DocuClaw

🦀 Universal Sovereign Data Infrastructure for Individuals & Teams

English | 简体中文 | Deutsch | Français | Español | Italiano | 日本語

Your invoices. Your contracts. Your letters. Your data. Your rules.

---

🚀 Why DocuClaw?

In a world drowning in SaaS lock-in and cloud surveillance, DocuClaw gives you back control.

Whether you're a freelancer managing personal tax receipts, a startup juggling B2B invoices across borders, or a growing SME facing GoBD compliance audits — DocuClaw is your local-first, privacy-native, AI-powered document brain.

📄 Physical Mail → 📸 Scan → 🤖 AI Extract → 📁 Local Markdown Archive
📧 Email Receipt → 🔗 Webhook → 🤖 AI Extract → 📁 Local Markdown Archive
🧾 API Invoice  → 🔌 Plugin  → 🤖 AI Extract → 📁 Local Markdown Archive

✨ Key Features

Feature	Description
🛡️ 100% Sovereign	All data stays on YOUR machine. Zero cloud dependency. Zero telemetry.
🏢 Multi-Entity	Manage personal docs, company invoices, and team files — all in one install.
🔌 Plugin Architecture	Country-specific parsers (DE, US, CN, ...) snap in like LEGO bricks.
📝 Markdown-Native	Every document becomes a searchable .md file with structured YAML frontmatter.
🤖 AI-Powered Extraction	Multimodal LLM extracts structured data from scans, photos, and emails.
✅ Compliance-Ready	Designed with GoBD (Germany), GDPR, and audit-trail principles baked in.
🔍 RAG-Ready	Full-text originals preserved for retrieval-augmented generation workflows.

---

🏗️ Architecture

DocuClaw follows a Core Engine + Pluggable Parsers architecture, designed for enterprise-grade extensibility:

┌─────────────────────────────────────────────┐
│                   CLI / API                  │
├─────────────────────────────────────────────┤
│               Core Engine                    │
│  ┌──────────┐  ┌──────────┐  ┌───────────┐ │
│  │  Schema   │  │ Storage  │  │  Registry │ │
│  │(Pydantic) │  │  Layer   │  │  (Plugin) │ │
│  └──────────┘  └──────────┘  └───────────┘ │
├─────────────────────────────────────────────┤
│             Parser Plugins                   │
│  ┌────────┐  ┌────────┐  ┌──────────────┐  │
│  │ DE 🇩🇪  │  │ US 🇺🇸  │  │ Custom ...  │  │
│  │Invoice │  │Invoice │  │  Your Parser │  │
│  └────────┘  └────────┘  └──────────────┘  │
├─────────────────────────────────────────────┤
│        Input Adapters (Future)               │
│  📷 Scanner │ 📧 Email │ 🔗 Webhook │ 🔌 API │
└─────────────────────────────────────────────┘

The Data Contract

Every document, whether a €10K enterprise invoice or a personal electricity bill, is normalized into a universal Markdown schema with structured YAML frontmatter:

---
id: doc_20260215_a1b2c3d4
entity_id: "org_acme_01"
entity_type: "company"
source_type: physical_mail
country: DE
document_type: b2b_invoice
date_received: "2026-02-15"
sender_name: "AWS EMEA SARL"
amount_total: 125.50
currency: EUR
status: pending
tags: [IT_Infrastructure, Q1_Expense]
---
### Raw Content
[Full OCR / email body preserved for compliance & RAG]

### AI Summary
This is the February AWS bill containing €20.04 input VAT...

---

⚡ Quick Start

Installation

# Clone the repository
git clone https://github.com/openclaw-ai/docuclaw.git
cd docuclaw

# Install dependencies
pip install -e .

Usage

# Process a German invoice scan
docuclaw process \
  --entity-id "org_mycompany_01" \
  --entity-type company \
  --country DE \
  --source-type physical_mail \
  --input ./scans/invoice_aws_feb.png

# Output: ./docuclaw_data/org_mycompany_01/2026/02/doc_20260215_xxxx.md

Python API

from docuclaw.schema import DocuClawDocument, EntityType, SourceType
from docuclaw.core.storage import MarkdownStorageEngine
from docuclaw.parsers.de_invoice_parser import DEInvoiceParser

# Initialize
storage = MarkdownStorageEngine(base_path="./docuclaw_data")
parser = DEInvoiceParser()

# Parse a document
doc = parser.parse(
    file_path="./scans/invoice.png",
    entity_id="org_mycompany_01",
    entity_type=EntityType.COMPANY,
)

# Persist as structured Markdown
output_path = storage.save(doc)
print(f"📄 Saved: {output_path}")

---

🧩 Writing Custom Parsers

Extend DocuClaw for any country or document type:

from docuclaw.parsers.base import BaseDocumentParser
from docuclaw.schema import DocuClawDocument

class USReceiptParser(BaseDocumentParser):
    """Parser for US retail receipts."""

