rtorrent-mcp
MCP server for rTorrent BitTorrent automation, enabling torrent management (add/list/pause/delete), multi-source searching (Nyaa, Pirate Bay, etc.), and post-processing workflows via natural language.
README
rTorrent MCP Server
<p align="center"> <a href="https://github.com/casey/just"><img src="https://img.shields.io/badge/just-ready_to_go-7c5cfc?style=flat-square&logo=just&logoColor=white" alt="Just"></a> <a href="https://github.com/astral-sh/ruff"><img src="https://img.shields.io/endpoint?url=https://raw.githubusercontent.com/astral-sh/ruff/main/assets/badge/v2.json" alt="Ruff"></a> <a href="https://python.org"><img src="https://img.shields.io/badge/Python-3.13+-3776AB?style=flat-square&logo=python&logoColor=white" alt="Python"></a> <a href="https://github.com/PrefectHQ/fastmcp"><img src="https://img.shields.io/badge/FastMCP-3.2-7c5cfc?style=flat-square" alt="FastMCP"></a> </p>
📖 Installation Guide — quick start, manual setup, and troubleshooting
rTorrent MCP FastMCP 3.1.0 server for anime BitTorrent automation with Austrian legal context, talking to rTorrent over XML-RPC/SCGI (not a generic site scraper).
What this is: A BitTorrent control plane: add/list/pause torrents, search indexers (Nyaa, etc.), workflows, and post-processing against your rTorrent instance. It is not a generic systems MCP, and it is not a qBittorrent Web API client.
Web UI (web_sota/): A small Vite + React dashboard + REST bridge (/api/*) on the same uvicorn process as MCP (status, torrent list, magnet add). It is a deliberately minimal alternative to the ruTorrent WebUI bundled with Dockersee Quick Start (subsection ruTorrent vs this projects webapp) and web_sota/README.md. Agents still use MCP tools for full workflows.
** Naming:** GitHub repo
rtorrent-mcp; Python packagertorrent_mcp. The old qBittorrent prototype used the historic nameqbtmcpthat client is not supported; control is rTorrent via XML-RPC. See docs/RTORRENT_REFERENCE.md and docs/RTORRENT_SETUP.md.
Logs & tool text: Prefer ASCII markers ([OK], [FAIL], [WARN]) instead of Unicode emoji so MCP clients, Windows consoles, and JSON stay predictable.
Features
- rTorrent docs: docs/RTORRENT_REFERENCE.md (architecture + env) and docs/RTORRENT_SETUP.md (Docker, plugins, long setup)
- FastMCP 3.1.0: Portmanteau tools, MCPB packaging, prompts, skills provider, sampling, agentic workflow tool, CI/CD
- 6 Consolidated Tools: torrent, search, nlp, legal, system, workflow management
- rTorrent integration: Torrent operations via XML-RPC to your instance (add/list/pause/resume/delete, etc.)
- Multi-Source Search: nyaa.si (anime), Pirate Bay (TV), YTS (movies), Anna's Archive (ebooks)
- Post-Processing: Automatic completion detection, filename normalization, Plex integration
- Metadata Services: IMDb and TVDB metadata retrieval for movies and TV shows
- Austrian Legal Compliance: Built-in legal risk assessment for Austrian users
- Natural language: Parse anime-related commands in English/German
- Quality scoring: Simple heuristics for release ordering (e.g. group/metadata cues)
- Tool schemas: Docstrings and structured parameters for MCP clients
- System tools: Health/status helpers and optional workspace/repo inspection where implemented
- Configuration: Environment variables and
.envsupport - Tests: Unit and integration tests under
tests/
Quick Start
git clone https://github.com/sandraschi/rtorrent-mcp
cd rtorrent-mcp
just
This opens an interactive dashboard showing all available commands. Run just bootstrap to install dependencies, then just serve or just dev to start.
Manual Setup
If you don't have just installed:
Prerequisites
- Python 3.10 or higher
- Docker Desktop (for rTorrent) - Recommended
- Claude Desktop (for MCP integration)
rTorrent stack (Docker, recommended)
What crazy-max/docker-rtorrent-rutorrent is
CrazyMax maintains a well-used Docker setup that packages rTorrent (the actual client), ruTorrent (a PHP web UI on top of rTorrent), and nginx as a front door. Nginx exposes XML-RPC on a TCP port so clients (this MCP server, scripts, other tools) can call rTorrents RPC at /RPC2 without you wiring SCGI sockets by hand. The image is aimed at install Docker, get a working rTorrent + classic WebUI, not at building rTorrent from source.
