Configure Custom Servers

When the server you need is not in the registry, you can add it manually by providing a JSON configuration. This works with any MCP-compatible server — internal tools, private APIs, or open-source servers.

Step 1: Open the Tools Page

Click Tools in the top navigation bar to view your installed tools and servers.

Step 2: Start Adding a Custom Server

Click Add Tools at the top of the page. In the panel that opens, you can configure a custom server by providing its JSON configuration.

You can also manage servers under Workspace settings > MCP Servers.

Step 3: Provide the Server Configuration

Enter a name for the server and paste your JSON configuration into the editor. The configuration tells Skilder how to start the server process on its cloud infrastructure.

Configuration Schema

Example: Custom API Tool Server

This configuration runs an MCP server that wraps a custom internal API:

{
  "command": "npx",
  "args": ["-y", "@your-org/internal-api-mcp-server"],
  "env": {
    "API_BASE_URL": "https://api.internal.example.com",
    "API_TOKEN": "your-api-token-here"
  }
}

Example: Python Server

This configuration starts a Python-based MCP server:

{
  "command": "python",
  "args": ["-m", "my_mcp_server"],
  "env": {
    "DATABASE_URL": "postgresql://db.example.com:5432/mydb",
    "LOG_LEVEL": "info"
  }
}

Example: Docker-Based Server

This configuration runs an MCP server packaged as a Docker image. The container runs within Skilder's cloud infrastructure.

{
  "command": "docker",
  "args": ["run", "--rm", "-i", "ghcr.io/your-org/mcp-server:latest"],
  "env": {
    "AUTH_SECRET": "your-secret-value"
  }
}

Step 4: Handle Secrets

If your configuration includes sensitive values (API keys, tokens, passwords), Skilder encrypts them before storage. Enter secret values directly in the env block — Skilder detects fields marked as secrets and protects them automatically.

You can also update secrets later from the server's detail page without re-entering the full configuration.

Step 5: Save and Verify

Click Save. Skilder starts the server on its cloud infrastructure and the server's tools appear in the Tools page.

Check the server status under Workspace settings > MCP Servers to confirm it is Running. If the status shows Error, click the server to review the error message. Common causes include:

• Incorrect command or args — Make sure the executable exists and the arguments are valid.
• Missing environment variables — Verify that all required env values are set.
• Image pull failures — For Docker-based servers, confirm the image reference is correct and accessible.

Updating a Custom Server

To change the configuration of a custom server after creation:

1. Open Workspace settings > MCP Servers and click the server to open its detail page.
2. Edit the configuration fields — command, arguments, or environment variables.
3. Click Save. Skilder restarts the server with the updated configuration.

Next Steps

• Browse & Install from Registry — See what pre-configured servers are available.
• Skills — Combine tools from your custom server into skills that agents can use.