MCP Manager is a Blazor-based dashboard that helps you discover, install, and manage Model Context Protocol (MCP) servers across multiple AI agents. It provides a unified interface to browse MCP registries, install servers to supported agents, and monitor active installations. The tool emphasizes a plugin-like architecture with pluggable agent connectors and registries, enabling you to extend support for additional agents or MCP sources as your needs evolve. Use the dashboard to see which MCP servers are installed on each agent, perform bulk operations, and keep your local agent ecosystem organized from a single UI.

Prerequisites:

• .NET 10.0 SDK installed on your machine
• Access to the internet to restore NuGet packages
• Optional: Docker if you want to run containerized (see Docker option in the README)

Install steps:

1. Clone the repository git clone https://github.com/JerrettDavis/McpManager.git cd McpManager

2. Restore and run the web project (development workflow)

   From repository root

   dotnet restore dotnet run --project src/McpManager.Web

3. Open the application Navigate to https://localhost:5001 (or the configured URL if you changed ASPNETCORE_URLS)

Production build (optional): 4) Publish for a specific runtime dotnet publish -c Release -r win-x64 --self-contained

or linux/osx variants

5. Run the published output ./McpManager.Web (on Linux/macOS) or McpManager.Web.exe (on Windows)

Notes and tips:

• The MCP Manager supports multiple agents and registries via a plug-in architecture. If you need to add a new agent or registry, implement the corresponding interface and register it in Program.cs.
• Ensure your environment variables (like ASPNETCORE_URLS) do not conflict with other services on your host.
• For production deployments, consider running behind a reverse proxy (NGINX/Apache) and enabling HTTPS.
• If you encounter port binding issues, verify that the host port (default 5000/5001) is free or adjust ASPNETCORE_URLS accordingly.
• Docker deployments are supported via the repository’s Docker guidance; use the official GHCR image as shown in the docs.