everything-mcp
The definitive MCP server for voidtools Everything - lightning-fast file search for AI agents.
README
<div align="center"> <h1>⚡ Everything MCP</h1> <p> <strong>The definitive MCP server for <a href="https://www.voidtools.com/">voidtools Everything</a> - lightning-fast file search for AI agents.</strong> </p> <p> <a href="https://pypi.org/project/everything-mcp/"><img alt="PyPI" src="https://img.shields.io/pypi/v/everything-mcp.svg?cacheSeconds=300&v=20260204"></a> <a href="https://pypi.org/project/everything-mcp/"><img alt="Python" src="https://img.shields.io/pypi/pyversions/everything-mcp.svg?cacheSeconds=300&v=20260204"></a> <a href="LICENSE"><img alt="License" src="https://img.shields.io/github/license/elis132/everything-mcp.svg?cacheSeconds=300&v=20260204"></a> </p> <p>Search millions of files in milliseconds. Built for <strong>Claude Code</strong>, <strong>Codex</strong>, <strong>Gemini</strong>, <strong>Kimi</strong>, <strong>Qwen</strong>, <strong>Cursor</strong>, <strong>Windsurf</strong>, and any MCP-compatible client.</p> </div>
Why This One?
| Feature | everything-mcp (this) | mamertofabian (271⭐) | essovius (0⭐) |
|---|---|---|---|
| Tools | 5 well-designed | 1 generic | 16 granular |
| Auto-detection | ✅ Finds Everything + es.exe automatically | ❌ Manual DLL path | ❌ Manual setup |
| Everything 1.5 | ✅ Auto-detects instance | ❌ No support | ⚠️ Manual flag |
| Content preview | ✅ Read first N lines | ❌ | ❌ |
| File type categories | ✅ 10 categories | ❌ | ✅ |
| Stats & counts | ✅ Size stats, extension breakdown | ❌ | Partial |
| Error handling | ✅ All tools return clean errors | ❌ Raw exceptions | ❌ |
| Test suite | ✅ pytest | ❌ | ❌ |
| Zero config | ✅ Works out of the box | ❌ Need SDK DLL path | ❌ Need es.exe in PATH |
Performance
Real benchmark from this machine (Windows, query: everything.exe):
everything-mcp(Everything index viaes.exe): 220.22 ms avg (5 runs)- Naive filesystem walk over
C:\: 66,539.03 ms (single run) - Observed speedup: ~302x faster
Reproduce locally (PowerShell):
@'
import os
import subprocess
import time
import statistics
ES = os.path.expandvars(r"%LOCALAPPDATA%\Everything\es.exe")
QUERY = "everything.exe"
es_runs = []
for _ in range(5):
t0 = time.perf_counter()
subprocess.run([ES, "-n", "100", QUERY], capture_output=True, text=True)
es_runs.append((time.perf_counter() - t0) * 1000)
t0 = time.perf_counter()
matches = []
for dirpath, _, filenames in os.walk(r"C:\\"):
for name in filenames:
if name.lower() == QUERY:
matches.append(os.path.join(dirpath, name))
walk_ms = (time.perf_counter() - t0) * 1000
es_avg = statistics.mean(es_runs)
print("ES avg ms:", round(es_avg, 2))
print("Walk ms:", round(walk_ms, 2))
print("Speedup x:", round(walk_ms / es_avg, 1))
print("Matches:", len(matches))
'@ | python -
Installation
Prerequisites
- Windows with Everything installed and running
- es.exe (Everything command-line interface):
- Everything 1.5 alpha: es.exe is included
- Everything 1.4: Download from github.com/voidtools/es
- Place
es.exein your PATH or in the Everything installation directory
- Python 3.10+ or uv
Via uv (recommended - no install needed)
uvx everything-mcp
Via pip
pip install everything-mcp
From source
git clone https://github.com/elis132/everything-mcp.git
cd everything-mcp
pip install -e ".[dev]"
Configuration
<details> <summary>Shared MCP JSON Template</summary>
Use this server definition anywhere a client asks for MCP JSON:
{
"mcpServers": {
"everything": {
"command": "uvx",