    @property
    def supported_countries(self) -> list[str]:
        return ["US"]

    @property
    def supported_document_types(self) -> list[str]:
        return ["receipt", "b2c_invoice"]

    def parse(self, file_path, entity_id, entity_type, **kwargs):
        # Your extraction logic here
        ...

---

🇪🇺 GDPR & EU Compliance

DocuClaw is architected from the ground up with EU General Data Protection Regulation (GDPR) compliance as a core design principle — not an afterthought.

Why DocuClaw is inherently GDPR-friendly

GDPR Requirement	How DocuClaw Fulfills It
Art. 5(1)(c) — Data Minimization	Only extracts and stores the structured fields you explicitly define. Zero telemetry, zero usage analytics, zero behavioral tracking.
Art. 5(1)(f) — Integrity & Confidentiality	All data processing happens locally on your machine. No network transmission = no interception risk.
Art. 5(2) — Accountability	Built-in audit logging with hash-chain integrity verification provides tamper-evident processing records.
Art. 17 — Right to Erasure	Data is stored as plain Markdown files on your local filesystem. Deletion is as simple as removing a file — no vendor tickets, no retention policies to fight.
Art. 25 — Data Protection by Design	Privacy-first architecture: local-only processing is not a feature toggle, it's the only mode. No cloud fallback exists.
Art. 44–49 — International Transfers	No data ever leaves your machine. No third-party servers, no sub-processors, no cross-border transfers. Full compliance by architectural design.

GoBD Compliance (Germany) 🇩🇪

For users in Germany, DocuClaw additionally supports GoBD (Grundsätze zur ordnungsmäßigen Führung und Aufbewahrung von Büchern, Aufzeichnungen und Unterlagen in elektronischer Form):

• Immutability: Hash-chain audit trail ensures archived documents cannot be silently altered
• Traceability: Every processing step is logged with timestamps and checksums
• Retention: Local storage with structured dating supports the 10-year retention requirement
• Accessibility: Markdown-native format ensures documents remain human-readable without proprietary software

No Cloud? No Problem.

Unlike SaaS alternatives, DocuClaw never requires you to:

• Sign a Data Processing Agreement (DPA) with a third party
• Conduct a Data Protection Impact Assessment (DPIA) for cloud transfers
• Maintain Records of Processing Activities (RoPA) for external processors
• Worry about the adequacy decisions of third-country data flows

Your data stays on your machine. Period. That's the simplest — and most secure — compliance strategy.

---

🤖 AI-Powered Output — From Archive to Action

DocuClaw doesn't just archive your documents — it turns them into an actionable knowledge base. Through AI agent integration (via OpenClaw or any compatible LLM agent), your structured Markdown data becomes a living system that can answer questions, automate workflows, and feed directly into the tools you already use.

💬 Ask Your Documents

Talk to your document archive like you'd talk to a colleague:

You:    "How much did I spend on AWS in Q4 2025?"
Agent:  "Based on 3 invoices archived in DocuClaw, your total AWS spend
         in Q4 2025 was €387.42 (Oct: €125.50, Nov: €131.88, Dec: €130.04)."

You:    "When does my office lease expire?"
Agent:  "Your lease contract (doc_20240301_lease) shows an expiration date
         of March 31, 2027, with a 3-month notice period starting Jan 1, 2027."

Because documents are stored as structured Markdown with YAML frontmatter, any LLM or RAG pipeline can instantly query, filter, and reason over your entire archive — without sending data to the cloud.

📅 Calendar, Reminders & To-Do Lists

Auto-extract actionable dates and deadlines from your documents:

Source Document	Auto-Generated Action
Invoice with due date	📅 Calendar event: "Pay AWS invoice €125.50" on Feb 28
Contract with renewal clause	⏰ Reminder: "Lease renewal notice deadline" 90 days before expiry
Insurance policy	✅ To-Do: "Review and renew car insurance by April 15"
Quarterly VAT summary	📋 Task: "Submit Q1 VAT return — total: €2,340.00"

Push these directly to Apple Calendar, Google Calendar, Todoist, Things, Notion, or any task manager via standard APIs.

