Docpick MCP server
✓ VerifiedExtracts structured JSON from documents using local OCR engines and LLMs with customizable Pydantic schemas.
Also known as QuartzUnit/docpick
- Stars
- 1
- Tools
- —
- Downloads, 30 days
- 2,671
- Latest version
- 0.1.2
- Last commit
- Apr 15, 2026
- License
- Apache-2.0
What is Docpick MCP?
Docpick extracts structured JSON from documents using local OCR engines and LLMs with customizable Pydantic schemas.
Summary generated from the project's README
About Docpick MCP server
Converting physical documents into structured JSON data locally is what Docpick enables for developers building private data pipelines. It functions as a schema-driven extraction system that combines local OCR engines like PaddleOCR, EasyOCR, or GOT-OCR2.0 with local LLM endpoints such as Ollama or vLLM. This setup ensures that sensitive information from invoices, receipts, or bank statements never leaves your machine, as the entire process runs on your own CPU or GPU.
Users can choose from eight built-in schemas for common documents or define custom Pydantic models to enforce specific data structures. The system includes built-in validation tools like checkdigit verification for IBAN or Korean business numbers and cross-field rules to ensure mathematical consistency. Whether processing a single file via the Python API or running batch jobs across entire directories with parallel workers, the pipeline manages OCR fallback and LLM retries to maintain high extraction accuracy.
What you can do with it
- 1Extracting invoice dataConvert commercial invoices into structured JSON while verifying tax IDs and line item sums.
- 2Validating identity documentsExtract passport information and verify it against ICAO 9303 standards and MRZ zones.
- 3Automating batch document processingRun parallel extraction tasks on a folder full of documents using a specific schema.
- 4Ensuring cross-document consistencyVerify that data remains consistent across an invoice, a bill of lading, and a packing list.
Written from the project's README. Every use case is checked against a quote from it.
How to install Docpick MCP
Formats verified 10 Nov 2026Add to CursorAdd to GooseInstall Docpick MCP in Claude Code
$ claude mcp add docpick -- uvx docpickInstall Docpick MCP in Cursor
~/.cursor/mcp.json
{
"mcpServers": {
"docpick": {
"command": "uvx",
"args": [
"docpick"
]
}
}
}Install Docpick MCP in VS Code
.vscode/mcp.json
{
"servers": {
"docpick": {
"type": "stdio",
"command": "uvx",
"args": [
"docpick"
]
}
}
}Install Docpick MCP in VS Code CLI
$ code --add-mcp '{"name":"docpick","type":"stdio","command":"uvx","args":["docpick"]}'Install Docpick MCP in Claude Desktop
claude_desktop_config.json
{
"mcpServers": {
"docpick": {
"command": "uvx",
"args": [
"docpick"
]
}
}
}Install Docpick MCP in Codex
$ codex mcp add docpick -- uvx docpickInstall Docpick MCP in Codex config.toml
~/.codex/config.toml
[mcp_servers.docpick] command = "uvx" args = ["docpick"]
Install Docpick MCP in Gemini CLI
$ gemini mcp add -s user docpick uvx -- docpickInstall Docpick MCP in Gemini settings.json
~/.gemini/settings.json
{
"mcpServers": {
"docpick": {
"command": "uvx",
"args": [
"docpick"
]
}
}
}Automated checks
Ran Nov 15, 2026Full report →Passed our automated checks. That is not a security audit: read the code and give the server the narrowest access it needs. How checks work
- Source repository linked and reachablequartzunit/docpick is public and reachablegithub.com/quartzunit/docpick ↗Pass
- Package names this repository as its sourcedocpick names quartzunit/docpick as sourcepypi.org/project/docpick/ ↗Pass
- Build provenance (Sigstore attestation)No build attestation published for this releaseNot checked
- Open source license (OSI)Apache-2.0spdx.org/licenses/Apache-2.0.html ↗Pass
- Commit or release in the last yearLast commit or release 2026-04-15github.com/quartzunit/docpick ↗Pass
- OpenSSF ScorecardNot checked
- Package name not close to a popular packageNo popular package one edit away from docpickPass
- No critical advisory in the package or its dependencies22 resolved dependencies checked; no critical advisoryosv.dev ↗Pass
- No hidden instructions to the model0 tool descriptions and the README read; no override phrases, credential paths or hidden charactersPass
- Destructive tools are labelledNot checked
- Remote endpoint answers the MCP handshakeNot checked
README
View on GitHubDocpick
Document in, Structured JSON out. Locally. With your schema.
docpick is a lightweight, schema-driven document extraction pipeline that combines local OCR engines with local LLMs to extract structured JSON from any document — invoices, receipts, bills of lading, tax forms, and more.