"args": ["everything-mcp"]
}
}
}
</details>
<details> <summary>Claude Code</summary>
Use the Claude Code CLI:
claude mcp add everything -- uvx everything-mcp
claude mcp list
Or add to .claude/settings.json:
{
"mcpServers": {
"everything": {
"command": "uvx",
"args": ["everything-mcp"]
}
}
}
</details>
<details> <summary>Claude Desktop</summary>
Add to %APPDATA%\Claude\claude_desktop_config.json:
{
"mcpServers": {
"everything": {
"command": "uvx",
"args": ["everything-mcp"]
}
}
}
</details>
<details> <summary>Codex CLI</summary>
Use the Codex CLI:
codex mcp add everything -- uvx everything-mcp
codex mcp list
</details>
<details> <summary>Gemini CLI</summary>
Use the Gemini CLI:
gemini mcp add -s user everything uvx everything-mcp
gemini mcp list
</details>
<details> <summary>Kimi CLI</summary>
Use the Kimi CLI:
kimi mcp add --transport stdio everything -- uvx everything-mcp
kimi mcp list
</details>
<details> <summary>Qwen CLI</summary>
Use the Qwen CLI:
qwen mcp add -s user everything uvx everything-mcp
qwen mcp list
</details>
<details> <summary>Cursor</summary>
Cursor currently uses MCP settings/deeplinks rather than a stable mcp add
CLI command. Add the JSON config in Cursor's MCP settings UI.
</details>
<details> <summary>Windsurf</summary>
Windsurf currently uses MCP settings rather than a stable mcp add CLI
command. On Windows, add the JSON config to:
%USERPROFILE%\.codeium\windsurf\mcp_config.json
</details>
<details> <summary>Generic MCP Clients</summary>
Any MCP-compatible client can use this format:
{
"mcpServers": {
"everything": {
"command": "uvx",
"args": ["everything-mcp"]
}
}
}
</details>
<details> <summary>Using pip Instead of uvx</summary>
{
"mcpServers": {
"everything": {
"command": "everything-mcp"
}
}
}
Or with explicit Python:
{
"mcpServers": {
"everything": {
"command": "python",
"args": ["-m", "everything_mcp"]
}
}
}
</details>
Environment Variables (Optional)
Everything MCP auto-detects your setup, but you can override:
| Variable | Description | Example |
|---|---|---|
EVERYTHING_ES_PATH |
Path to es.exe | C:\Program Files\Everything\es.exe |
EVERYTHING_INSTANCE |
Everything instance name | 1.5a |
{
"mcpServers": {
"everything": {
"command": "uvx",
"args": ["everything-mcp"],
"env": {
"EVERYTHING_INSTANCE": "1.5a"
}
}
}
}
Tools
1. everything_search - The Workhorse
Search files and folders using Everything's full query syntax.
| Parameter | Default | Description |
|---|---|---|
query |
(required) | Everything search query |
max_results |
50 | 1–500 |
sort |
date-modified-desc |
See sort options below |
match_case |
false | Case-sensitive |
match_whole_word |
false | Whole words only |
match_regex |
false | Regex mode |
match_path |
false | Match full path |
offset |
0 | Pagination offset |
Everything Search Syntax:
*.py → All Python files
ext:py;js;ts → Multiple extensions
ext:py path:C:\Projects → Python files in Projects
size:>10mb → Larger than 10 MB
size:1kb..1mb → Between 1 KB and 1 MB
dm:today → Modified today
dm:last1week → Modified in the last week
dc:2024 → Created in 2024
"exact name.txt" → Exact filename match
project1 | project2 → OR search
!node_modules → Exclude term
ext:py !test !__pycache__ → Python, excluding tests
content:TODO → Files containing TODO (requires content indexing)
regex:^test_.*\.py$ → Regex search
parent:src ext:py → Python files in 'src' folders
dupe: → Duplicate filenames
empty: → Empty folders
2. everything_search_by_type - Category Search
Search by pre-defined file type categories.
