MCPJungle is an MCP server that acts as a centralized self-hosted registry and gateway for managing multiple Model Context Protocol servers behind a single endpoint. It connects to both remote streamable HTTP servers and local STDIO-based servers, storing configuration and tool states in PostgreSQL or SQLite databases. Developers, engineering teams, and AI agent builders use MCPJungle to eliminate scattered client-side configurations across Claude Desktop, Cursor, Copilot, and custom agent frameworks. Instead of manually configuring every tool and prompt in every individual development environment, operators register all internal and third-party MCP servers inside MCPJungle once. The platform then exposes a unified HTTP endpoint that multiplexes tool discovery, resource lookup, and tool execution. It also enables grouping tools into distinct profiles, globally toggling tool availability, and managing access control or OpenTelemetry observability in enterprise deployments. By consolidating disparate tool providers into one stable access layer, MCPJungle standardizes how private agents interact with database tools, APIs, and developer utilities.
Category: AI & LLM Tooling
Tags: postgresql, registry, self-hosted, sqlite
bash curl -O https://raw.githubusercontent.com/mcpjungle/MCPJungle/refs/heads/main/docker-compose.yaml docker compose up -d 2. Install the mcpjungle command-line tool via Homebrew: bash brew install mcpjungle/mcpjungle/mcpjungle 3. Register an upstream MCP server using the CLI: bash mcpjungle register --name context7 --url https://mcp.context7.com/mcp 4. Configure Claude Desktop to connect to MCPJungle by updating its configuration file: json { "mcpServers": { "mcpjungle": { "command": "npx", "args": [ "mcp-remote", "http://localhost:8080/mcp", "--allow-http" ] } } }Part of MCP Servers
MCPJungle is a self-hosted registry and gateway designed to run multiple Model Context Protocol servers behind a single HTTP endpoint. Instead of configuring each server separately in every client, users register their tools, resources, and prompts in MCPJungle once and connect their clients to its unified URL.
MCPJungle works with any AI client or agent framework that supports the Model Context Protocol, including Claude Desktop, Cursor, and GitHub Copilot. Clients connect to MCPJungle over its streamable HTTP endpoint or through stdio bridge commands like npx mcp-remote.
You can run the MCPJungle server using Docker Compose to launch the gateway along with a PostgreSQL database. The mcpjungle command-line interface can be installed on macOS via Homebrew or downloaded directly as a standalone binary from the project releases page on GitHub.
MCPJungle supports persistent storage using PostgreSQL or SQLite. Local personal setups can use SQLite or Docker Compose with PostgreSQL, while team or enterprise deployments typically connect to an external managed PostgreSQL cluster to centralize server definitions, permissions, and tool states.
Yes, MCPJungle is an open-source project hosted on GitHub. It provides open-source container images, binaries, and documentation, allowing individual developers and teams to deploy, self-host, and inspect the registry gateway in their own infrastructure without external vendor dependencies.