PY
Python
For Python 3.9 and later. Pass a file path, an open file or a public link, and catch typed errors such as InsufficientCreditsError.
$ pip install parserbee For developers · Python, Node.js and MCP
Document parsing API for invoices, receipts and purchase orders
Send a PDF, scan or phone photo to one endpoint and get back clean JSON with the fields you chose. Call our invoice OCR API from your code, or let your AI assistant call it through the MCP server.
25 free pages No credit card 1 page credit per page MIT-licensed libraries
pip install parserbeenpm install parserbeenpx -y parserbee-mcp
Request
curl -X POST https://app.parserbee.com/api/v1/extract \
-H "x-api-key: $PARSERBEE_API_KEY" \
-F "template_id=your_template_id" \
-F "file=@invoice.pdf"Response 200 OK
{
"success": true,
"data": {
"invoice_number": "INV-2026-001",
"vendor_name": "Acme Corp",
"total_amount": 1500.00
},
"usage": { "pages_processed": 1, "credits_charged": 1 }
}Get your API key in a minute. 25 free pages, no credit card.
Libraries and SDKs
Official libraries and tools
Open source under the MIT license. Install one, add your API key, and make your first call in a few minutes.
JS
Node.js and TypeScript
For Node.js 18 and later. ES modules and CommonJS, with TypeScript types you can shape to your own Parser Template fields.
$ npm install parserbeeMCP
MCP server
Lets Claude Desktop, Claude Code, Cursor, VS Code and other MCP clients extract documents with your Parser Templates.
$ npx -y parserbee-mcpAPI
REST API
One HTTPS endpoint that any language can call. Send the key in the x-api-key header and a file or a file URL in the body.
POST https://app.parserbee.com/api/v1/extractPM
Postman collection
Every endpoint, ready to send. The collection saves your template ID and request ID between calls, so you can click through a full run.
OAS
OpenAPI 3.1 spec
The full API description in one JSON file. Import it into your API client or generate a client for another language.
Quick start
Your first extraction in three steps
- 1
Create a free account
Sign up in under a minute. You get 25 free pages and no credit card is needed.
- 2
Make a Parser Template
Upload one sample document and the AI suggests the fields, or pick a ready-made one from the template library. Copy the template ID from the template page.
- 3
Copy your API key
Open Settings in the dashboard and copy your key. It starts with
pk_. Save it asPARSERBEE_API_KEY.
curl -X POST https://app.parserbee.com/api/v1/extract \
-H "x-api-key: $PARSERBEE_API_KEY" \
-F "template_id=your_template_id" \
-F "file=@invoice.pdf"from parserbee import ParserBee
client = ParserBee() # reads PARSERBEE_API_KEY
result = client.extract("your_template_id", "invoice.pdf")
print(result["data"]["invoice_number"])import { ParserBee } from "parserbee";
const client = new ParserBee(); // reads PARSERBEE_API_KEY
const result = await client.extract("your_template_id", { file: "invoice.pdf" });
console.log(result.data.invoice_number);curl -X POST https://app.parserbee.com/api/v1/extract \
-H "x-api-key: $PARSERBEE_API_KEY" \
-H "Content-Type: application/json" \
-d '{"template_id": "your_template_id", "file_url": "https://example.com/invoice.pdf"}'{
"success": true,
"request_id": "req_abc123",
"data": {
"invoice_number": "INV-2026-001",
"invoice_date": "2026-10-02",
"vendor_name": "Acme Corp",
"total_amount": 1500.00,
"line_items": [
{
"description": "Consulting",
"quantity": 10,
"amount": 1500.00
}
]
},
"credits_remaining": 449,
"processing_time_ms": 2400,
"usage": {
"pages_processed": 1,
"credits_charged": 1,
"doc_size_bytes": 204800
}
}The keys inside data are the field names of your Parser Template. Change the template and the JSON changes with it; your code does not need a new parser for each supplier's layout.
MCP server
Extract documents from your AI assistant
The ParserBee MCP server works in Claude Desktop, Claude Code, Cursor, VS Code and any other Model Context Protocol client, and gives the assistant six tools. Ask in plain words, for example: “Read every receipt in this folder and give me a table of merchant, date and total.”
extract_documentExtract a local file or a public link with a Parser Templatelist_templatesList your Parser Templates and their fieldsget_templateRead one Parser Templateget_creditsCheck your page credit balancelist_recent_extractionsThe last 10 extraction requestsget_extractionOne extraction, with its data
Claude Desktop: Settings > Developer > Edit Config. Claude Code: --scope user makes ParserBee available in every folder, not only the current one; start a new session after you add it. Cursor: ~/.cursor/mcp.json. VS Code: .vscode/mcp.json. You need Node.js 18 or later; more setup notes are in the parserbee-mcp README.
