feat: add mcp_server_hqd for multi-source enterprise data search

server/mcp_server_hqd/.env_example

@@ -0,0 +1,8 @@
+# HQD MCP Remote endpoint
+HQD_MCP_ENDPOINT=https://sd6k08f59gqcea6qe13vg.apigateway-cn-beijing.volceapi.com/mcp
+
+# Auth token for remote HQD MCP endpoint
+HQD_AUTH_TOKEN=
+
+# Server settings
+PORT=8000

server/mcp_server_hqd/.gitignore

@@ -0,0 +1,8 @@
+__pycache__/
+*.py[oc]
+build/
+dist/
+wheels/
+*.egg-info
+.venv
+.env

server/mcp_server_hqd/.python-version

@@ -0,0 +1 @@
+3.10

server/mcp_server_hqd/README.md

@@ -0,0 +1,116 @@
+# HQD Multi-Source Search MCP Server
+
+## Version Information
+
+v0.1
+
+## Product Description
+
+### Short Description
+
+Query enterprise data from HQD (High-Quality Dataset) multi-source search service, covering basic info, risk, operations, IP, and litigation data.
+
+### Long Description
+
+HQD Multi-Source Search MCP Server is a thin proxy that connects to the remote HQD MCP service deployed on Volcengine. It provides unified access to 5 enterprise data sources through a two-phase interaction pattern: metadata discovery (`describe_datasource`) followed by data retrieval (`query_datasource`). All queries are forwarded to the remote endpoint — no local data processing is performed.
+
+## Category
+
+Data Intelligence
+
+## Tags
+
+Enterprise Data, Multi-Source Search, HQD
+
+## Tools
+
+This MCP Server product provides the following Tools (capabilities):
+
+### Tool 1: describe_datasource
+
+Get metadata for data sources including dimensions, metrics, and filters. Agents should call this first to understand available data structures before querying.
+
+### Tool 2: query_datasource
+
+Query data from a specific datasource with filtering, aggregation, and pagination. Supports filter operators: `eq`, `like`, `in`, `not_in`, `between`, `range`, `keyword`.
+
+## Compatible Platforms
+
+- Python
+
+## Authentication Method
+
+Bearer Token
+
+### Obtaining Auth Token
+
+Contact the HQD service administrator to obtain an authentication token.
+
+### Environment Variable Configuration
+
+| Variable Name | Value |
+| ---------- | ---------- |
+| `HQD_AUTH_TOKEN` | Auth token for the remote HQD MCP endpoint |
+| `HQD_MCP_ENDPOINT` | Remote HQD MCP endpoint (optional, has default) |
+
+## Python MCP Server
+
+### Dependencies
+
+The device running MCP server needs to install the following dependencies:
+
+- [Python](https://www.python.org/downloads/) 3.10 or higher.
+- [`uv`](https://docs.astral.sh/uv/) & [`uvx`](https://docs.astral.sh/uv/guides/tools/).
+
+### Deployment and Configuration
+
+```json
+{
+  "mcpServers": {
+    "mcp-server-hqd": {
+      "command": "uvx",
+      "args": [
+        "--from",
+        "git+https://github.com/volcengine/mcp-server#subdirectory=server/mcp_server_hqd",
+        "mcp-server-hqd"
+      ],
+      "env": {
+        "HQD_AUTH_TOKEN": "Your HQD Auth Token"
+      }
+    }
+  }
+}
+```
+
+> Note: Please replace `Your HQD Auth Token` above with the authentication token provided by the HQD service administrator.
+
+## Using Clients
+
+The following clients are supported for interacting with MCP Server. For specific configurations, please refer to the client documentation:
+
+- Cursor
+- [Trae](https://www.trae.com.cn/)
+- Claude Desktop
+- Ark
+
+Supports [Cline](https://cline.bot/) plugin
+
+## Available Datasources
+
+| ID | Name |
+|----|------|
+| `enterprise_basic_wide` | Enterprise Basic Information |
+| `enterprise_risk_wide` | Enterprise Risk Information |
+| `enterprise_operation_wide` | Enterprise Operations Information |
+| `enterprise_ip_wide` | Enterprise Intellectual Property |
+| `enterprise_litigation` | Enterprise Litigation Information |
+
+## Conversation Initiation Example
+
+- List all available data sources.
+- Query the basic information of the enterprise named "ByteDance".
+- Search for enterprises with registered capital over 10 million in Beijing.
+
+## License
+
+[MIT](../../LICENSE)