This repos root docker-compose.yml pins crazymax/rtorrent-rutorrent:latest, maps XML-RPC to 12224 and ruTorrent to 12222, and uses volumes under ./config, your downloads folder, ./watch, and ./logs (see the compose file for exact bind paths on Windows).
Install (minimal)
- Install Docker Desktop and ensure it is running.
- Clone this repository (or copy
docker-compose.ymland related layout). - From the repository root:
docker compose up -d
(Use
docker-compose up -dif your Docker install only provides the hyphenated CLI.) - Check the container: docker logs rtorrent-mcp
- Endpoints (defaults in this repo):
- XML-RPC (for MCP):
http://localhost:12224/RPC2 - ruTorrent WebUI:
http://localhost:12222Point the MCP server at the RPC endpoint withRTORRENT_HOST/RTORRENT_PORT(see docs/RTORRENT_REFERENCE.md).
ruTorrent vs this projects webapp (web_sota/)
ruTorrent (bundled in CrazyMaxs image) is the full UI: plugins, RSS, autotools, labels, and a lot of surface area. Many people find it overcomplicated and the UI dated; it is still the right place when you need plugin workflows (RSS rules, auto-move, unpack, etc.) that we do not replicate.
Our webapp under web_sota/ is intentionally rudimentary: a small Vite + React dashboard on a REST bridge (/api/*) served by the same Python process as MCPsee web_sota/README.md. Today it is a light substitute for day-to-day glances: health, rTorrent probe, torrent list, magnet add. It is not a feature-complete ruTorrent replacement. Use it when you want something simple; keep ruTorrent (or MCP tools) when you need depth.
Run the stack (backend + Vite) with:
.\web_sota\start.ps1
Default dev URLs are documented in web_sota/README.md (Vite + uvicorn ports).
Optional: ruTorrent plugins (CrazyMax image)
The upstream image ships ruTorrent with many plugins; common automation-related ones include RSS/feeds, autotools, scheduler, unpack, ratio/seedingtime. See docs/RTORRENT_SETUP.md for a longer list and configuration notes.
Installation
rTorrent in Docker (CrazyMax image), ports, and webapp vs ruTorrent are covered under Quick Start rTorrent stack (Docker, recommended) above. This section is for the Python MCP package and optional desktop wiring.
Prerequisites
- uv installed (RECOMMENDED)
- Python 3.12+
Quick Start
Run immediately via uvx:
uvx rtorrent-mcp
Claude Desktop Integration
Add to your claude_desktop_config.json:
"mcpServers": {
"rtorrent-mcp": {
"command": "uv",
"args": ["--directory", "D:/Dev/repos/rtorrent-mcp", "run", "rtorrent-mcp"]
}
}
Platform setup
Quick Setup (Windows - Docker Recommended)
Prerequisites: Docker Desktop must be installed and running.