{
"mcpServers": {
"parserbee": {
"command": "npx",
"args": ["-y", "parserbee-mcp"],
"env": { "PARSERBEE_API_KEY": "pk_your_api_key" }
}
}
}claude mcp add parserbee --scope user -e PARSERBEE_API_KEY=pk_your_api_key -- npx -y parserbee-mcp{
"mcpServers": {
"parserbee": {
"command": "npx",
"args": ["-y", "parserbee-mcp"],
"env": { "PARSERBEE_API_KEY": "pk_your_api_key" }
}
}
}{
"servers": {
"parserbee": {
"command": "npx",
"args": ["-y", "parserbee-mcp"],
"env": { "PARSERBEE_API_KEY": "pk_your_api_key" }
}
}
}Reference
The API at a glance
Base URL https://app.parserbee.com. Every request needs the x-api-key header. The full reference, with request and response examples, is in the API docs.
| Method | Endpoint | What it does |
|---|---|---|
| POST | /api/v1/extract | Extract data from an uploaded file or a public file URL with a Parser Template |
| GET | /api/v1/templates | List the Parser Templates in your account |
| GET | /api/v1/templates/{id} | One Parser Template with all of its fields |
| GET | /api/v1/credits | Page credit balance, plan and subscription status |
| GET | /api/v1/activity | Your most recent extraction requests |
| GET | /api/v1/activity/{id} | One extraction request, with its extracted data |
Billing
1 page credit per page processed, minimum 1 per document. Each response shows what the call cost.
Files
PDF, PNG, JPG, WebP, HEIC and common office files, up to 50 MB. Scans and phone photos work.
Input
multipart/form-data with template_id and file, or JSON with template_id and file_url.
Uptime
Live status and incident history are on the status page.
Errors
| Status | Code | Meaning |
|---|---|---|
| 400 | MISSING_FILE, MISSING_TEMPLATE_ID, INVALID_FILE | A field is missing, or the file type or size is not accepted |
| 401 | INVALID_API_KEY | The x-api-key header is missing or wrong |
| 402 | INSUFFICIENT_CREDITS | No page credits left; the body also holds credits_remaining |
| 403 | SUBSCRIPTION_INACTIVE | The account has no active plan or trial |
| 404 | TEMPLATE_NOT_FOUND | The template ID does not exist in your account |
| 422 | PDF_TOO_LONG | The PDF has more pages than your plan allows in one file |
| 500 | PROCESSING_FAILED | The document could not be processed |
Errors return JSON with error (a message) and code. The Python and Node.js libraries turn each one into its own error class.
Your first 25 pages are on us
Create an account, build a Parser Template from one sample document, and call the API in minutes. No credit card, and the free pages never expire.
FAQ
Developer questions
Is there a free plan for the ParserBee API?
Every new account gets 25 free pages in total, with full API access and no credit card. The free pages do not expire. Paid plans start at $29 per month; see pricing.
How is API usage billed?
1 page credit per page processed, with a minimum of 1 per document. A photo or a one-page PDF costs 1 credit and a 10-page PDF costs 10. Each response shows pages_processed and credits_charged, and GET /api/v1/credits returns your balance. The API, the dashboard and the integrations all use the same page credits.
Which file types does the API accept?
PDF, PNG, JPG, WebP and HEIC, plus Word (.doc), PowerPoint, Excel (.xlsx), CSV, TXT, RTF, ODT, HTML and EML files, up to 50 MB each. Scanned PDFs and phone photos work. DOCX files are not accepted: save them as PDF first.
Do I have to write a JSON schema?
No. A Parser Template is the schema. Build one from a single sample document with the AI template builder, edit its fields in the dashboard, or start from the template library. The API returns your template's field names, so the JSON has the same shape for every supplier's layout.
Is the API synchronous?
Yes. One request returns the extracted data in the same response, so there is no job ID to poll and no webhook to set up.
How do I process many documents?
Send one request per document from a loop or a queue, and check credits_remaining in each response. For a one-off batch with no code, Data Lab in the dashboard takes a set of files and exports one Excel, CSV or Google Sheets file.
Which programming languages are supported?
There are official libraries for Python and Node.js (with TypeScript types). Any other language can call the REST API with a normal HTTP client, and the OpenAPI 3.1 spec lets you generate a client.
Does ParserBee work with Zapier, Make and n8n?
Yes. ParserBee has a native Zapier integration (see the Zapier guide). Make and n8n call the API with their HTTP modules, and ready-made n8n workflows are on GitHub.
Can an AI assistant use ParserBee?
Yes, through the ParserBee MCP server. Add it to Claude Desktop, Claude Code, Cursor, VS Code or any other app that supports the Model Context Protocol, and the assistant can extract documents with your Parser Templates, list your templates and check your page credits.
Stop retyping documents. Start calling one endpoint.
Invoices, receipts, purchase orders, bank statements and more, as JSON your code can trust.