Multi Web Search MCP
This server does not currently have a deployable release on Glama. Follow the repository's setup instructions to install and configure it for your MCP client, including any required credentials and local dependencies.
Multi Web Search MCP
Multi Web Search MCP 是 Codex 的联网搜索 MCP 服务。 项目面向多供应商搜索接口。 项目提供统一搜索入口。 项目返回结构化搜索结果。 项目支持自动故障转移。 项目支持短期熔断。 项目支持请求缓存。 项目支持并发限制。 项目提供 fast、balanced、deep 三种搜索模式。
项目特点
多供应商适配。
自动故障转移。
结构化结果。
短期熔断。
请求缓存。
并发限制。
三种搜索模式。
密钥不进入工具结果。
Related MCP server: web-search-mcp
搜索接口
项目适配以下接口:
DeepSeek Search API
DeepSeek Anthropic Messages
MiMo Chat Completions
GLM Chat Completions
DashScope
OpenAI-compatible Responses
Gemini Google Search
自定义 Chat Completions 接口
工具
web_search:单次搜索。web_search_batch:批量搜索。单次最多 8 个查询。search_providers:查看供应商状态。reload_search_config:重载配置。clear_search_cache:清理缓存。
快速开始
运行环境为 Python 3.11 以上版本。
依赖列表位于 requirements.txt。
依赖安装命令:
python3 -m pip install -r requirements.txt启动脚本为 launch.sh。
启动脚本读取同一目录的 server.py。
启动命令:
./launch.sh配置
配置文件为 providers.json。
配置文件包含供应商名称。
配置文件包含接口类型。
配置文件包含启用状态。
配置文件包含优先级。
配置文件包含 API Key 来源。
配置文件内置多个供应商模板。
供应商通过 enabled 控制启用状态。
供应商通过 priority 控制优先级。
API Key 来源字段为 api_key_env。
API Key 来源字段为 api_key_file。
API Key 来源字段为 api_key_toml。
配置修改后自动加载。
个人配置建议保存为 providers.local.json。
MULTI_WEB_SEARCH_CONFIG 可指定配置文件路径。
providers.local.json 已加入 .gitignore。
Codex MCP 配置
配置示例:
[mcp_servers.multi_web_search]
command = "/path/to/multi-web-search/launch.sh"
startup_timeout_sec = 10.0
tool_timeout_sec = 45.0
[mcp_servers.multi_web_search.tools.web_search]
output_token_limit = 12000
[mcp_servers.multi_web_search.tools.web_search_batch]
output_token_limit = 20000安全
API Key 从环境变量读取。
API Key 从密钥文件读取。
API Key 从 TOML 配置读取。
工具结果不返回 API Key。
仓库不包含 API Key。
仓库不包含个人配置。
仓库不包含用户数据。
仓库不包含搜索记录。
安全策略位于 SECURITY.md。
测试
检查命令:
make check测试命令:
make test支持
贡献指南位于 CONTRIBUTING.md。
项目的 Star 支持后续开发。
English
Multi Web Search MCP is a web search MCP server for Codex. It provides one search entry for multiple providers. It supports automatic failover. It returns structured search results. It supports request caching. It supports concurrency limits. It provides three search modes: fast, balanced, deep. API keys come from environment variables. API keys come from key files. API keys come from TOML fields. MCP tool results do not include API keys. The repository does not include API keys. The repository does not include personal configuration. The repository does not include user data. The repository does not include search history.
License
许可证为 MIT。
许可证文本位于 LICENSE。
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
This server cannot be deployed
Maintenance
Related MCP Connectors
Web search for AI agents — one tool across 6 engines, routed to the cheapest + cached.
Provides AI assistants with access to Seltz's powerful Web Search capabilities.
Web search and page fetch for AI agents. Results come from a shared cache; cached calls cost less.
LLM-ready web search + instant answers + URL-to-clean-text fetch for agents and RAG.
Related MCP Servers
- FlicenseNot gradedqualityDmaintenanceEnables AI agents to search the web and extract content using multiple search providers, with caching, retry logic, and options for JavaScript-heavy page rendering.-
- AlicenseNot gradedqualityDmaintenanceProvides multi-provider web search capabilities with fallback chains, semantic reranking, and content extraction for grounded agent retrieval.1MIT
- AlicenseAqualityBmaintenanceProvides a unified web search tool for AI agents by running multiple search providers concurrently and returning normalized results. It integrates with Claude Code, Codex CLI, OpenCode, and Hermes.1379 PyPI17MIT
- AlicenseNot gradedqualityBmaintenanceEnables AI clients to perform unified web searches across multiple search providers with automatic key rotation, failover, and circuit breaking, and to inspect provider status and capabilities.241 npm3MIT