# Download the project files
# Place docker-compose.yml and install.bat in your desired directory
# Run the installation script
install.bat
# The script will:
# - Create necessary directories
# - Configure rTorrent with SCGI support
# - Start the Docker containers
# - Test the connection
Management Commands:
start.bat # Start rTorrent containers
stop.bat # Stop rTorrent containers
status.bat # Check container status and health
uninstall.bat # Remove everything
Alternative: WSL2 Setup
# Enable WSL2 (run as Administrator)
wsl --install -d Ubuntu
# Inside WSL2 Ubuntu
sudo apt update
sudo apt install rtorrent
Linux/macOS Setup
# Ubuntu/Debian
sudo apt update
sudo apt install rtorrent
# CentOS/RHEL/Fedora
sudo yum install rtorrent
# or
sudo dnf install rtorrent
# macOS
brew install rtorrent
# Verify SCGI support
rtorrent -h | grep -i scgi
Basic Configuration
For Docker (Windows):
-
Create rTorrent configuration
# Create config directory mkdir C:\rtorrent-mcp\config # Create rtorrent.rc configuration @" # SCGI configuration for MCP server scgi_port = 0.0.0.0:5000 # Basic settings session.path.set = /config/session directory.default.set = /downloads log.execute = /config/rtorrent.log # Performance settings max_uploads.set = 50 max_connections.set = 200 max_peers.set = 100 # Austrian Legal Compliance system.method.set_key = event.download.inserted_new, anime_category, "d.custom1.set=anime" "@ | Out-File -FilePath "C:\rtorrent-mcp\config\rtorrent.rc" -Encoding UTF8 -
Restart container to apply configuration
docker-compose restart -
Verify connection
# Test SCGI connection from Windows Invoke-RestMethod -Uri "http://localhost:5000/RPC2" -Method POST -ContentType "text/xml" -Body '<?xml version="1.0"?><methodCall><methodName>system.listMethods</methodName></methodCall>'
For WSL2/Linux/macOS:
-
Create rTorrent configuration
mkdir -p ~/.rtorrent cat > ~/.rtorrent.rc << 'EOF' # SCGI configuration for MCP server scgi_port = localhost:5000 # Basic settings session.path.set = ~/.rtorrent/session directory.default.set = ~/Downloads log.execute = ~/.rtorrent/rtorrent.log # Performance settings max_uploads.set = 50 max_connections.set = 200 max_peers.set = 100 EOF -
Start rTorrent daemon
# Start in background rtorrent -d # Or with systemd (create service) sudo systemctl start rtorrent sudo systemctl enable rtorrent -
Verify connection
# Test SCGI connection curl -X POST -H "Content-Type: text/xml" \ -d '<?xml version="1.0"?><methodCall><methodName>system.listMethods</methodName></methodCall>' \ http://localhost:5000/RPC2
For Windows, macOS, Docker, and advanced configuration options, see docs/RTORRENT_SETUP.md.
Configuration
-
Create a
.envfile (or set environment variables)# rTorrent settings RTORRENT_HOST=localhost RTORRENT_PORT=12224 RTORRENT_PATH=/var/lib/rtorrent/session # Nyaa.si settings NYAA_BASE_URL=https://nyaa.si # Application settings DEBUG=false LOG_LEVEL=INFO
Running the Server
# Run with stdio transport (for Claude Desktop)
python -m rtorrent_mcp.server --transport stdio
# Or with HTTP transport
python -m rtorrent_mcp.server --transport http
# Custom config file
python -m rtorrent_mcp.server --config /path/to/config.env
# Direct module execution
python src/rtorrent_mcp/server.py
Development
Testing
# Run all tests
uv run pytest
# Run with coverage report
uv run pytest --cov=rtorrent_mcp --cov-report=html
Code Style
# Format code with ruff
uv run ruff format .
# Lint code with ruff
uv run ruff check . --fix
# Type checking with pyright
uv run pyright
# Security scanning
uv run bandit -r src/
uv run safety scan
Features in Detail
Smart Anime Search
# Basic search
await search_anime("Detective Conan", resolution="720p", group="ASW")
# Advanced search with filters
await search_anime(
query="One Piece",
resolution="1080p",
group="Erai-raws"
)
rTorrent Integration
# Add torrent from magnet link
magnet = "magnet:?xt=urn:btih:..."
await add_torrent(magnet, category="anime")
# Monitor and manage downloads
await list_torrents()
await pause_torrent("torrent_hash")
await resume_torrent("torrent_hash")
await delete_torrent("torrent_hash", delete_files=True)
# Check connection status
await get_status()
(AT) Austrian Legal Compliance
# Check if content is safe for Austria
is_safe = await check_austrian_legal_status(torrent_info)
if is_safe:
await add_torrent(torrent_info["magnet"])
else:
logger.warning("Content may not be legal in Austria")
Extended Search Capabilities
# Search manga
await search_manga("One Piece", subcategory="translated")
# Search Japanese TV shows
await search_japanese_tv("Terrace House", subcategory="translated")
# Search movies on YTS
await search_movies("The Matrix", quality="1080p", sort_by="seeds")
# Search ebooks on Anna's Archive (60M+ books!)
await search_ebooks_annas("Python Programming", content_type="books")
# Search comics on Pirate Bay
await search_comics("Watchmen", max_results=20)
Metadata Services
# Get IMDb metadata for a movie
await get_imdb_metadata("The Matrix", year=1999)
# Search IMDb for multiple matches
await search_imdb("Matrix", year=1999)
# Get TVDB metadata (requires API subscription)
await get_tvdb_metadata("Breaking Bad", year=2008)
# Get detailed Anna's Archive torrent info
await get_annas_detail("https://annas-archive.org/...")
