Connect Lido's MCP server to Claude or Cursor and OCR any invoice — scanned, photographed, or digital. Extract every field without templates or training.
claude mcp add lido -- npx -y @lido-app/mcp-serverThe Lido MCP server is the fastest way to extract structured data from invoices using an AI coding assistant. Install with one command, describe what you need in plain English, and get organized rows and columns back — no templates, no configuration. 50 free pages included.
Three steps from install to extracted data
Run a single command in your terminal. Works with Claude Code, Claude Desktop, Cursor, and Windsurf.
Tell Claude or Cursor to extract data from your invoices. Describe what you need in plain English.
Lido reads the document layout, extracts the fields and tables, and returns structured data you can use immediately.
Extract data from invoices without leaving your editor
Lido reads document layouts the way a human would — interpreting headers, tables, and field relationships by context. No zones, no rules, no training.
Run a single npm command and the MCP server connects to your AI assistant. No API keys to configure, no SDKs to install, no dashboard to set up.
Get clean rows and columns from any document. Line items, totals, dates, vendor names — extracted and organized into data you can pipe into your code.
OCR built in. Photographed receipts, scanned invoices, and image-only PDFs all get the same accurate extraction as digital documents.
Process hundreds of invoices in a single conversation. Point your AI assistant at a folder and let Lido extract them all.
SOC 2 Type 2 certified and HIPAA compliant. Documents encrypted in transit and at rest, automatically deleted within 24 hours.
Tell your AI assistant what to extract in plain English
"OCR this scanned invoice and extract vendor, invoice number, date, line items, and total."
"This is a photo of an invoice I took on my phone. Extract all the line items and the grand total."
"OCR this faded invoice scan — I need the vendor name, PO number, and payment terms."
"Process all scanned invoices in /scans/ and output vendor, amount, and due date for each."
Last updated: June 2026
Model Context Protocol (MCP) is an open standard created by Anthropic that allows AI assistants to connect with external tools and data sources directly. Rather than transferring data manually between your browser and your AI, MCP servers give the AI immediate access to specialized functionality.
The Lido MCP server grants Claude, Cursor, and other MCP-compatible tools the capability to extract structured data from documents. Upon installation, your AI assistant receives four new tools: authenticating with your Lido account, extracting data from files, accessing extraction tips for different document types, and monitoring your usage.
This means you can provide your AI with a stack of invoices and ask it to extract the data you need — all within the same conversation where you are writing code, analyzing data, or building automation.
Standard document extraction tools require switching to a web dashboard, uploading files, building extraction templates, and then transferring the results back to your workspace. MCP removes that context switching.
Using an MCP server, your AI assistant runs the complete process: it reads your document, calls the extraction API, and delivers structured data — all in the same terminal or editor where you are already working. You describe what you need in English, and the AI handles the extraction logic.
Lido’s approach is especially practical because it operates without templates. Most extraction tools expect you to define zones, rules, or training sets for each document layout. Lido interprets the visual structure of each document individually, the way a person would scan a page and understand which numbers are totals, which text is a vendor name, and which rows form a table.
For step-by-step setup instructions, see how to extract document data with Claude using Lido MCP. For background on the protocol itself, read what is MCP (Model Context Protocol).
Related: compare MCP servers for invoice processing via MCP, OCR via MCP, accounting document processing via MCP.
Any tool that supports Model Context Protocol. This includes Claude Code (terminal), Claude Desktop (app), Cursor, Windsurf, and other MCP-compatible clients. The server installs the same way in all of them.
No. Lido reads document layouts automatically — it interprets headers, tables, and field relationships by visual context rather than predefined zones or rules. This means it works on any invoice layout from any source without configuration.
95-99% accuracy on digital documents and 90-98% on scanned documents. Lido uses AI vision models that read the visual structure of each page, so accuracy is consistent across different layouts and formats.
50 pages free, no credit card required. After that, plans start at $29/month for 100 pages. Enterprise pricing is available for high-volume processing.
Lido is SOC 2 Type 2 certified and HIPAA compliant. Documents are encrypted with AES-256 at rest and TLS 1.2+ in transit. All uploaded documents are automatically deleted within 24 hours and are never used to train AI models.
Yes. The MCP server includes OCR for scanned documents and images. Photographed receipts, scanned invoices, and image-only PDFs all get the same structured extraction as digital documents.
One command. 50 free pages. No credit card, no templates, no configuration.
claude mcp add lido -- npx -y @lido-app/mcp-server