MCP SERVICE

PDF Toolkit.

Bring PDF and image processing into your AI assistant. With MCP, convert, split, merge, watermark, and track document tasks in plain language.

Create an API key View integration docs

View on GitHub

MCP service URL

Streamable HTTP

MCP Ready

https://api.tensormaster.com/pdf/mcp

OAuth 2.1 + PKCE

API Key compatible

WHY PDF TOOLKIT

File processing built for Agents

Let your AI assistant call a reliable, traceable PDF workflow directly instead of switching between separate tools. It fits document processing, archiving, and e-commerce operations.

Multi-format conversion

Convert PDFs or images to Word, Excel, and Markdown for immediate editing.

Composable processing

Split, merge, watermark PDFs, and create PDFs from images in one service.

Trackable async tasks

Separate task creation, status checks, and result retrieval for large files and batches.

CLIENT QUICK START

Quick notes for popular clients

Menu names may vary slightly, but the flow is the same: enter the URL, authorize, and confirm that the tools are available.

Doubao

Enter the URL in Remote MCP / Connector, click “Authorize”, agree to the scopes, and confirm the connection.

Claude

Ask Claude to add the connector in Settings, restart it, then use /mcp to authorize the PDF service.

Codex (ChatGPT)

Open Settings → Plugins → Add MCP server, choose Streamable HTTP, save, then click Authenticate.

HOW TO CONNECT

Choose the connection method for your client

OAuth is best for clients such as Doubao, Claude, and ChatGPT. Use an API Key when the client only supports custom headers.

OAuth authorization (recommended)

Complete browser authorization once in an OAuth-capable MCP Client. Token refresh is handled automatically.

1.Add a remote MCP server and enter the service URL above.
2.Click “Authorize” or “Authenticate”, then register or sign in to MiaoDashi.
3.Review the PDF tool scopes, click “Agree”, and return to confirm the connection.

Manual API Key

Use this with MCP Clients, curl, or local tools that support custom headers.

1.Open API keys & authorizations in your account and create an API Key.
2.Add the Authorization header in your MCP Client.
3.Copy the JSON below into the Agent's configuration field.

{
  "mcpServers": {
    "pdf": {
      "type": "streamable-http",
      "url": "https://api.tensormaster.com/pdf/mcp",
      "headers": {
      "Authorization": "Bearer md_api_xxx"
      }
    }
  }
}

AVAILABLE TOOLS

One toolkit for the full workflow

MCP Clients discover the tools and schemas automatically. Describe the goal and let the Agent organize the calls.

PDF / image to DOCX

convert_file_to_word

Try online

PDF / image to XLSX

convert_file_to_excel

Try online

PDF / image to Markdown

convert_file_to_markdown

Try online

Split a PDF by page count

split_pdf

Try online

Merge PDFs in array order

merge_pdfs

Try online

Create a PDF from images

images_to_pdf

Try online

Add a text watermark

add_pdf_watermark

Try online

Query task status and history

get_pdf_task / list_pdf_tasks

MCP

FAQ

Frequently asked questions

Do I need to upload files to the MCP service?+

MCP requests use reachable temporary HTTPS file URLs. The client or Agent should map attachments to file_url, file_urls, or image_urls before calling a tool.

Which should I use: OAuth or an API Key?+

Prefer OAuth when browser authorization is supported. Use an API Key when the client can only configure custom headers. Both methods follow the account permissions and credit balance.

How do I retrieve a task result?+

Save the task_id after creating a task, then use get_pdf_task to query queued, running, succeeded, or failed. Do not create duplicate tasks just to poll.

File processing consumes account credits. See the corresponding tool page for the current credit cost.