Vibeship-spawner-skills contract-analysis

id: contract-analysis name: Contract Analysis category: legal description: Use when reviewing contracts, extracting key terms, identifying risks, or building contract analysis tools - covers NLP approaches, clause identification, and risk scoring

patterns: golden_rules: - rule: "Never skip human review" reason: "AI finds patterns, lawyers find issues" - rule: "Benchmark against playbook" reason: "Standard terms define 'acceptable'" - rule: "Flag missing clauses" reason: "What's NOT there matters most" - rule: "Track versions" reason: "Changes between drafts reveal intent" - rule: "Context matters" reason: "Same clause means different things in different contracts"

analysis_pipeline: - "INGEST - PDF/DOCX, OCR" - "PARSE - Structure, Sections" - "EXTRACT - Clauses, Dates, Parties" - "ANALYZE - Compare to Playbook" - "REPORT - Risk Summary"

critical_clauses: indemnification: - "Mutual vs. one-way" - "Scope of covered claims" - "Cap on liability" - "Exclusions and carve-outs" limitation_of_liability: - "Cap amount (ideally = contract value)" - "Exclusions from cap" - "Consequential damages waiver" - "Carve-outs for IP infringement" termination: - "Termination for convenience" - "Cure period for material breach" - "Wind-down obligations" - "Data return/destruction" intellectual_property: - "Ownership of work product" - "License scope and restrictions" - "Background IP rights" - "IP indemnification" confidentiality: - "Definition of confidential information" - "Permitted disclosures" - "Term of obligations" - "Return/destruction requirements" data_protection: - "Data processing scope" - "Subprocessor rights" - "Breach notification timeline" - "Data subject rights compliance"

risk_levels: low: "Standard terms, minor variations" medium: "Some deviations, negotiable" high: "Significant risk, requires attention" critical: "Unacceptable terms, must negotiate"

anti_patterns:

• pattern: "Keyword-only search" problem: "Misses synonyms and context" solution: "Use NLP with semantic understanding"
• pattern: "No playbook baseline" problem: "Can't identify deviations" solution: "Define standard acceptable terms"
• pattern: "Ignoring missing clauses" problem: "Silent risks" solution: "Check for expected clause presence"
• pattern: "Single-pass review" problem: "Misses cross-references" solution: "Analyze clause interactions"
• pattern: "No version tracking" problem: "Can't see negotiation history" solution: "Compare all drafts"

implementation_checklist: extraction: - "Extract all parties and defined terms" - "Identify and classify key clauses" - "Extract dates, amounts, and deadlines" analysis: - "Score each clause against playbook" - "Flag missing expected clauses" - "Highlight unusual or non-standard terms" comparison: - "Compare against previous version if available" - "Track changes between negotiation rounds" reporting: - "Calculate overall risk score" - "Generate summary of critical issues" - "Create prioritized recommendations" - "Have legal counsel review findings"

handoffs:

• skill: gdpr-privacy trigger: "data processing agreements"
• skill: patent-drafting trigger: "IP licensing terms"

ecosystem: nlp: - "SpaCy - Entity extraction" - "Hugging Face Transformers - Classification" - "LegalBERT - Legal NLP" extraction: - "docx2txt - DOCX parsing" - "PyMuPDF - PDF extraction" - "Tesseract - OCR" comparison: - "difflib - Text comparison" - "python-docx-compare"

sources: tutorials: - "AI Contract Analysis Best Practices" - "NLP for Contract Review" articles: - "AI Contract Review 2025"