Integrations · AI Connections

Use pdfAssistant Inside Your AI Assistant

pdfAssistant connects directly to your AI assistant through the Model Context Protocol (MCP) — an open standard that lets AI models securely discover and execute external tools. Process, convert, merge, compress, redact, OCR, watermark, and encrypt PDFs without leaving your conversation. Authenticate in under 60 seconds with secure OAuth single sign-on.

Paid plan required · Not available on the free tier

Supported Platforms

Connect from the AI client you already use.

One endpoint, three first-class integrations. Pick yours to see the setup guide.

pdfAssistant with Claude

Full MCP integration with Claude Desktop via Streamable HTTP. Set up in one CLI command.

View setup guide →

pdfAssistant with ChatGPT

Connect via Developer Mode with remote Streamable HTTP transport. OAuth login in seconds.

View setup guide →

pdfAssistant with Codex

Native CLI integration with Codex. Map the server URL and authenticate via OAuth.

View setup guide →

How It Works

Three steps. Under 60 seconds.

From a fresh AI client to a fully connected PDF toolbox in your conversation. See paid plans →

1. Link Your Account

Open your AI client's settings, add the pdfAssistant remote server endpoint, and click Link Account.

2. Secure OAuth Login

A popup opens. Log in with your pdfAssistant credentials via OAuth to authorize your AI assistant to use your tools and credits. Your passwords and API keys are never exposed.

3. Start Processing

Drop a document into your chat and describe what you need. Your assistant calls pdfAssistant's tools, processes the file in the cloud, and delivers the result in your conversation.

What You Can Do

Every PDF operation, exposed to your AI.

The complete toolset your AI assistant can call once pdfAssistant is connected via MCP.

CategoryAvailable ToolsExample Prompt
ConvertWord, Excel, PowerPoint to PDF; Webpage URL to PDF; HTML file to PDF; Email to PDF; PDF to Word/Excel/PowerPoint; PDF to image formats (PNG, JPEG, TIFF, GIF, BMP); Graphics to PDF“Convert this live pricing webpage URL and this PowerPoint presentation into PDFs.”
Combine & SplitMerge multiple PDFs, Split PDF by page range, ZIP/unZIP files“Merge these three contracts into a single file and ZIP the final output.”
OptimizeCompress PDF, Linearize PDF (web-optimize), Flatten forms, Flatten annotations, Flatten layers, Flatten transparencies, Convert colors, Convert to PDF/X, Rasterize PDF“Compress this PDF to under 5MB and linearize it so it opens fast on the web.”
Secure & EditEncrypt PDF, Add restrictions, Remove encryption, Remove restrictions, Text redaction (via literal strings, presets, or regex), Add text/images to PDF, Attach files to PDF, Add watermark“Use a regex pattern to redact all Social Security Numbers from this document text.”
ExtractExtract text, Extract images, Export form data, File read/write/append/list helpers“Extract all the data from the form fields in this document.”
AnalyzeCheck document metadata (accessibility tags, security settings, creator/origin info, text-extractability status)“Check the metadata of this file to see if it is fully tagged for accessibility.”
Sign & FormsDigitally sign PDF, Import form data, Convert XFA to AcroForms“Convert this old XFA form into a standard AcroForm so I can sign it.”
ArchiveConvert to PDF/A (archival standard)“Convert this contract to PDF/A for long-term archival.”

Why MCP

From Document Comprehension to Document Manipulation

Most AI assistants can talk about a document — summarize it, extract key points, answer questions about its content. But they cannot safely perform structural modifications to the file itself. They cannot compress an oversized presentation, flatten form layers, redact sensitive data with regex patterns, or convert raw images into searchable text.

pdfAssistant bridges that gap. When connected via MCP, your AI moves from basic text comprehension to active document manipulation — with a professional-grade toolbox it can combine on the fly to handle real document workflows end to end.

Security & Privacy

Built for documents you can't afford to leak.

OAuth 2.0 authentication. TLS 1.3 in transit. Zero retention after the job completes.

24-Hour File Deletion

Uploaded files are processed in encrypted storage and permanently deleted within 24 hours of task completion — not stored long-term.

pdfAssistant Doesn't Train On Your Data

pdfAssistant never uses your documents, prompts, or extracted metadata to train AI models. Note: when you connect third-party LLMs (Claude, ChatGPT, Codex, etc.) over MCP, the content you share with them is subject to their own retention and training policies.

Encrypted in Transit

All file payloads and tool calls use TLS 1.3 encryption end-to-end.

Architecture

Built on Open Standards

pdfAssistant's MCP server runs on a high-performance Streamable HTTP transport architecture using the JSON-RPC 2.0 protocol. All connections route through a single secure endpoint:

https://chat.pdfassistant.ai/mcp

Authentication is handled via standard OAuth 2.0 — your credentials are never exposed to your local LLM environment or third-party interfaces.

Ready to put pdfAssistant to work in your AI assistant?

Connect once. Process forever.

MCP server access requires a paid pdfAssistant plan