everything-mcp
Enables LLM agents to perform instant whole-disk filename searches using Everything's index, supporting native query syntax and pagination.
README
everything-mcp
MCP server that exposes Everything (voidtools) instant file search to LLM agents such as Claude Code. It searches file and folder names across the entire disk in milliseconds using Everything's index — a complement to Glob/Grep, which only see the current project.
Features
- Whole-disk filename search in milliseconds via Everything's live index.
- Native Everything query syntax (
ext:,path:,dm:today,size:>10mb,folder:,file:,|,!) passed straight through — no need for extra tools. - Pagination with
offsetand atotal_resultscount, so an agent knows when to page. everything_statusdiagnostic tool to check the index is ready before searching.- Windows via the Everything SDK dll over local IPC — no open ports, no per-search subprocess.
Requirements
- Windows with Everything running (living in the tray is enough).
Everything64.dllfrom the Everything SDK.- Python ≥3.11 and uv.
Setup
-
Clone the repo
git clone https://github.com/Santisoutoo/everything-mcp.git cd everything-mcp -
Provide the DLL. Download the Everything SDK and copy
dll/Everything64.dllintolib/(not versioned in the repo), or point theEVERYTHING_DLLenvironment variable at the dll wherever it lives. -
Register it in Claude Code (user scope, available in every session):
claude mcp add --scope user everything -- uv run --directory C:\path\to\everything-mcp everything-mcpReplace
C:\path\to\everything-mcpwith the absolute path where you cloned the repo.
Tools
search_files
search_files(query, max_results=50, offset=0, match_case=False,
match_whole_word=False, match_path=False, regex=False, sort="default")
| Parameter | Meaning |
|---|---|
query |
Everything query. Space = AND; wildcards * ?; filters like ext:pdf;docx, path:"C:\Users", parent:"D:\media", folder:, file:, dm:today, size:>10mb; operators ` |
max_results |
Cap on returned rows (1–1000). |
offset |
Skip N results (pagination). |
match_case |
Case-sensitive matching. |
match_whole_word |
Require each term to match a whole word. |
match_path |
Match terms against the full path, not just the name. |
regex |
Treat query as a regular expression. |
sort |
One of default (index order), name, path, size (largest first), date_modified (newest first). |
Returns {total_results, results: [{path, is_folder, size, modified}]}. total_results is the full
match count; page with offset when it exceeds max_results. Searches names only, not file contents.
everything_status
Returns {available, db_loaded, version, indexed_items}, or {available: false, error} when
Everything is not running. Useful for an agent to confirm the index is ready before searching.
Development
uv sync
uv run pytest # smoke tests (self-skip when Everything is not running)
uv run ruff check . # lint
uv run mypy src # type-check
uv run everything-mcp # start the server over stdio
License
Released under the MIT License.
推荐服务器
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。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。
VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
Exa MCP Server
模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。