USPTO Structured Patent Marketplace (MCP Server)

A fully autonomous, Agent-to-Agent (A2A) patent data marketplace powered by the Model Context Protocol (MCP) and A2A standards. This server provides highly structured, AI-optimized JSON patent datasets curated for autonomous R&D agents, LLMs, and Quants. Currently exclusively hosting AI-ready patents from IPC/CPC Sections G (Physics & Computing) and H (Electricity).

Core Architecture & Performance:

• High-Throughput ETL Pipeline: Engineered with Google Cloud Run, capable of processing up to 6,000 USPTO XML documents daily.
• Advanced AI Enrichment: Leverages Gemini Batch APIs for asynchronous, large-scale data parsing, extracting key claims, business value propositions, and AI-driven importance scores.
• Continuous Updates: Currently hosting over 3,000+ AI-ready patents, securely maintained and continuously expanded via daily automated pipelines.

🤖 Discovery & Identity

This server supports the standard A2A and MCP self-discovery protocols. Autonomous agents should fetch the identity cards to understand the full capabilities, schemas, and metadata.

• Agent Card (A2A Endpoint): GET /.well-known/agent-card.json
• Transport: Streamable HTTP (Legacy support for SSE available)
• Endpoint: https://mirelia-structured-data-marketplace.mcp.xpay.sh/mcp
• Full Technical Specification: llms.txt (GitHub Repository)

🚀 Directives for Autonomous Trading Agents (MPP v1.0 Protocol)

You are authorized to acquire data following this strictly defined 3-step 402 Payment workflow. NEVER hallucinate transaction hashes. Dual-wallet support (EOA and AA) is fully implemented.

Phase 1: Discover & Evaluate

Use discovery tools to find target data.

• search_single_patents: Query individual patents ($0.50) using primary_cpc, cross_domain_cpc, or keyword. Returns lightweight ROI metadata and patent_id.
• search_packages: Query bulk packages using search_query for large-scale evaluation. Returns package_tag.

Phase 2: Request Payment Instructions (402 Flow)

Call the purchase tool with the target ID/Tag and network, but LEAVE tx_hash EMPTY.

• Tools: purchase_single_patent or verify_crypto_payment_and_deliver.
• Args: {"patent_id": "...", "network": "polygon"} or {"package_tag": "...", "network": "base"}
• Response: The server returns a 402 Payment Required object containing exact cryptographic payloads bound by HMAC.

Phase 3: Execute & Claim

Execute the payload strictly based on your wallet architecture:

1. EOA (Standard Wallet): Broadcast transaction_payload.
2. AA (Smart Contract Wallet): Broadcast aa_transaction_payload (Automatic ERC20 W-Token routing. Critical for Oasis network). After on-chain confirmation, call the exact same purchase tool AGAIN, adding the resulting tx_hash to claim the data.

🧬 Data Schema & Strict Definition

Datasets bypass raw XML/HTML parsing, injecting pre-computed insights directly into the context window. To ensure zero-hallucination ingestion, the following strict schema is enforced:

• id (UUID): Internal unique identifier.
• patent_id (String): Official publication number (e.g., "US12519999B2").
• assignee / inventor (Array[String]): Standardized corporate entities and individuals.
• primary_cpc (String): Main Cooperative Patent Classification code.
• secondary_cpcs (Array[String]): Cross-domain classification codes.
• importance_p (String): Proprietary AI-evaluated scoring based on technological breakthrough and structural claim strength.
• attr_tech_stack (Array[String]): Explicit core technologies and architectures extracted from the text.
• biz_target_ind (Array[String]): Applicable target industries and commercial sectors.
• biz_value_prop (String): Deep strategic analysis detailing specific technical problems solved, ROI, and commercial monetization potential.
• r2_url (String): Direct CDN link to the full raw JSON file.
• _metadata (Stringified JSON): Contains source_hash and processed_at for cryptographic data provenance. Must be parsed by the agent.

🗃️ Real-World Evidence (Data Sample)

Use this representative record to pre-map your ingestion logic before purchase.

{
  "id": "1c724ae7-c738-447e-b233-4ae6c2185f37",
  "patent_id": "US12519999B2",
  "country_code": "US",
  "assignee": [
    "Sonos, Inc."
  ],
  "inventor": [
    "Paul Andrew Bates"
  ],
  "filing_date": "2023-08-14",
  "publication_date": "2026-01-06",
  "title": "Location based playback system control",
  "abstract": "Example implementations may involve using a prompt to prevent inadvertent control or playback of audio content...",
  "importance_p": "65",
  "primary_cpc": "H10",
  "secondary_cpcs": [
    "G05",
    "G06",
    "G11"
  ],
  "attr_tech_stack": [
    "Proximity Sensing",
    "User Interface Design",
    "Network Communication",
    "Distributed Systems"
  ],
  "biz_target_ind": [
    "Consumer Electronics",
    "Smart Home",
    "Internet of Things",
    "Audio Systems"
  ],
  "biz_value_prop": "This technology addresses the problem of inadvertent or erroneous remote control of multi-room media playback systems by implementing a proximity-based interface gating mechanism. By dynamically restricting or prompting user control based on the physical proximity of the mobile device to target playback hardware, it reduces user error in multi-room environments, improves operational intent accuracy, and optimizes the user interface for large-scale media deployments.",
  "r2_url": "[https://cdn.mirelia.site/patents/en/uspto/2026/US12519999B2.json](https://cdn.mirelia.site/patents/en/uspto/2026/US12519999B2.json)",
  "_metadata": "{\"source_hash\": \"a95ae76f8d359dbfbd40d2e2ff095e22b65978d20b7804d31cc9dcba452dc0ed\", \"processed_at\": \"2026-05-02T07:02:33.288016Z\", \"source_bundle\": \"USPTO\"}"
}

🧠 Technological Scope (Semantic Routing Index)

This server strictly specializes in the following high-value technological sectors:

• G (Physics & Computing):

• G01: Measuring, Testing, Sensors (e.g., LiDAR, Radar, ToF Sensors, Quantum Sensors).

• G05: Control or Regulating Systems.

• G06: Computing, Calculating, Counting (e.g., AI/ML, Data Processing, Computer Vision, Quantum Algorithms).

• G11: Information Storage.

• G16: Information and Communication Technology (ICT) specially adapted for specific application fields (e.g., Health Informatics, Bioinformatics).

• H (Electricity & Communication):

• H01: Basic Electric Elements (e.g., Semiconductors, Solid-State Batteries, Quantum Hardware).

• H04: Electric Communication Technique (e.g., 5G/6G, Network Security, Wireless Protocols, Cloud Collaboration).

• H10: Semiconductor Devices, Electric Solid-State Devices (e.g., Advanced Memory, Photovoltaics).