Categories: audio, video, image, document, code, archive, executable, font, 3d, data
| Parameter | Default | Description |
|---|---|---|
file_type |
(required) | Category name |
query |
"" |
Additional filter |
path |
"" |
Directory restriction |
max_results |
50 | 1–500 |
sort |
date-modified-desc |
Sort order |
3. everything_find_recent - What Changed?
Find recently modified files. Sorted newest-first.
Periods: 1min, 5min, 10min, 15min, 30min, 1hour, 2hours, 6hours, 12hours, today, yesterday, 1day, 3days, 1week, 2weeks, 1month, 3months, 6months, 1year
| Parameter | Default | Description |
|---|---|---|
period |
1hour |
Time period |
path |
"" |
Directory restriction |
extensions |
"" |
Extension filter (e.g. py,js,ts) |
query |
"" |
Additional filter |
max_results |
50 | 1–500 |
4. everything_file_details - Deep Inspection
Get metadata and optional content preview for specific files.
| Parameter | Default | Description |
|---|---|---|
paths |
(required) | File paths to inspect (1–20) |
preview_lines |
0 | Lines of text to preview (0–200) |
Returns: Full metadata (size, dates, permissions, hidden status). Directories: item count and listing. Text files with preview: first N lines of content.
5. everything_count_stats - Quick Analytics
Get count and size statistics without listing individual files.
| Parameter | Default | Description |
|---|---|---|
query |
(required) | Search query |
include_size |
true | Calculate total size |
breakdown_by_extension |
false | Sampled per-extension stats (files only) |
Examples
"Find all Python files modified today in my project"
→ everything_find_recent(period="today", extensions="py", path="C:\Projects\myapp")
"How much disk space do my log files use?"
→ everything_count_stats(query="ext:log", include_size=true, breakdown_by_extension=true)
"Show me the first 50 lines of that config file"
→ everything_file_details(paths=["C:\Projects\app\config.yaml"], preview_lines=50)
"Find all duplicate filenames in Documents"
→ everything_search(query='dupe: path:"C:\Users\me\Documents"')
"Find all images larger than 5MB"
→ everything_search(query="ext:jpg;png;gif size:>5mb")
Troubleshooting
"es.exe not found"
- Ensure Everything is installed: https://www.voidtools.com/
- Download es.exe: https://github.com/voidtools/es/releases
- Place es.exe in your PATH or set
EVERYTHING_ES_PATH
"Everything IPC window not found"
- Ensure Everything is running (check system tray)
- If using Everything 1.5 alpha, set
EVERYTHING_INSTANCE=1.5a - Ensure you're not running Everything Lite (no IPC support)
"No results for valid queries"
- Verify Everything's index is built (needs time on first run)
- Try the same query in Everything's GUI
- Check that the drive/path is included in Everything's index settings
Debugging
# View server logs
everything-mcp 2>everything-mcp.log
# MCP Inspector
npx @modelcontextprotocol/inspector uvx everything-mcp
Architecture
┌──────────────┐ MCP (stdio) ┌──────────────────┐
│ AI Agent │◄────────────────────►│ Everything MCP │
│ (Claude, │ │ Server │
│ Codex, etc) │ │ │
└──────────────┘ │ 5 Tools: │
│ • search │
│ • search_by_type│
│ • find_recent │
│ • file_details │
│ • count_stats │
└────────┬─────────┘
│ async subprocess
▼
┌──────────────────┐
│ es.exe │
│ (CLI interface) │
└────────┬─────────┘
│ IPC / Named Pipes
▼
┌──────────────────┐
│ Everything │
│ Service │
│ (voidtools) │
│ │
│ Real-time NTFS │
│ file index │
└──────────────────┘
Development
# Install with dev dependencies
pip install -e ".[dev]"
# Run tests
pytest
# Lint
ruff check src/ tests/
Contributing
Contributions welcome! Areas for improvement:
- Direct named pipe IPC (bypass es.exe for lower latency)
- Everything SDK3 integration for Everything 1.5
- Content search integration
- File watching / change notifications
- Bookmark and tag support (Everything 1.5)
License
MIT - see LICENSE
Acknowledgments
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.