Post-Processing System
# Check for completed downloads
completed = await check_completed_downloads()
# Process a completed download
await process_completed_download("torrent_hash")
# Start automatic post-processing (background polling)
await start_post_processing()
# Stop post-processing
await stop_post_processing()
# Normalize a filename
normalized = await normalize_filename("Show.Name.S01E01.RELEASE-GROUP.mkv", category="tv")
Natural Language Processing
# English commands
await sandra_anime_command("get me this weeks asw anime, 720p")
# German commands
await sandra_anime_command("lade detective conan asw 720p")
# Parse commands without executing
await parse_anime_command("asw attack on titan 1080p")
# Get command help
await get_command_help()
System Tools
# Server help / tool listing
await help()
# System status and health check
await get_system_status()
# Analyze the repository
await analyze_repo()
Configuration Options
Environment Variables
| Variable | Default | Description |
|---|---|---|
RTORRENT_HOST |
localhost |
rTorrent SCGI host |
RTORRENT_PORT |
12224 |
rTorrent XML-RPC port |
RTORRENT_PATH |
/var/lib/rtorrent/session |
rTorrent session path |
NYAA_BASE_URL |
https://nyaa.si |
Nyaa.si base URL |
LOG_LEVEL |
INFO |
Logging level (DEBUG, INFO, WARNING, ERROR, CRITICAL) |
DEBUG |
false |
Enable debug mode |
ALLOWED_CATEGORIES |
["Anime"] |
Allowed content categories |
ALLOWED_RESOLUTIONS |
["720p", "1080p"] |
Allowed video resolutions |
MAX_TORRENT_SIZE_GB |
10 |
Maximum allowed torrent size in GB |
POST_PROCESSING_ENABLED |
false |
Enable automatic post-processing |
POST_PROCESSING_POLL_INTERVAL |
60 |
Seconds between polling for completed downloads |
DELETE_TORRENT_AFTER_COMPLETE |
true |
Remove torrent after completion |
NORMALIZE_FILENAMES |
true |
Normalize filenames before moving |
INGESTION_ANIME_PATH |
- | Path to temporary ingestion folder for anime |
INGESTION_TV_PATH |
- | Path to temporary ingestion folder for TV shows |
INGESTION_MOVIES_PATH |
- | Path to temporary ingestion folder for movies |
OMDB_API_KEY |
- | OMDb API key for IMDb metadata (free at omdbapi.com) |
TVDB_API_KEY |
- | TVDB API key for TV metadata (requires subscription) |
API_KEY |
- | Bearer/X-API-Key auth for REST API (optional, set to enable) |
NYAA_ASW_USERNAME |
AkihitoSubsWeeklies |
ASW user page on nyaa.si for direct lookup |
PIRATEBAY_BASE_URL |
https://thepiratebay10.xyz |
The Pirate Bay domain (changes frequently) |
RTORRENT_SAMPLING_BASE_URL |
http://127.0.0.1:11434/v1 |
OpenAI-compatible LLM endpoint (Ollama default) |
RTORRENT_SAMPLING_MODEL |
llama3.2 |
LLM model for agentic workflow |
RTORRENT_SAMPLING_USE_CLIENT_LLM |
- | Set to 1 to prefer host LLM over server-side |
PLEX_URL |
- | Plex server URL (enables library refresh after post-process) |
PLEX_TOKEN |
- | Plex authentication token |
JELLYFIN_URL |
- | Jellyfin server URL (enables library scan after post-process) |
JELLYFIN_API_KEY |
- | Jellyfin API key |
Media Service Integration
Architecture
Two paths, depending on whether *arr is in the loop:
Direct (anime via nyaa — no *arr)
rTorrent (rtorrent-mcp initiated)
→ PostProcessor (normalize + move to ingestion)
→ MediaIntegrator (scan Plex/Jellyfin)
For content downloaded directly through rtorrent-mcp (anime, manga from nyaa), the MediaIntegrator fires Plex/Jellyfin scans so files appear in your media libraries without waiting for a scheduled scan.