server/mcp_server_hqd/README_zh.md

@@ -0,0 +1,116 @@
+# 高质量数据集多源搜索 MCP Server
+
+## 版本信息
+
+v0.1
+
+## 产品描述
+
+### 短描述
+
+查询 HQD（高质量数据集）多源搜索服务中的企业数据，涵盖基本信息、风险、经营、知识产权和诉讼数据。
+
+### 长描述
+
+HQD 多源搜索 MCP Server 是一个轻量代理，连接部署在火山引擎上的远端 HQD MCP 服务。通过两阶段交互模式提供对 5 个企业数据源的统一访问：先通过 `describe_datasource` 发现元数据，再通过 `query_datasource` 检索数据。所有查询均转发到远端服务，本地不进行数据处理。
+
+## 分类
+
+数据智能
+
+## 标签
+
+企业数据、多源搜索、HQD
+
+## Tools
+
+本 MCP Server 产品提供以下 Tools (工具/能力):
+
+### Tool 1: describe_datasource
+
+获取数据源的可查询元数据信息，包括维度（dimensions）、指标（metrics）和过滤条件（filters）。Agent 应先调用此工具了解数据结构，再进行查询。
+
+### Tool 2: query_datasource
+
+查询指定数据源的实际数据，支持过滤、聚合和分页。支持的过滤操作符：`eq`（精确）、`like`（模糊）、`in`（批量）、`not_in`（排除）、`between`（范围）、`range`（数值范围）、`keyword`（全文搜索）。
+
+## 可适配平台
+
+- Python
+
+## 鉴权方式
+
+Bearer Token
+
+### 获取 Auth Token
+
+请联系 HQD 服务管理员获取认证令牌。
+
+### 环境变量配置
+
+| 变量名 | 值 |
+| ---------- | ---------- |
+| `HQD_AUTH_TOKEN` | 远端 HQD MCP 服务的认证令牌 |
+| `HQD_MCP_ENDPOINT` | 远端 HQD MCP 端点地址（可选，有默认值） |
+
+## Python 版 MCP server
+
+### 依赖项
+
+运行 MCP server 的设备需要安装以下依赖项。
+
+- [Python](https://www.python.org/downloads/) 3.10 或更高版本。
+- [`uv`](https://docs.astral.sh/uv/) & [`uvx`](https://docs.astral.sh/uv/guides/tools/)。
+
+### 部署与配置
+
+```json
+{
+  "mcpServers": {
+    "mcp-server-hqd": {
+      "command": "uvx",
+      "args": [
+        "--from",
+        "git+https://github.com/volcengine/mcp-server#subdirectory=server/mcp_server_hqd",
+        "mcp-server-hqd"
+      ],
+      "env": {
+        "HQD_AUTH_TOKEN": "Your HQD Auth Token"
+      }
+    }
+  }
+}
+```
+
+> 注：请将上方 `Your HQD Auth Token` 替换为 HQD 服务管理员提供的认证令牌。
+
+## 使用客户端
+
+支持通过以下客户端与 MCP Server 交互，具体配置可查阅该客户端文档。
+
+- Cursor
+- [Trae](https://www.trae.com.cn/)
+- Claude Desktop
+- 方舟
+
+支持 [Cline](https://cline.bot/) 插件。
+
+## 可用数据源
+
+| ID | 名称 |
+|----|------|
+| `enterprise_basic_wide` | 企业基本信息宽表 |
+| `enterprise_risk_wide` | 企业风险信息宽表 |
+| `enterprise_operation_wide` | 企业经营信息宽表 |
+| `enterprise_ip_wide` | 企业知识产权宽表 |
+| `enterprise_litigation` | 企业诉讼信息 |
+
+## 对话发起示例
+
+- 列出所有可用的数据源。
+- 查询名为"字节跳动"的企业基本信息。
+- 搜索北京市注册资本超过1000万的企业。
+
+## 许可
+
+[MIT](../../LICENSE)

server/mcp_server_hqd/pyproject.toml

@@ -0,0 +1,19 @@
+[project]
+name = "mcp-server-hqd"
+version = "0.1.0"
+description = "MCP server for HQD Multi-Source Enterprise Data Search"
+readme = "README.md"
+requires-python = ">=3.10"
+license = {text = "MIT"}
+dependencies = [
+    "mcp[cli]>=1.5.0",
+    "requests>=2.31.0",
+    "python-dotenv>=1.0.1",
+]
+
+[project.scripts]
+mcp-server-hqd = "mcp_server_hqd.server:main"
+
+[build-system]
+requires = ["hatchling"]
+build-backend = "hatchling.build"

server/mcp_server_hqd/src/mcp_server_hqd/config.py

@@ -0,0 +1,45 @@
+"""Configuration for HQD MCP Server (remote proxy mode)."""
+
+import os
+import logging
+from dataclasses import dataclass
+
+from dotenv import load_dotenv
+
+logger = logging.getLogger(__name__)
+
+load_dotenv()
+
+
+@dataclass
+class HqdConfig:
+    """Configuration for HQD MCP proxy server.
+
+    Environment variables:
+        HQD_MCP_ENDPOINT: Remote HQD MCP server endpoint (streamable-http)
+        HQD_AUTH_TOKEN: Auth token for remote endpoint (sent as Authorization header)
+        PORT: Local server port (default: 8000)
+    """
+
+    endpoint: str
+    auth_token: str
+    port: int
+
+
+def load_config() -> HqdConfig:
+    """Load configuration from environment variables."""
+    endpoint = os.getenv(
+        "HQD_MCP_ENDPOINT",
+        "https://sd6k08f59gqcea6qe13vg.apigateway-cn-beijing.volceapi.com/mcp",
+    )
+    if not endpoint:
+        raise ValueError("Missing required environment variable: HQD_MCP_ENDPOINT")
+
+    return HqdConfig(
+        endpoint=endpoint,
+        auth_token=os.getenv("HQD_AUTH_TOKEN", ""),
+        port=int(os.getenv("PORT", "8000")),
+    )
+
+
+config = load_config()