Provides tools for interacting with Git repositories, including status, diff, log, branch management, add, commit, pull, push, and show operations.
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type @ followed by the MCP server name and your instructions, e.g., "@GPT Windows Connector list the files in my project workspace"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
Website: https://lucasmcp.com/
Lucas is an MCP-native bridge that lets compatible AI assistants work on computers you explicitly authorize: files, terminal, Git, browser, and desktop applications. The computer makes an outbound secure WebSocket connection to the Lucas Gateway, so no inbound port is required on the user's computer.
Create a Lucas account or sign in with Google.
Install and open Lucas Node on Windows.
In Dashboard → Computers, enter the computer Node ID and its current 8-digit Connection Code.
Approve the requesting account locally and choose its access policy / Allowed Folders.
Connect an MCP-compatible AI client to Lucas and authorize the same account.
The AI can execute only through computers and local permissions that account is allowed to use.
Previously approved accounts remain connected until their local access is revoked. A computer can authorize multiple Lucas accounts.
Related MCP server: WinCommander
Plan | Monthly | Annual | Requests / month | Active Computers | AI accounts |
Free | $0 | — | 1,000 | 1 | 1 |
Pro | $9.99 | $99.99 | 25,000 | 3 | 1 |
Pro+ | $19.99 | $199.99 | 100,000 | 6 | 3 |
Expansion Pack (Pro+ only): $14.99/month or $149.99/year, adding 100,000 Requests, 6 active Computers, and 1 AI account.
When a new account is created through a referral link, both users receive 1,000 bonus Requests. When the referred user first becomes a paid customer, the referrer receives an additional 10,000 Requests.
Local Node approval is the authority for computer access.
Allowed Folders constrain direct file access and workspace operations.
Passwords are hashed with Argon2.
Browser sessions use HttpOnly authentication cookies.
Node transport uses a persistent device credential and WSS.
Sensitive fields are redacted from dashboard audit output.
Admin accounts are never created by public “first signup” behavior; production deployments must configure GWC_SUPER_ADMIN_EMAIL.
Shell and desktop actions still execute with the Windows account's OS permissions. For higher-risk workloads, use a dedicated Windows account or VM and grant the minimum Lucas permissions required.
Website: https://lucasmcp.com/
Dashboard: https://lucasmcp.com/dashboard
Pricing: https://lucasmcp.com/pricing
MCP: https://lucasmcp.com/mcp
Node WSS: wss://lucasmcp.com/ws/node
Health/readiness: https://lucasmcp.com/health
Python 3.11+:
git clone https://github.com/Neal86/Lucas.git
cd Lucas
python -m venv .venv
# Windows: .venv\Scripts\activate
# macOS/Linux: source .venv/bin/activate
pip install -e ".[dev]"
pytest -qRun the gateway:
gwc-gatewayThe production deployment uses the included Dockerfile. Keep GWC_DATA_DIR on persistent storage so the database and generated JWT secret survive container replacement.
Set GWC_PUBLIC_BASE_URL=https://lucasmcp.com.
Set GWC_SUPER_ADMIN_EMAIL.
Keep GWC_DATA_DIR on persistent storage.
Configure Turnstile and SMTP/email verification.
Configure Google OAuth redirect URI exactly.
Configure all six Stripe Price IDs and the Stripe webhook secret.
Confirm /health reports critical readiness.
Require the latest main CI to pass on Windows and Ubuntu before deployment.
Customer-facing Privacy, Terms, Refund, and Contact pages are linked from the Lucas website.
This server cannot be deployed
Shared memory and actions for Claude, Kiro, OpenAI, Cursor, and other MCP-compatible AI clients.
Remote MCP server for supportsheep: run AI interviews and manage support content for your blog.
Talk to your public-facing AI from any MCP client — Claude, ChatGPT, Cursor, Cline, Windsurf.
OCR, transcription, file extraction, and image generation for AI agents via MCP.
