Sponsored by Deepsite.site

XGR.Network MCP

创建者
xgr-network2 months ago
XGR.Network MCP is a remote MCP server that gives AI agents access to the XGR stack. It supports XDaLa workflow preparation, XGRChain evidence lookup, Explorer data access and programmable process automation through a public streamable HTTP endpoint.
概览

XGR MCP Gateway

XGR MCP Gateway provides AI-native access to the XGR.Network stack through the Model Context Protocol. It connects MCP-compatible agents such as Claude, ChatGPT, IDE assistants and custom hosts to XGRChain, XDaLa sessions, Explorer data, XRC standards and owner-reviewed on-chain action preparation.

The gateway is designed for agent-assisted workflow creation: agents can inspect live chain evidence, reason over XDaLa process state, draft XRC artifacts and prepare handoffs that a user reviews and signs locally.

The gateway never requests, receives, stores or controls user or third-party private keys and cannot sign on behalf of users. User write intents are prepared as handoffs and signed locally. An optional starter-gas service may use a dedicated server-controlled funding key solely to send fixed XGR gas grants from the service wallet.

Public MCP endpoints

Mainnet:

https://mcp.xgr.network/mcp

Testnet:

https://mcp.testnet.xgr.network/mcp

Use mainnet to inspect real XGRChain and XDaLa activity. Use testnet to safely draft, validate and experiment with agent-driven workflows before production use.

MCP client configuration

{
  "mcpServers": {
    "xgr-mainnet": {
      "type": "streamable-http",
      "url": "https://mcp.xgr.network/mcp"
    },
    "xgr-testnet": {
      "type": "streamable-http",
      "url": "https://mcp.testnet.xgr.network/mcp"
    }
  }
}

For clients that still expect http instead of streamable-http, use the same endpoint URL and select the HTTP/remote MCP transport offered by your client.

What agents can do

  • Read chain state: inspect live XGRChain status, blocks, account state and transaction evidence.
  • Discover official XGR resources: use get_xgr_network_info for canonical XGR.Network, XGRChain, XDaLa, RPC, Explorer, MCP, documentation and ecosystem metadata.
  • Inspect XDaLa sessions: find, list and explain sessions, process steps, payloads, receipts and execution outcomes.
  • Search Explorer data: query account history, value transfers, block history, transaction statistics and session analytics.
  • Work with XRC standards: explore XRC-137 rules, XRC-729 orchestrations, process graphs, reuse patterns and failure analytics.
  • Draft process artifacts: prepare XRC-137/XRC-729 artifacts and bundles against built-in schemas and validation rules.
  • Prepare owner-signed actions: create review-and-sign handoffs for deployments and session starts without exposing private keys.
  • Request starter gas where enabled: fund an eligible low-balance address with one fixed 1 XGR grant before deployment, session start, contract calls or other on-chain actions.

Official network metadata

get_xgr_network_info returns versioned, machine-readable metadata for XGR.Network and its ecosystem. Agents should use it when users request official project information, network configuration, RPC or Explorer endpoints, MCP connection details, documentation, XRC standard context or source repositories.

get_chain_status remains the live JSON-RPC status tool and additionally returns compact official entry points for the connected XGRChain mainnet.

Why XGR.Network MCP

XGR.Network MCP is built around deterministic process infrastructure rather than generic chat automation. It gives agents structured access to the XGR stack while keeping user signing, user custody and final approval outside the gateway.

This makes it suitable for:

  • AI-assisted Web3 workflow design
  • XDaLa process preparation and inspection
  • compliance-oriented process evidence
  • deterministic validation and execution flows
  • EVM-compatible process automation
  • agent interfaces for XGRChain data and XRC standards

Documentation

Full reference lives in the central XGR documentation:

Self-hosting

The gateway can be operated against your own XGRChain RPC and Explorer instance. Self-hosting requires an Explorer deployment with a read-only Postgres mirror for transaction search and session analytics tools.

Typical setup flow:

npm install
npm run typecheck
npm run build
npm run start:http

Required runtime configuration is documented in Setup & Configuration.

Security model

  • The gateway never requests, receives, stores or controls user or third-party private keys.
  • User transactions are prepared as handoffs and signed locally by the user or the user's custody setup.
  • The optional starter-gas service is a narrow exception: it signs only transfers from a dedicated server-controlled funding wallet and cannot sign on behalf of users.
  • Starter-gas grants are fixed at 1 XGR, limited by recipient balance, one confirmed grant per address, hourly and daily caps, and a bounded retry policy.
  • Starter-gas grant state is stored atomically in SQLite with reserved, broadcast, confirmed and failed states.
  • Production user signing remains under the control of the user's wallet or custody setup.