*arr-managed (movies/TV)
*arr (searches, decides what to grab)
→ *arr sends magnet/torrent to rTorrent (via Download Client config)
→ rTorrent downloads
→ *arr polls rTorrent' or watches folder
→ *arr imports + renames
→ *arr notifies Plex/Jellyfin
For *arr-managed content, configure rTorrent as a download client directly in Radarr/Sonarr (Settings > Download Clients > rTorrent). The *arr handles everything: dispatch, completion detection, import, and media server notification. No rtorrent-mcp integration needed.
Enable Plex/Jellyfin scanning
Set the URL + API key in .env:
# Plex
PLEX_URL=http://localhost:32400
PLEX_TOKEN=your_plex_token
# Jellyfin
JELLYFIN_URL=http://localhost:8096
JELLYFIN_API_KEY=your_jellyfin_key
Only services with both URL and key set are contacted — others are skipped silently.
Manual trigger
await torrent_management(action="notify_media", torrent_hash="...", category="tv")
This fires the scan pipeline for an already-processed torrent without re-running the file move.
Documentation
API Reference
For detailed API documentation, run the server and visit:
http://localhost:10910/api/health
REST endpoints on port 10910 (same process as MCP /mcp):
GET /api/health— liveness + versionGET /api/capabilities— tools/resources/skills surface (dynamic discovery)GET /api/skills/GET /api/skills/{name}— bundled SKILL.md listing/contentGET /api/llm/discover— probe Ollama :11434 / LM Studio :1234 / vLLM :8000POST /api/ai/chat— chat completion via the configured sampling endpointGET /api/rtorrent/status/GET /api/rtorrent/torrents/POST /api/rtorrent/magnetGET /api/fleet/apps— probe the fleet webapp reservoir for live peersGET /api/v1/diagnostics/GET /api/v1/system/info— CUA smoke diagnostics
Set API_KEY in .env to require Authorization: Bearer <key> on /api/*.
Product Requirements Document
See PRD.md for product background, requirements, and technical notes.
Extended Search Guide
See EXTENDED_SEARCH_GUIDE.md for complete guide to using all search capabilities including manga, movies, ebooks, comics, and metadata services.
Post-Processing Setup
See POST_PROCESSING_SETUP.md for complete post-processing configuration guide, including ingestion folder setup and Plex integration.
Status Report
See STATUS_REPORT.md for current project status, metrics, and development roadmap.
Development
-
Clone this repository and
cdinto it (or open an existing clone), then install development dependencies:git clone https://github.com/sandraschi/rtorrent-mcp.git cd rtorrent-mcp uv sync --dev -
Run tests:
uv run pytest -
Build documentation:
uv run mkdocs serveThen visit http://localhost:8001
Claude Desktop integration
The recommended mcpServers snippet is under Installation Claude Desktop Integration.
Manual MCP configuration
For advanced users or custom setups, manually configure Claude Desktop:
Location: ~/Library/Application Support/Claude/claude_desktop_config.json (macOS)
Location: %APPDATA%/Claude/claude_desktop_config.json (Windows)
Location: ~/.config/Claude/claude_desktop_config.json (Linux)
Add this configuration to your claude_desktop_config.json:
{
"mcpServers": {
"rtorrent-mcp": {
"command": "python",
"args": ["-m", "rtorrent_mcp.server", "--transport", "stdio"],
"cwd": "/path/to/your/rtorrent_mcp",
"env": {
"PYTHONPATH": "/path/to/your/rtorrent_mcp/src",
"RTORRENT_HOST": "localhost",
"RTORRENT_PORT": "12224",
"NYAA_BASE_URL": "https://nyaa.si"
}
}
}
}
Configuration Notes:
- Replace
/path/to/your/rtorrent_mcpwith your actual repository path - Adjust environment variables as needed for your setup
- The server will start automatically when Claude Desktop launches
Contributing
- Fork the repository
- Create a feature branch (
git checkout -b feature/-feature) - Commit your changes (
git commit -m 'Add some feature') - Push to the branch (
git push origin feature/-feature) - Open a Pull Request
🛡️ Industrial Quality Stack
This project adheres to SOTA 14.1 industrial standards for high-fidelity agentic orchestration:
- Python (Core): Ruff for linting and formatting. Zero-tolerance for
printstatements in core handlers (T201). - Webapp (UI): Biome for sub-millisecond linting. Strict
noConsoleLogenforcement. - Protocol Compliance: Hardened
stdout/stderrisolation to ensure crash-resistant JSON-RPC communication. - Automation: Justfile recipes for all fleet operations (
just lint,just fix,just dev). - Security: Automated audits via
banditandsafety.
