pdf-mcp

Turn tables in PDF and Word documents into Excel, CSV, or JSON. Use the simple browser app with no agent setup, or connect the same deterministic extraction tools to an AI agent through MCP.

The browser app is for direct table conversion. It previews the extracted rows, highlights basic structure warnings, and keeps files on your computer. Advanced profile checking can additionally validate named fields, preserve source evidence, and return accepted, needs_review, or rejected for a recurring document workflow. Cell values come from document parsers, not generated model output.

Need a profile and measured baseline for real documents? See the Document Reliability Pilot. Agents can use the structured offer at offers/document-to-excel-pilot.json.

Website: simple conversion, local tools, and the free hosted agent beta.

Current limitation: PDF extraction supports born-digital documents. Scanned or image-only PDFs require OCR, which is not included. An accepted decision means the configured checks passed; it does not prove that the source document itself is correct.

Simple browser app

This is the easiest way to use pdf-mcp. It requires no Python, terminal, MCP client, or agent configuration after downloading the app.

1. Open the v0.4.0 release.
2. Download the pdf-mcp-app ZIP for 64-bit Windows, an Apple silicon Mac, or 64-bit Linux.
3. Unzip it and open pdf-mcp-app.
4. Your browser opens. Drop in a PDF or Word document, choose Excel, CSV, or JSON, and select Convert document.

Use Stop app in the browser when finished. The app listens only on your computer. Temporary document copies are deleted immediately after conversion, and prepared downloads expire after 30 minutes or when the app stops. The community beta downloads are currently unsigned, so the operating system may ask you to confirm that you want to open them.

If Python is already installed, the same interface can be started with:

python -m pip install "pdf-agent-mcp @ https://github.com/wesseltl/pdf-mcp/releases/download/v0.4.0/pdf_agent_mcp-0.4.0-py3-none-any.whl"
pdf-mcp-app

The simple app performs raw table extraction. “No basic structure problems detected” is not an accuracy guarantee. Check important values before using them, or use a profile-checked pilot for a recurring business workflow.

For agents: local or free hosted beta

The normal pdf-agent-mcp command remains local, MIT licensed, network-free, and telemetry-free. For people who explicitly want a measured agent beta, pdf-agent-cloud-mcp uploads only the selected document for one authenticated operation.

Profile checking and evaluation currently run in the local edition or as part of the paid pilot. The free hosted beta currently exposes raw text, table, and CSV extraction only.

The free hosted beta includes 25 operations per calendar month. Temporary uploads are deleted when each request completes. Usage metrics exclude filenames, document contents, extracted text, and table cells. Applications are open while endpoint deployment is completed. Apply without attaching a document; invitations begin only after the endpoint is verified:

Apply for free agent beta access

After acceptance, install and configure the separate bridge with the endpoint and key you receive:

python -m pip install "pdf-agent-mcp[cloud] @ https://github.com/wesseltl/pdf-mcp/releases/download/v0.4.0/pdf_agent_mcp-0.4.0-py3-none-any.whl"

{
  "mcpServers": {
    "pdf-cloud": {
      "command": "pdf-agent-cloud-mcp",
      "env": {
        "PDF_MCP_CLOUD_URL": "https://endpoint-provided-with-beta-access.example",
        "PDF_MCP_CLOUD_API_KEY": "key-provided-once"
      }
    }
  }
}

Use only redacted or non-sensitive files. See the free beta terms, privacy notice, and machine-readable beta/free-hosted-beta.json.

What it turns a document into

A PDF or Word table like this:

Item     Qty   Price
Widget    3    12.50
Gadget    1    40.00
Bolt     10     0.25

can be checked against invoice-lines-v1 and returned as canonical records with an explicit decision:

{
  "decision": "accepted",
  "profile": {"id": "invoice-lines-v1", "version": "1.0.0"},
  "records": [
    {
      "values": {"item": "Widget", "quantity": "3", "unit_price": "12.50"},
      "evidence": {
        "item": {"page": 1, "table_index": 0, "row": 1, "column": 0, "bbox": [242.9, 136.0, 287.7, 154.0]}
      }
    }
  ]
}

The tools it gives an agent

MCP setup for Claude Desktop

The fastest way to use this is with an MCP client like Claude Desktop. Three steps:

1. Install it

python -m pip install "pdf-agent-mcp[mcp] @ https://github.com/wesseltl/pdf-mcp/releases/download/v0.4.0/pdf_agent_mcp-0.4.0-py3-none-any.whl"

2. Add it to your client's config

Claude Desktop's config lives here:

• macOS: ~/Library/Application Support/Claude/claude_desktop_config.json
• Windows: %APPDATA%\Claude\claude_desktop_config.json

Add the server:

{
  "mcpServers": {
    "pdf": { "command": "pdf-agent-mcp" }
  }
}

3. Restart Claude Desktop. You'll see a tools icon appear, meaning the server is connected.

That's it. Now ask about any .pdf or .docx on your machine:

You:  Check /Users/me/invoices/2024-001.pdf with invoice-lines-v1.

Agent (calls extract_with_profile):
  decision: accepted
  records: 3
  issues: 0

The agent must route needs_review and rejected results to a person instead of treating them as trusted business data.

Restricting file access: to stop the agent reading anything outside one folder, set PDF_MCP_ALLOWED_DIR. See SECURITY.md.

Use it with other MCP clients

The same server works in any MCP client, only the config differs. Use pdf-agent-mcp as the command.

Cursor — ~/.cursor/mcp.json (global) or .cursor/mcp.json (per project). Same shape as Claude Desktop, and it hot-reloads (no restart):

{ "mcpServers": { "pdf": { "command": "pdf-agent-mcp" } } }

VS Code / GitHub Copilot — .vscode/mcp.json. Note the different key (servers, not mcpServers) and the required type. Tools only run in Copilot Agent mode:

{ "servers": { "pdf": { "type": "stdio", "command": "pdf-agent-mcp" } } }

Windsurf — ~/.codeium/windsurf/mcp_config.json (create it if missing). Same shape as Claude Desktop:

{ "mcpServers": { "pdf": { "command": "pdf-agent-mcp" } } }

Cline — add it from the extension's MCP settings panel in VS Code (command: pdf-agent-mcp).

Profile-checked extraction

The package includes lab-coa-v1 and invoice-lines-v1 as reference profiles:

extract-document-with-profile examples/invoice.pdf invoice-lines-v1 result.xlsx

Exit status is 0 for accepted, 2 for needs_review, and 3 for rejected. XLSX output contains Review, Data, and Evidence sheets. Formula-like document values are escaped in spreadsheet exports rather than executed.

Profiles are ordinary, versioned JSON contracts. See the profile format and machine-readable schema. Results follow the versioned extraction result schema.

Measure a profile

Measure exact field and record accuracy against local, customer-approved expected rows:

evaluate-document-profile evaluations/sample-invoice.json