- Zero cloud dependency — runs entirely on your machine (CPU or GPU)
- Custom schemas — define your own Pydantic models or use 8 built-in document schemas
- Validation built-in — checkdigit verification, cross-field rules, cross-document consistency
- Apache 2.0 — no GPL/AGPL dependencies
Install
pip install docpick # core (LLM extraction only)
pip install docpick[paddle] # + PaddleOCR (recommended)
pip install docpick[easyocr] # + EasyOCR (Korean-optimized)
pip install docpick[got] # + GOT-OCR2.0 (GPU, vision-language)
pip install docpick[all] # all OCR backends
Requirements: Python 3.11+ / LLM endpoint (vLLM, Ollama, or OpenAI-compatible)
Quick Start
Python API
from docpick import DocpickPipeline
from docpick.schemas import InvoiceSchema
pipeline = DocpickPipeline()
result = pipeline.extract("invoice.pdf", schema=InvoiceSchema)
print(result.data) # Structured dict matching schema
print(result.validation) # Validation errors/warnings
print(result.confidence) # Per-field confidence scores
CLI
# Extract structured data
docpick extract invoice.pdf --schema invoice --output result.json
# OCR only (no LLM)
docpick ocr document.png --lang ko,en
# Validate extracted JSON
docpick validate result.json --schema invoice
# Batch process a directory
docpick batch ./documents/ --schema invoice --output ./results/ --concurrency 4
# List available schemas
docpick schemas list
# Show schema details
docpick schemas show invoice
Built-in Schemas
| Schema | Document Type | Key Validations |
|---|---|---|
invoice |
Commercial invoices | Line item sums, tax ID checkdigit, date order |
receipt |
Retail/restaurant receipts | Total = subtotal + tax + tip |
bill_of_lading |
Ocean/air B/L | Container weight sums, ISO 6346, HS code format |
purchase_order |
Purchase orders | PO total = line items, delivery date order |
kr_tax_invoice |
Korean e-tax invoice (세금계산서) | Business number checkdigit (x2), supply/tax/total sums |
bank_statement |
Bank statements | IBAN mod97, period date order |
id_document |
Passport/ID (ICAO 9303) | MRZ, ISO 3166 country codes, date ranges |
certificate_of_origin |
Certificate of Origin | ISO 3166 alpha-2 country codes |
Custom Schemas
Define your own schema with Pydantic:
from pydantic import BaseModel
from docpick import DocpickPipeline
from docpick.validation.rules import SumEqualsRule, RequiredFieldRule
class MyDocument(BaseModel):
"""Custom document schema."""