License
This project is licensed under the MIT License - see the LICENSE file for details.
Acknowledgments
- rTorrent - The lightweight torrent client
- Nyaa.si - For the anime torrents
- [FastMCP](https://FastMCP 3.1.0anthropic.com) - The MCP framework
- Claude Desktop - For MCP integration
Legal compliance (examples)
await check_legal_status("austria")
await check_legal_status("germany")
Interpretation of results is on you; this is not legal advice.
Release group priorities (defaults)
Heuristic ordering used by search helpers (tune in config as needed):
- ASW
- SubsPlease
- Erai-raws
- EMBER
- Judas
Austrian context (AT)
- Tooling includes AT-oriented legal risk hints in outputs; verify locally.
- Command parsing supports English and German where implemented.
Configuration
Copy .env.example to .env and configure:
RTORRENT_HOST=localhost
RTORRENT_PORT=12224
NYAA_BASE_URL=https://nyaa.si
ALLOWED_CATEGORIES=Anime
ALLOWED_RESOLUTIONS=720p,1080p
DEFAULT_RESOLUTION=720p
PREFERRED_RELEASE_GROUP=ASW
LOG_LEVEL=INFO
Testing
Run Tests
# Run all tests with coverage
pytest
# Run specific test categories
pytest -m unit # Unit tests only
pytest -m integration # Integration tests only
# Run with verbose output
pytest -v
# Generate coverage report
pytest --cov=rtorrent_mcp --cov-report=html
Test Structure
tests/
conftest.py # Test configuration and fixtures
unit/ # Unit tests (isolated components)
test_rtorrent_client.py
integration/ # Integration tests (full workflows)
test_mcp_integration.py
PowerShell Test Runner
Windows users can use the PowerShell test runner:
# Run all tests
.\scripts\run-tests.ps1
# Run with coverage
.\scripts\run-tests.ps1 -Coverage
# Run unit tests only
.\scripts\run-tests.ps1 -Unit
Modern Development Commands
With UV installed, from a clone of this repo at the repository root, you can use these modern commands:
# Install all dependencies (including dev tools)
uv sync --dev
# Run linting and formatting
uv run ruff check . --fix
uv run ruff format .
# Run type checking
uv run pyright
# Run security scans
uv run bandit -r src/
uv run safety scan
# Run tests with coverage
uv run pytest --cov=src/rtorrent_mcp --cov-report=html
# Build package
uv build
# Validate package
uv run twine check dist/*
CI, tests, and checklist
- CI: GitHub Actions runs lint, type check, and tests (see
.github/workflows/) - Tests:
uv run pytest(coverage optional viapytest --cov) - Self-review:
docs/MCP_PRODUCTION_CHECKLIST.mdis a checklist for hardening; it is not a third-party certification.
Treat this project like any other self-hosted tool: verify behaviour in your environment and keep dependencies updated.
Legal Disclaimer
This tool is designed for Austrian legal context where personal downloading is generally tolerated. Users in other jurisdictions should research local copyright laws. High-risk countries (Germany, Japan) require additional precautions.
Dependencies
- FastMCP 3.1.0+: MCP server framework with stdio transport
- UV: Modern Python package manager for fast, reliable builds
- aiohttp: Async HTTP client for indexer APIs
- beautifulsoup4: HTML parsing for search results
- xmlrpc.client: rTorrent SCGI communication (Python stdlib)
- psutil: System monitoring and health checks
- pydantic: Data validation and settings management
- python-dotenv: Environment configuration
Development Dependencies
- ruff: Fast Python linter and formatter
- pyright: Type checking and static analysis
- bandit: Security vulnerability scanner
- safety: Dependency vulnerability scanner
- pytest: Testing framework with coverage
- build & twine: Package building and publishing
Author
Maintainer: sandraschi / rtorrent-mcp contributors.
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。