OpenLander: An MCP-Native Self-Hosted Deployment Platform for the AI Era

OpenLander is an open-source self-hosted deployment platform designed specifically for AI coding tools. It supports the MCP protocol, enabling agents like Cursor and Claude Code to directly operate deployment workflows and achieve full automation from code to production. It addresses the pain point of difficult deployment in AI development, providing a cost-effective solution that combines Coolify's Docker foundation, Vercel's simple user experience, and native MCP support. It is suitable for scenarios such as individual developers and AI teams.

Today, while AI programming tools (e.g., Cursor, Claude Code) simplify development, deployment still requires knowledge of Docker, SSL, etc., or paying high cloud service fees. OpenLander's core concept combines Coolify's Docker foundation + Vercel's simple experience + native MCP operation support, providing a web dashboard and AI agent interaction capabilities. In terms of cost: deploying 5 services on traditional cloud platforms costs about $100 per month, while a Mac Mini + OpenLander requires a one-time investment of $600, which breaks even after 6 months and then incurs zero cost.

OpenLander adopts a layered model of Project (workspace) and Service (deployable unit). A Service contains complete information such as code repository, image, Dockerfile, etc. The deployment workflow is simplified as: paste Git URL → click deploy → get accessible URL. If the build fails, logs can be checked via the dashboard or MCP to troubleshoot; when a container crashes, the platform records status changes and logs, and AI agents can perform repairs and redeployments.

MCP (Model Context Protocol) is a key feature of OpenLander, supporting direct interaction with AI tools like Cursor and Claude Code to achieve: log reading (analyze build/operation issues), status monitoring (container health and resources), automatic repair (rebuild/rollback), and deployment operations (full workflow automation). This design extends the capabilities of AI tools from writing code to managing deployment, realizing end-to-end DevOps integration.

OpenLander is built on Docker and Traefik, supporting Linux/macOS (Windows via WSL2). It uses PostgreSQL for data storage and Docker Compose for orchestration. Quick start command: OPENLANDER_POSTGRES_PASSWORD='change-me' docker compose up -d --build. After starting, visit http://localhost:10114 to complete initialization (language, admin password, GitHub integration, etc.). Note: The 0.1 version has a single-process limit; it is recommended to deploy a single instance in a production environment.

Features: Web dashboard (project overview, deployment history, real-time log streaming, web terminal, etc.); deployment and operation capabilities (one-click Git deployment, Traefik automatic routing, blue-green deployment, public sharing, etc.).

Comparison:

Feature	Coolify/Dokploy	OpenLander
Interface	Web dashboard	Web dashboard + MCP agent workflow
Build failure handling	Manual repair by user	Expose logs for agent analysis
AI support	None	MCP protocol supports Cursor, etc.
Server awareness	Manual configuration	Automatic detection of ports/proxies/containers

OpenLander is suitable for: 1. Individual developers (low-cost self-hosting of multiple projects); 2. AI-assisted development teams (extending agents to deployment); 3. Edge computing scenarios (local/edge deployment); 4. Privacy-sensitive applications (fully self-hosted).

Recommendations: Use pre-built images for memory-constrained hosts; set OPENLANDER_IMAGE=ghcr.io/openlander-ai/openlander:0.1.0 to lock the version and avoid automatic updates.

OpenLander uses the AGPL-3.0 license and is hosted on GitHub; community contributions are welcome.

Summary: OpenLander represents a new direction for deployment platforms—from manual operation to agent collaboration. By extending AI tool capabilities to DevOps via MCP, it provides a complete solution for software development in the AI era, and is worth trying for developers pursuing cost-effectiveness, privacy control, and automation.