百度AI搜索 MCP服务
基于百度千帆AI搜索API的MCP服务,可在Cursor等支持MCP的应用中使用百度AI搜索功能。
功能特性
- 智能问答: 基于百度搜索的AI智能回答,支持实时信息获取
- 深度搜索: 可选开启深度搜索获取更全面的信息
- 新闻获取: 快速获取各领域最新新闻
- 时间过滤: 支持按时间范围筛选搜索结果
- 参考来源: 返回回答的参考来源链接
安装
方式一:使用 uvx(推荐)
无需安装,直接在 Cursor MCP 配置中使用:
{
"mcpServers": {
"baidu-ai-search": {
"command": "uvx",
"args": ["baidu-ai-search-mcp"],
"env": {
"BAIDU_API_KEY": "your-api-key-here"
}
}
}
}
方式二:使用 pip 安装
pip install baidu-ai-search-mcp
然后在 Cursor MCP 配置中:
{
"mcpServers": {
"baidu-ai-search": {
"command": "python",
"args": ["-m", "baidu_ai_search_mcp"],
"env": {
"BAIDU_API_KEY": "your-api-key-here"
}
}
}
}
配置
获取 API Key
- 访问 百度千帆控制台
- 创建应用并获取 API Key(格式:
bce-v3/ALTAK***/xxx)
Cursor 配置文件位置
- Windows:
%USERPROFILE%\.cursor\mcp.json - macOS/Linux:
~/.cursor/mcp.json
完整配置示例
{
"mcpServers": {
"baidu-ai-search": {
"command": "uvx",
"args": ["baidu-ai-search-mcp"],
"env": {
"BAIDU_API_KEY": "bce-v3/ALTAK-xxxxx/xxxxxxxxxx",
"BAIDU_MODEL": "ernie-3.5-8k"
}
}
}
}
配置完成后,重启 Cursor 即可使用。
可用工具
baidu_ai_ask
向百度AI搜索提问,获取基于实时搜索的智能回答。
参数:
question(必需): 要询问的问题enable_deep_search(可选): 是否开启深度搜索,默认 falsetime_filter(可选): 时间过滤 (week/month/semiyear/year)
示例:
使用 baidu_ai_ask 工具查询"Python 3.12有什么新特性?"
baidu_ai_search
智能搜索,返回搜索结果摘要和链接。
参数:
query(必需): 搜索查询词max_results(可选): 最大结果数量,默认 5
baidu_ai_news
获取最新新闻资讯。
参数:
topic(可选): 新闻主题(如"科技"、"财经"、"体育")time_range(可选): 时间范围 (week/month),默认 week
支持的模型
| 模型 | 特点 |
|---|---|
| ernie-3.5-8k | 默认模型,速度快,性价比高 |
| ernie-4.0-turbo-8k | 更智能,适合复杂问题 |
| ernie-4.0-turbo-128k | 支持长文本上下文 |
| deepseek-r1 | 深度推理,适合复杂分析 |
| deepseek-v3 | DeepSeek最新模型 |
通过 BAIDU_MODEL 环境变量设置使用的模型。
API 配额
- 每天 100次 免费调用
- 超出后按量计费
- 详见 计费说明
开发
本地开发安装
git clone https://github.com/yourusername/baidu-ai-search-mcp.git
cd baidu-ai-search-mcp
pip install -e .
运行测试
# 设置环境变量
export BAIDU_API_KEY="your-api-key"
# 直接运行
python -m baidu_ai_search_mcp
常见问题
Q: 提示 "未配置 BAIDU_API_KEY"
A: 请确保在 mcp.json 的 env 中正确配置了 API Key。
Q: 请求超时
A: 深度搜索模式耗时较长,可尝试关闭 enable_deep_search。
Q: 返回错误 "Authentication error"
A: API Key 格式不正确或已过期,请检查配置。
License
MIT
Metadata
Release files for baidu-ai-search-mcp 0.1.0
For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.
Source distribution (sdist)
| File | Size | Uploaded | Provenance | Actions |
|---|---|---|---|---|
| baidu_ai_search_mcp-0.1.0.tar.gz | 8.3 kB | Without provenance | Details for baidu_ai_search_mcp-0.1.0.tar.gz |
Built distribution (wheel)
| File | Interpreter | ABI | Platform | Actions Reset |
|---|---|---|---|---|
| baidu_ai_search_mcp-0.1.0-py3-none-any.whl | Python 3 | none | any | Details for baidu_ai_search_mcp-0.1.0-py3-none-any.whl |
Total release size: 18.4 kB
Release files / baidu_ai_search_mcp-0.1.0.tar.gz
| Download URL | baidu_ai_search_mcp-0.1.0.tar.gz |
|---|---|
| Size | 8.3 kB |
| Tags | Source |
|
SHA-256 checksum How to use checksums |
a6b400b0d57fccff3f7a0d79a26b5473491b51f723d6289088ad19b6ea23efe1
|
|
BLAKE2b-256 checksum How to use checksums |
90470da985d16269fc342d3d091673434e6b38b4df905c6711ea4e9fcd067b7f
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
No |
| Uploaded via |
twine/6.2.0 CPython/3.13.9
|
Release files / baidu_ai_search_mcp-0.1.0-py3-none-any.whl
| Download URL | baidu_ai_search_mcp-0.1.0-py3-none-any.whl |
|---|---|
| Size | 10.2 kB |
| Tags | Python 3 |
|
SHA-256 checksum How to use checksums |
ea380c17ad0ace4040023e3c56d55aea0961f7e5826686b3c03f3027e3e79d35
|
|
BLAKE2b-256 checksum How to use checksums |
f385e2ded502199ceab00b6e5f7f349ba954383a417ee5b22c6afb56466a3c82
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
No |
| Uploaded via |
twine/6.2.0 CPython/3.13.9
|
Security for baidu-ai-search-mcp 0.1.0
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
No provenance found
None of the files in this release were uploaded with provenance metadata.
A note on package security
PyPI is a repository for distributing packages, not a service for auditing them. While third-party security researchers and our staff do their best to detect and remove malicious packages, PyPI cannot guarantee the quality, security or suitability of any given package.
To learn more about protecting yourself from malicious actors, visit our package security documentation.