🧾 Tax Filing & Financial Reports

Generate tax-ready outputs directly from your archived documents:

• Expense Summaries: Categorized by type, vendor, tax rate, and period
• VAT/GST Reports: Pre-calculated input tax, output tax, and net amounts
• Annual Tax Packages: Formatted for your accountant or tax advisor
• ELSTER-ready data (Germany): Export structured data compatible with German tax filing
• Custom Financial Reports: Monthly P&L, quarterly cash flow, annual overviews

🔗 Third-Party System Integration

DocuClaw generates and submits data in the exact format required by external systems:

System Type	Examples	What DocuClaw Generates
Accounting Software	DATEV, Xero, QuickBooks, Lexware	Booking entries, invoice records, expense imports
ERP Systems	SAP, Odoo, ERPNext	Structured purchase orders, vendor records
Government Portals	ELSTER (DE), HMRC (UK), IRS (US)	Tax declarations, compliance reports
Banking Platforms	SWIFT, SEPA	Payment instructions, reconciliation data
CRM Systems	Salesforce, HubSpot	Contract metadata, vendor information

All data stays local until you decide to export or submit. DocuClaw generates the output — you control when and where it goes.

---

🗺️ Roadmap

Our vision for DocuClaw is to become the ultimate Sovereign Data Hub for your personal and business documents. Here is what we are building next:

Phase 1: Core Engine & Expanded Parsers (Current)

• Milestone 1: Core schema, storage engine, parser framework, CLI skeleton
• Milestone 2: Email ingestion adapter (IMAP/POP3)
• Milestone 3: Real multimodal LLM integration (Ollama, OpenAI Vision)
• Milestone 4: Web UI dashboard (local-only, no cloud)
• Milestone 5: GoBD-compliant audit trail with hash chains
• Milestone 6: Multi-entity permission model & team collaboration
• Milestone 7: Webhook & API ingestion endpoints
• Multi-Country Parser Ecosystem: Specialized extraction logic for highly-bureaucratic regions:
  • 🇩🇪 Germany (e.g., Steuerbescheid, GoBD compliance considerations)
  • 🇫🇷 France (e.g., CAF, URSSAF, CPAM documents)
  • 🇮🇹 Italy (e.g., Raccomandata, Fattura Elettronica)
  • 🇪🇸 Spain, 🇺🇸 United States (Medical bills, IRS notices), 🇯🇵 Japan (Hanko documents).
• Advanced OCR Pipeline: Better layout recognition for complex tabular data (e.g., invoices).

Phase 2: Omnichannel Ingestion (Meeting data where it lives)

• Seamless Email Integration:
  • One-click OAuth for Gmail, Outlook, and iCloud.
  • Standard IMAP support for privacy-focused providers (ProtonMail) and regional giants (GMX, Web.de).
• Native OS & Media Sync:
  • Apple Photos Integration: Auto-import receipts and documents directly from your macOS/iOS photo library.
  • Local Watchdogs: Auto-process files dropped into specific local folders (perfect for network scanners).

Phase 3: Pluggable AI Engines (Bring Your Own Brain)

• Cloud AI Integration: Easy API key setup for OpenAI (GPT-4o), Anthropic (Claude), and Google (Gemini).
• Local-First LLMs: Out-of-the-box support for local inference engines like Ollama and LM Studio. Process highly sensitive documents (like medical records) completely offline.

Phase 4: Automated Export Workflows (The Data Router)

• Calendar & Tasks: Automatically push deadlines (e.g., invoice due dates) to Google Calendar, Apple iCal, or Todoist.
• Tax & Accounting Sync: Export parsed financial data to tools like DATEV, Lexoffice, SevDesk (EU), or QuickBooks (US).
• Knowledge Base Integrations: Seamlessly sync structured Markdown data into Obsidian or Notion for your Second Brain.

---

📦 Part of the OpenClaw Ecosystem

DocuClaw is a core component of openclaw.ai — an open-source ecosystem for sovereign AI-powered productivity tools.

Project	Description
DocuClaw	Sovereign document intelligence & archival
DeepReader	AI-powered web content ingestion
ClawHub	Plugin marketplace & community hub

---

🤝 Contributing

We welcome contributions! Whether it's a new country parser, a bug fix, or documentation improvements.

# Development setup
git clone https://github.com/openclaw-ai/docuclaw.git
cd docuclaw
pip install -e ".[dev]"

# Run tests
pytest

# Run linters
ruff check .
mypy docuclaw/

See CONTRIBUTING.md for detailed guidelines.

---

📄 License

Licensed under the MIT License. Use it freely. Own your data.

---

Built with 🦀 by the OpenClaw community
"Your data should work for you, not against you."