company_name: str | None = None
total_amount: float | None = None
tax_amount: float | None = None
net_amount: float | None = None
items: list[dict] | None = None
class ValidationRules:
rules = [
RequiredFieldRule("company_name"),
SumEqualsRule(["net_amount", "tax_amount"], "total_amount"),
]
pipeline = DocpickPipeline()
result = pipeline.extract("my_document.pdf", schema=MyDocument)
Or use a JSON Schema file:
docpick extract document.pdf --schema my_schema.json
Validation
Check Digit Algorithms
| Algorithm | Use Case |
|---|---|
kr_business_number |
Korean business registration number (10 digits) |
luhn |
Credit card numbers |
iso_6346 |
Shipping container numbers |
iban_mod97 |
International bank account numbers |
awb_mod7 |
Air waybill numbers |
mrz |
Machine Readable Zone (passport/ID) |
Cross-Field Rules
| Rule | Description |
|---|---|
SumEqualsRule |
Sum of fields equals target (with tolerance) |
DateBeforeRule |
Date A must precede Date B |
RequiredFieldRule |
Field must be non-null and non-empty |
FieldEqualsRule |
Two fields must be equal |
RangeRule |
Numeric field within min/max bounds |
RegexRule |
Field matches regex pattern |
Cross-Document Validation
Validate consistency across related documents (e.g., Invoice + B/L + Packing List):
from docpick.validation.cross_document import create_trade_document_validator
validator = create_trade_document_validator()
result = validator.validate({
"invoice": invoice_data,
"bl": bl_data,
"packing_list": packing_list_data,
"certificate": certificate_data,
})
print(result.is_valid)
OCR Engines
| Engine | Type | GPU | Languages | Best For |
|---|---|---|---|---|
| PaddleOCR | Traditional OCR | Optional | 111 | General documents (default) |
| EasyOCR | Traditional OCR | Optional | 80+ | Korean text |
| GOT-OCR2.0 | Vision-Language | Required | Multi | Complex layouts |
| VLM | Vision-Language | Required | Multi | Direct image → JSON |
2-Tier Auto Engine
The default auto engine uses confidence-based fallback:
- Tier 1 (CPU): PaddleOCR → EasyOCR
- Tier 2 (GPU): GOT-OCR2.0 → VLM
If Tier 1 average confidence falls below threshold (default 0.7), automatically escalates to Tier 2.
LLM Providers
| Provider | Endpoint | Default Model |
|---|---|---|
| vLLM | http://localhost:8000/v1 |
Qwen/Qwen3.5-32B-AWQ |
| Ollama | http://localhost:11434 |
qwen3.5:7b |
Configure via CLI or YAML:
docpick config set llm.provider ollama
docpick config set llm.base_url http://localhost:11434
docpick config set llm.model qwen3.5:7b
Error Handling
The pipeline is designed to be resilient:
- OCR failure → automatic fallback to next available engine
- LLM JSON parse failure → automatic retry with correction prompt (up to 1 retry)
- Partial results → returns whatever was extracted, with errors logged in
result.errors - Document load failure → returns empty result with error message
result = pipeline.extract("damaged.pdf", schema=InvoiceSchema)
if result.errors:
print("Pipeline warnings:", result.errors)
if result.data:
print("Partial extraction:", result.data)
Batch Processing
Process entire directories with parallel workers:
from docpick.batch import BatchProcessor
from docpick.schemas import InvoiceSchema
processor = BatchProcessor(concurrency=4)
result = processor.process_directory(
"./invoices/",
schema=InvoiceSchema,
recursive=True,
)
print(f"Processed {result.succeeded}/{result.total} files")
for path, extraction in result.results.items():
print(f"{path}: {extraction.data.get('total_amount')}")
Architecture
flowchart TD
A["📄 Document\n(PDF / Image)"] --> B["DocumentLoader\n(pypdfium2)"]
B --> C["Tier 1: OCR\n(PaddleOCR / EasyOCR)\nCPU"]
C --> D{"Confidence\n≥ threshold?"}
D -->|"yes"| F["LLM Extractor\n(vLLM / Ollama)\nSchema prompt"]
D -->|"no"| E["Tier 2: VLM\n(GOT / VLM)\nGPU"]
E --> F
F --> G["Pydantic Validation"]
G --> H["✅ ExtractionResult"]
License
Apache 2.0 — all dependencies are Apache 2.0 or MIT licensed.
Part of the QuartzUnit ecosystem — composable Python libraries for data collection, extraction, search, and AI agent safety.
Docpick MCP questions
How do I add Docpick MCP to Claude Code?
Run claude mcp add docpick -- uvx docpick in your terminal. Add --scope user to make it available in every project.
How do I add Docpick MCP to Cursor?
Open ~/.cursor/mcp.json and add the server under mcpServers, using the Cursor config shown in the install section above.
Does Docpick require cloud services?
No, it runs entirely on your machine with zero cloud dependency.
Can I use my own data structures for extraction?
Yes, you can define your own schemas using Pydantic models or JSON Schema files.
How does the system handle low-confidence OCR results?
It uses a 2-tier auto engine that automatically escalates to more advanced vision-language models if confidence falls below a threshold.