSON schemas, and examples
`adcp-sdk-typescript/`	TypeScript SDK: types, validators, client, and scoring reference implementation
`webmcp-reference/`	WebMCP browser integration: expose commerce tools to in-browser AI agents via `navigator.modelContext`
`amdp-spec/`	AMDP v0.1.0 — Agent Mandate Discovery Protocol: cross-vertical authorization spec, JSON schemas, and example mandates
`agentic-kit/`	Agentic Kit — the trust layer: building blocks for verifiable + auditable AI agents
`lean-verification/`	Lean 4 formal verification — 47 theorems proving auction/budget/wallet/policy correctness, `lake build`-checkable

Quick Start

bash

npm install @protocol-commerce/adcp-sdk

typescript

import { AdcpClient } from '@protocol-commerce/adcp-sdk';

const client = new AdcpClient({
  serverUrl: 'https://mcp.nexbid.dev',
  apiKey: 'your-api-key',
});

const results = await client.search({
  query: 'organic olive oil',
  intent: 'purchase',
  geo: 'CH',
});

The Problem

AI agents are becoming the primary interface for product discovery and purchase. The protocols powering this shift — OpenAI's ACP, Google's UCP — are proprietary and platform-controlled. Publishers have no representation. There is no open standard.

Protocol Commerce fills this gap.

Read the full Technical Manifesto →

Key Principles

Open by Default — MIT-licensed, auditable, forkable
Publisher Parity — Publishers are first-class protocol participants
Privacy Native — Context signals, not user profiles
Transparent Scoring — Public formula: score = 0.4×bid + 0.4×relevance + 0.2×quality

All 7 principles →

Reference Implementation: Nexbid

Nexbid is the production-grade reference implementation of Protocol Commerce, built by digital opua GmbH (CHE-435.289.702, Switzerland).

Metric	Status
MCP Integration	Native — full commerce lifecycle
Formal Verification	Lean 4 — core security properties proven
Privacy	Cookie-free, Swiss-hosted, nDSG + GDPR compliant
AI Scoring	Multi-provider, model-agnostic
Phase	Production-grade closed beta

Connect Any LLM

json

{
  "mcpServers": {
    "nexbid": {
      "url": "https://mcp.nexbid.dev/mcp",
      "transport": "streamable-http"
    }
  }
}

Any MCP-compatible LLM (Claude, GPT-4, Gemini) can directly search, browse, and purchase products through the Nexbid marketplace.

Why Formal Verification?

Nexbid is the only commerce platform whose core security properties are mathematically proven in Lean 4 — not just tested, but proven correct for all possible inputs.

→ Technology details · How Nexbid compares

License

MIT — Use it, fork it, build on it. No strings attached.

query	string	required	Natural language product or recipe query
content_type	string	optional	Filter by content type: product, recipe, or all (default)
intent	string	optional	User intent for the search
budget_max_cents	integer	optional	Maximum budget in cents (e.g. 20000 for CHF 200)
budget_min_cents	integer	optional	Minimum budget in cents
currency	string	optional	Currency for budget filtering
geo	string	optional	ISO 3166-1 alpha-2 country code (default: CH)
category	string	optional	Filter by product category
brand	string	optional	Filter by brand name
max_results	integer	optional	Maximum number of results (1-50, default: 10)
previous_queries	array	optional	Previous queries in this search session for multi-turn refinement (oldest first, max 10). Example: ["running shoes", "waterproof only"]
agent_id	string	optional	Agent identifier — used for analytics-attribution (e.g. "claude-3.5", "gpt-4o", "perplexity"). Optional but strongly recommended so publishers/advertisers can see which AI agents drive their traffic.
session_id	string	optional	Session token to correlate multiple tool-calls from the same conversation. Optional. Useful for multi-turn analytics.

product_id	string	required	Product UUID
agent_id	string	optional	Agent identifier for analytics-attribution.
session_id	string	optional	Session token for multi-turn correlation.