License

Licensed under the Apache License 2.0.

Mainnet XGR purchase tools

The optional purchase tools are mainnet-only, disabled by default, and are never registered by testnet. Enable them only with XGR_PURCHASE_TOOLS_ENABLED=true, XGR_PURCHASE_NETWORK=mainnet, a valid HTTPS API URL (or local HTTP), and a maximum at or below 249.99 EUR.

quote_xgr_purchase is optional planning only. create_xgr_purchase_order creates the real order in one call to POST /api/orders; the website confirmation modal is not a separate API phase. The backend is authoritative for the final price, crypto amount, custody wallet, payment reference, and reservation. The MCP holds no user or third-party private keys and does not pay; an external wallet agent can execute the returned exact payment instruction.

The 249.99 EUR maximum is an MCP policy for autonomous orders. Separately, the XGR_Web backend requires a complete billing address from a newly calculated net_eur >= 250; it is not a backend order maximum. A backend repricing/address error is returned without retrying the order.

Purchase tools distinguish fixed-XGR orders from conservative USDC/USDT budget orders. The live backend POST determines the binding market price and exact amount_crypto; EUR is only a policy/reference value. A budget order whose exact returned amount exceeds its cap is not paid and its existing reservation expires normally.

Starter-gas tools

Where enabled, get_xgr_starter_gas_options exposes the current fixed grant and eligibility policy. request_xgr_starter_gas sends one fixed 1 XGR grant from a dedicated service wallet to an eligible low-balance EVM address. Agents should use it before deployments, session starts, contract calls or other on-chain actions when the target address lacks native XGR for gas.

The service does not request an address-ownership proof and does not use proof-of-work. Abuse is bounded through one confirmed grant per address, an hourly cap, a daily cap, recipient-balance eligibility, a low-balance service wallet and a limited number of retries after pre-broadcast failures.

服务器配置

{
  "mcpServers": {
    "xgr-network": {
      "url": "https://mcp.xgr.network/mcp",
      "type": "streamable-http"
    }
  }
}
推荐的 MCP Server
TraeBuild with Free GPT-4.1 & Claude 3.7. Fully MCP-Ready.
RedisA Model Context Protocol server that provides access to Redis databases. This server enables LLMs to interact with Redis key-value stores through a set of standardized tools.
BlenderBlenderMCP connects Blender to Claude AI through the Model Context Protocol (MCP), allowing Claude to directly interact with and control Blender. This integration enables prompt assisted 3D modeling, scene creation, and manipulation.
Jina AI MCP ToolsA Model Context Protocol (MCP) server that integrates with Jina AI Search Foundation APIs.
Playwright McpPlaywright MCP server
DeepChatYour AI Partner on Desktop
Y GuiA web-based graphical interface for AI chat interactions with support for multiple AI models and MCP (Model Context Protocol) servers.
CursorThe AI Code Editor
WindsurfThe new purpose-built IDE to harness magic
MCP AdvisorMCP Advisor & Installation - Use the right MCP server for your needs
ChatWiseThe second fastest AI chatbot™
Baidu Map百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Tavily Mcp
MiniMax MCPOfficial MiniMax Model Context Protocol (MCP) server that enables interaction with powerful Text to Speech, image generation and video generation APIs.
EdgeOne Pages MCPAn MCP service designed for deploying HTML content to EdgeOne Pages and obtaining an accessible public URL.
Howtocook Mcp基于Anduin2017 / HowToCook (程序员在家做饭指南)的mcp server,帮你推荐菜谱、规划膳食,解决“今天吃什么“的世纪难题; Based on Anduin2017/HowToCook (Programmer's Guide to Cooking at Home), MCP Server helps you recommend recipes, plan meals, and solve the century old problem of "what to eat today"
Visual Studio Code - Open Source ("Code - OSS")Visual Studio Code
Serper MCP ServerA Serper MCP Server
AiimagemultistyleA Model Context Protocol (MCP) server for image generation and manipulation using fal.ai's Stable Diffusion model.
Amap Maps高德地图官方 MCP Server
Zhipu Web SearchZhipu Web Search MCP Server is a search engine specifically designed for large models. It integrates four search engines, allowing users to flexibly compare and switch between them. Building upon the web crawling and ranking capabilities of traditional search engines, it enhances intent recognition capabilities, returning results more suitable for large model processing (such as webpage titles, URLs, summaries, site names, site icons, etc.). This helps AI applications achieve "dynamic knowledge acquisition" and "precise scenario adaptation" capabilities.