Run GUI, Headless, and MCP Modes¶

Koharu can run as a normal desktop app, a headless local server with a Web UI, or an MCP server for AI agents. These are not separate backends. They all sit on top of the same local runtime and HTTP server.

What stays the same across modes¶

No matter how you launch Koharu, the runtime model is the same:

• the server binds to 127.0.0.1
• the UI and API are served by the same local process
• the page pipeline, model loading, and exports use the same internal code paths

That is why desktop editing, headless automation, and MCP tooling stay aligned.

Mode summary¶

Run the desktop app¶

Launch Koharu normally from your installed application.

Even in desktop mode, Koharu still starts a local HTTP server internally. The embedded window talks to that local server rather than calling the pipeline directly.

This is the default mode and is the best choice for most users.

Run headless mode¶

Headless mode starts the local server without opening the desktop GUI.

# macOS / Linux
koharu --port 4000 --headless

# Windows
koharu.exe --port 4000 --headless

After startup, open the Web UI at http://localhost:4000.

Headless mode stays in the foreground until you stop it, typically with Ctrl+C.

Run with a fixed port¶

By default, Koharu uses a random local port. Use --port when you need a stable address for bookmarks, scripts, reverse proxies, or MCP clients.

# macOS / Linux
koharu --port 9999

# Windows
koharu.exe --port 9999

If you do not specify --port, Koharu still starts the server, but the chosen port is dynamic.

Connect to the local API¶

When Koharu is running on a fixed port, the main endpoints are:

• Web UI: http://localhost:9999/
• RPC / HTTP API: http://localhost:9999/api/v1
• MCP server: http://localhost:9999/mcp

Replace 9999 with the port you chose.

Because Koharu binds to loopback, these endpoints are local by default. If you want remote access from another machine, you need to expose that port yourself through your own network setup.

For endpoint-level details, see HTTP API Reference.

Connect to the MCP server¶

Koharu includes a built-in MCP server using the same loaded documents, models, and page pipeline as the rest of the app.

Point your MCP client or agent at:

http://localhost:9999/mcp

This is useful when you want an agent to:

• inspect text blocks
• run OCR or translation
• export rendered pages
• automate review or batch workflows

For client-specific setup examples, see Configure MCP Clients.

For the built-in tool list itself, see MCP Tools Reference.

Force CPU mode¶

Use --cpu when you want to disable GPU inference explicitly.

# macOS / Linux
koharu --cpu

# Windows
koharu.exe --cpu

This is useful for compatibility testing, driver issues, or low-risk debugging when GPU setup is uncertain.

Download runtime dependencies only¶

Use --download if you want Koharu to prefetch runtime dependencies and exit without starting the app.

# macOS / Linux
koharu --download

# Windows
koharu.exe --download

In the current implementation, this path initializes:

• runtime libraries used by the local inference stack
• the default vision and OCR models

It does not predownload every optional local translation LLM. Those are still fetched when you select them in Settings.

Enable debug output¶

Use --debug when you want console-oriented startup with log output.

# macOS / Linux
koharu --debug

# Windows
koharu.exe --debug

On Windows, debug and headless runs also influence how Koharu attaches to or creates a console window.