product_id	string	optional	Product UUID to purchase. Mutually exclusive with line_items.
quantity	integer	optional	Quantity for single-item purchase (default: 1, ignored in cart-mode)
line_items	array	optional	Array of line items for cart purchase. Mutually exclusive with product_id.
cart_id	string	optional	Optional cart UUID. Server generates one if omitted.
cart_total_max_cents	integer	optional	Cart-level hard-bound on total amount (only with line_items)
cart_ttl_seconds	integer	optional	Cart mandate lifetime in seconds (default 900, max 3600)
checkout_mode	string	optional	Checkout mode. Default: prefill_link. wallet_pay requires a connected wallet with active mandate.
publisher_content_id	string	optional	Optional UUID of the publisher content (recipe, editorial, review, etc.) that surfaced this product to the user. When set, a successful purchase emits a premium-tier enriched_snippet_impression and the publisher gets credited 90% of the conversion value. Pass-through from nexbid_search, where it is returned alongside each product as `match_publisher_content_id` (when applicable).

query	string	required	Natural language product query
category	string	optional	Filter by product category
brand	string	optional	Filter by brand name
budget_max_cents	integer	optional	Maximum budget in cents
geo	string	optional	ISO 3166-1 alpha-2 country code
max_results	integer	optional	Maximum results (1-50, default: 10)

publisher_id	string	optional	Filter by publisher UUID
slot_type	string	optional	Filter by slot type
pricing_model	string	optional	Filter by pricing model
geo	string	optional	ISO 3166-1 alpha-2 country code
max_results	integer	optional	Maximum results (default: 20)

dev.nexbid/discovery

dev.nexbid/discovery MCP Server

🛠️ Key Features

🚀 Use Cases

⚡ Developer Benefits

⚠️ Limitations

Topics

nexbid_search

nexbid_product

nexbid_categories

nexbid_purchase

nexbid_order_status

list_products

get_product

list_inventory

get_inventory_item

create_media_buy

submit_creatives

activate

pause

cancel

list_media_buys

get_campaign_report

get_compliance_status

settle

track_enriched_snippet

Protocol Commerce

Contents

Quick Start

The Problem

Key Principles

Reference Implementation: Nexbid

Connect Any LLM

Why Formal Verification?

Sponsors

License

Links

Install

Remote endpoint

Related servers

inventory_id	string	required	Target inventory slot UUID
advertiser_id	string	required	Advertiser account UUID
bid_cents	integer	required	Bid amount in cents
pricing_model	string	required	Pricing model for the buy
creative_ref	string	optional	Creative asset URL or reference
creative_data	object	optional	Creative metadata (JSON)
budget_cents	integer	required	Total budget in cents
start_date	string	optional	Campaign start date (ISO 8601)
end_date	string	optional	Campaign end date (ISO 8601)
brand	string	optional	Advertiser brand name (for rules check)
category	string	optional	Product category (for rules check)
quality_score	number	optional	Quality score 0-1 (default: 0.5)
context_relevance	number	optional	Context relevance 0-1 (default: 0)
campaign_id	string	optional	Parent campaign UUID. When set, the campaign's default pacing + frequency cap cascade into this buy unless explicitly overridden.
pacing_strategy	string	optional	Override pacing strategy. Falls back to campaign default, then "even".
frequency_cap_session	any	optional	Per-session impression cap for this buy. NULL means defer to inventory cap. Falls back to campaign default, then 3.
agent_mandate	any	optional	Optional signed agent mandate (ADR-066). Verified by the auction-entry gate.
agent_id	string	optional	Optional agent identifier (telemetry; feeds ADR-066 behavioral-risk shadow logging).

media_buy_id	string	required	Media buy UUID
creative_type	string	required	Creative type (e.g. "banner", "native", "snippet")
creative_ref	string	optional	Creative asset URL
creative_data	object	optional	Creative metadata (JSON)