인터페이스는 현지화되어 있으며, Skill 소개와 설명 문서는 선택한 언어로 AI가 자동 번역합니다. 코드·명령어·경로·전문 용어는 원문을 유지하며 원작자 문서가 우선합니다.
Skills 둘러보기
데이터 및 연구 · SKILL.md

analyzing-cyber-insurance

analyzing-cyber-insurance

Structures cyber insurance evaluation with coverage assessment, limit adequacy, and claims scenario analysis. Use when evaluating cyber coverage, analyzing policy terms, or assessing cyber insurance adequacy.

Independent GitHub snapshot · Automated classification · Original documentation governs

Skill 문서

GitHub

name: analyzing-cyber-insurance language: en description: Structures cyber insurance evaluation with coverage assessment, limit adequacy, and claims scenario analysis. Use when evaluating cyber coverage, analyzing policy terms, or assessing cyber insurance adequacy. tags:

  • analysis
  • insurance
  • valuation metadata: author: casemark practice_areas:
    • Insurance
    • Actuarial Science
    • Reinsurance document_types:
    • Analysis Report skill_modes:
    • Analysis

Analyzing Cyber Insurance

When To Use

  • Evaluating a new or renewal cyber insurance policy for coverage adequacy
  • Comparing competing cyber policy quotes across carriers
  • Assessing whether existing cyber limits align with the insured's risk profile
  • Running claims scenarios to stress-test policy response under breach events
  • Reviewing policy language for sublimit traps, retroactive date gaps, or exclusion pitfalls
  • Advising on tower structure or excess placement for large cyber programs

Inputs To Gather

  • Policy documents: Full policy form including declarations page, endorsements, and any manuscript amendments
  • Insured profile: Industry vertical, annual revenue, record count (PII/PHI/PCI), geographic footprint, and IT infrastructure summary
  • Loss history: Prior cyber claims, near-misses, and any regulatory actions or litigation
  • Risk assessment data: Most recent penetration test results, SOC 2 or ISO 27001 status, incident response plan maturity
  • Program structure: Primary and excess layers, attachment points, carrier line-up, and any shared/quota arrangements
  • Benchmark data: Peer company limits, industry loss trends (e.g., Advisen, NetDiligence claims studies), and broker market intelligence
  • Coverage requirements: Contractual insurance obligations, regulatory mandates, and board risk appetite statements

Workflow

  1. Map coverage grants against risk exposures

    • Parse insuring agreements for first-party (business interruption, data restoration, extortion, forensic costs, notification/credit monitoring, crisis management/PR) and third-party (network security liability, privacy liability, regulatory defense and penalties, media liability, PCI-DSS fines) coverages
    • Flag any coverage grants that are absent given the insured's exposure profile
    • Note whether coverage triggers are occurrence-based, claims-made, or claims-made-and-reported, and confirm retroactive date adequacy
  2. Analyze key policy terms and conditions

    • Identify sublimits that effectively cap high-frequency loss categories (e.g., ransomware/extortion sublimit, dependent business interruption sublimit, voluntary notification sublimit)
    • Review waiting period / hours clause for business interruption — compare against insured's actual RTO expectations
    • Examine consent requirements for breach counsel, forensic vendors, and ransom payments [VERIFY carrier panel requirements]
    • Assess war/terrorism and nation-state exclusions — note whether carve-backs for cyberterrorism exist
    • Review infrastructure/cloud exclusion wording — confirm coverage extends to hosted environments (AWS, Azure, SaaS dependencies)
    • Check prior acts, pending/prior litigation, and prior knowledge exclusions for gaps
  3. Evaluate limit adequacy

    • Model loss scenarios across severity tiers:
      • Low: Single employee credential compromise, limited data exposure (~5K records)
      • Medium: Ransomware event with 72-hour operational downtime, partial data exfiltration (~500K records)
      • High: Systemic breach with regulatory investigation across multiple jurisdictions, class action exposure (~5M+ records)
    • For each scenario, stack estimated costs: forensic investigation, notification, credit monitoring, legal defense, regulatory fines, business interruption loss, and reputational/crisis management spend
    • Compare modeled losses against aggregate and per-occurrence limits, noting erosion risk from sublimited coverages
    • Benchmark limits against industry peers by revenue band and data volume [VERIFY current benchmark sources]
  4. Assess program structure and pricing

    • If tower program: evaluate attachment points, follow-form consistency across excess layers, and any differences in conditions (DIC) or differences in limits (DIL) issues
    • Flag anti-stacking provisions, batch/related-claims clauses, and interrelated wrongful acts definitions that could consolidate multiple events into a single limit
    • Compare premium-per-million across layers and against market benchmarks
    • Identify retention (SIR/deductible) appropriateness relative to the insured's balance sheet and incident frequency
  5. Review exclusions and conditions for coverage disputes risk

    • Catalog exclusions most likely to generate coverage disputes: failure to maintain minimum security standards, criminal/fraudulent acts, contractual liability, bodily injury/property damage, unencrypted device exclusions
    • Assess notice/reporting requirements — identify strict compliance conditions that could void coverage
    • Note any cooperation and subrogation obligations that impose operational burdens during active incidents

Output

Structure the analysis report with the following sections:

  • Executive Summary: One-page overview of coverage adequacy rating (Strong / Adequate / Gaps Identified / Inadequate), top three findings, and priority recommendations
  • Coverage Matrix: Table mapping each insuring agreement to the insured's risk exposures, noting sublimits, retentions, and waiting periods per coverage part
  • Limit Adequacy Analysis: Scenario-based loss modeling results with comparison to purchased limits and benchmark data
  • Terms and Conditions Review: Annotated list of policy provisions with risk ratings (Green / Yellow / Red) based on potential to restrict or deny coverage
  • Exclusion Risk Assessment: Ranked list of exclusions by likelihood and severity of coverage dispute
  • Program Structure Evaluation: Visual or tabular representation of tower structure with commentary on layering efficiency and pricing
  • Recommendations: Prioritized action items — coverage enhancements to negotiate, limit increases to pursue, endorsements to request, and alternative market options

Quality Checks

  • Every sublimit, retention, and waiting period from the declarations page is accounted for in the coverage matrix
  • Loss scenarios use defensible cost assumptions with sources cited (NetDiligence, Ponemon/IBM, Verizon DBIR) [VERIFY latest study year and figures]
  • Exclusion analysis references specific policy section numbers and endorsement form numbers
  • Retroactive date, extended reporting period (tail), and policy period are confirmed and noted
  • All jurisdictional regulatory exposure is identified — GDPR, CCPA/CPRA, HIPAA, state breach notification statutes as applicable [VERIFY relevant jurisdictions for insured's footprint]
  • Recommendations distinguish between items achievable in current negotiation vs. next renewal cycle
  • Report flags any areas where actuarial loss modeling or independent broker review is warranted beyond the scope of this analysis
저장한 Skills