Qleric Contract Extractor (Open Source)

Contract extraction engine - extract 60+ fields from PDF contracts with Claude AI.

Disclaimer

This repository contains a curated, handpicked subset of the full Qleric Contract Extractor solution, made available as open source to comply with licensing requirements (e.g., GPL dependencies like PyMuPDF used in the SaaS version). It is manually synced with the SaaS application approximately once a month to reflect key updates. For access to the complete, production-ready solution, please refer to the main Qleric platform.

Features

• 60-field extraction taxonomy - dates, parties, financial terms, liability, IP, compliance, and more
• 3-tier extraction - Essential (9 fields), Professional (18 fields), Enterprise (25 fields)
• Custom field selection - mix-match any fields from the 60-field bank (max 25 per extraction)
• Smart chunking - handles contracts >50 pages with intelligent section boundary detection
• Table extraction - payment schedules, fee tables with automatic normalization
• Grounding verification - validates extracted values against source text

Installation

pip install -r requirements.txt
cp example.env .env
# Edit .env and add your ANTHROPIC_API_KEY

Security

1. Copy example.env to .env and add your actual API key.
2. Never commit your .env file — it contains sensitive credentials.
3. Ensure .env is listed in your .gitignore.

Quick Start (Python)

from contract_extractor import ContractExtractor

extractor = ContractExtractor()

# Extract from PDF bytes
with open("contract.pdf", "rb") as f:
    result = extractor.extract_from_pdf(f.read(), tier="professional")

print(result["analysis"])

Custom Field Selection

Pick specific fields from the 60-field bank instead of using predefined tiers:

result = extractor.extract_from_pdf(
    pdf_bytes,
    custom_fields=[
        "effective_date",
        "expiration_date", 
        "liability_cap",
        "gdpr_obligations",
        "termination_for_convenience"
    ]
)

API Server (Optional)

Run the included Flask API for HTTP access:

python simple_api.py

Send an extraction request:

curl -X POST -F "file=@contract.pdf" -F "tier=professional" http://localhost:5000/analyze

Extraction Tiers

Tier	Fields	Use Case
Essential	9	Quick due diligence
Professional	18	Standard review
Enterprise	25	Full extraction
Custom	1-25	Mix-match from 60-field bank

Available Fields

Category	Fields
Core	effective_date, expiration_date, parties
Financial	total_contract_value, payment_terms, currency
Termination	termination_notice_period, renewal_terms, governing_law
Liability	liability_cap, indemnification_clauses, insurance_requirements, limitation_of_liability
Performance	deliverables, sla_terms, performance_metrics, acceptance_criteria
IP	ip_ownership, license_scope, usage_restrictions
Compliance	confidentiality_period, non_compete_terms, arbitration_clause, audit_rights, data_protection
Administrative	notice_address, amendment_process, assignment_rights, force_majeure
+ Extended	30+ additional fields (GDPR, termination types, risk management, commercial terms)

Environment Variables

ANTHROPIC_API_KEY=your_api_key_here

# Optional: Custom extraction prompt
CLAUDE_EXTRACTION_PROMPT=your_custom_prompt

License

This project is licensed under the GNU Affero General Public License v3.0 (AGPL-3.0).

If you modify this code and provide it as a service, you must make your modifications available under the same license.

Commercial licensing available at qleric.com.

Credits

Built with:

• PyMuPDF - PDF processing
• Anthropic Claude - LLM extraction
• spaCy - NLP utilities