Doc2x

This MCP server wraps the Doc2x v2 PDF/image interfaces into semantic tools that you can invoke from an MCP client. It runs as a local or stdio-based service, enabling streamlined document parsing, conversion, and layout extraction without exposing low-level API calls.

How to use

You run the MCP server as a local process and connect to it through your MCP client. You can start the server in one of two ways: using npx to run the prebuilt package, or running the local source after building. Once running, you can execute the available tools to submit tasks, poll for statuses, wait for results, and download outputs. Use the client to invoke the tool names described in the Tools list, and pass the appropriate parameters as required by each tool.

Typical usage pattern you will follow: start the server, then submit a document or image processing task, monitor its progress, and finally retrieve results or artifacts. You will provide an API key to authorize requests, and you can optionally adjust timeouts and poll intervals to suit your environment.

How to install

Prerequisites you need before installing:

• Node.js version 18 or newer
• Access to a terminal or shell with network outbound access for downloading packages and API calls

Option A — run via npx
```json
{
  "command": "npx",
  "args": ["-y", "@noedgeai-org/doc2x-mcp"],
  "env": {
    "DOC2X_API_KEY": "sk-xxx",
    "DOC2X_BASE_URL": "https://v2.doc2x.noedgeai.com"
  }
}

Note: This uses the prebuilt MCP server package available on the npm registry. You provide your API key and optionally override the base URL.

Option B — run from local source First, install dependencies and build the project, then start the server with your API key. The following sequence assumes the repository is checked out locally at the default path.

doc2x-mcp
npm install
npm run build
DOC2X_API_KEY=sk-xxx npm start

Or, run it directly via node from the built distribution:

node <ABS_PATH>/doc2x-mcp/dist/index.js

Environment variables you pass here should include the API key and (optionally) the base URL as shown above.

## Additional configuration notes

Configuration is provided through environment variables. The following variables are supported: DOC2X\_API\_KEY (required), DOC2X\_BASE\_URL (optional, default https://v2.doc2x.noedgeai.com), DOC2X\_HTTP\_TIMEOUT\_MS (optional, default 60000), DOC2X\_POLL\_INTERVAL\_MS (optional, default 2000), DOC2X\_MAX\_WAIT\_MS (optional, default 600000), and DOC2X\_DOWNLOAD\_URL\_ALLOWLIST (optional, default ".amazonaws.com.cn,.aliyuncs.com,.noedgeai.com"; set to \* to allow any host, though this is not recommended). Ensure you provide the API key securely and do not expose it in logs.

## Available tools

### doc2x\_parse\_pdf\_submit

Submit a PDF parse task to extract semantic content from a PDF document.

### doc2x\_parse\_pdf\_status

Query the status of a previously submitted PDF parse task.

### doc2x\_parse\_pdf\_wait\_text

Wait for and retrieve textual content parsed from a PDF.

### doc2x\_convert\_export\_submit

Submit a request to convert and export a document or data set.

### doc2x\_convert\_export\_result

Fetch the result of a previously submitted export task.

### doc2x\_convert\_export\_wait

Wait for export task completion and retrieve outputs.

### doc2x\_download\_url\_to\_file

Download a file from a URL to a local file path.

### doc2x\_parse\_image\_layout\_sync

Synchronously parse the layout of an image to determine structure.

### doc2x\_parse\_image\_layout\_submit

Submit an image layout analysis task.

### doc2x\_parse\_image\_layout\_status

Check the status of an image layout analysis task.

### doc2x\_parse\_image\_layout\_wait\_text

Wait for and retrieve text from image layout analysis.

### doc2x\_materialize\_convert\_zip

Materialize and convert a ZIP package produced by Doc2x workflows.

### doc2x\_debug\_config

Retrieve or adjust debug/configuration options for the MCP server.