ClawHub
Back to skill

Security audit

Insurance Operations Automation

Security checks for vulnerabilities and agentic risk

Overview

This skill is a high-stakes insurance automation guide that encourages automated claims and underwriting decisions without enough guardrails for regulated data or human review.

Review carefully before installing or using this skill in production. It should be treated as planning material only unless your organization adds explicit controls for licensed human approval, permissible underwriting factors, claims authority limits, fraud review, audit logging, data minimization, retention, and jurisdiction-specific insurance/privacy compliance.

A.I.G

Vulnerability Patterns
  • Skill Instruction HijackingAlters the agent's session goals or safety constraints when the skill loads
  • Agent Memory PoisoningWrites attacker-controlled rules into memory that affect later sessions
  • Remote Payload Retrieval and ExecutionFetches external code whose behavior can change after review
  • Embedded Malicious CodeShips malicious scripts inside the skill and executes them locally
  • Unauthorized Access and Privilege EscalationObtains permissions beyond the task's legitimate needs

SkillSpector

Vulnerability Patterns
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (4)

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
92% confidence
Finding

The skill promotes end-to-end automation over underwriting, claims, policy management, and compliance using highly sensitive insurance and policyholder data, but provides no warning about privacy, jurisdictional restrictions, data minimization, consent, retention, or mandatory human oversight. In this context, an agent could be deployed to process regulated personal, financial, health, or fraud-related data in ways that violate privacy law or create harmful operational decisions without review.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The claims pipeline explicitly describes automatic approval of low-value claims without any warning about financial loss, fraud risk, unfair claims handling, or regulatory obligations. Because claims adjudication is a regulated, high-stakes activity, automating approval decisions without mandatory controls or human review can cause improper payments, discriminatory outcomes, and compliance violations.

Content

No source excerpt is available for this finding.

Autonomous Decision Making

Medium
Category
Excessive Agency
Confidence
94% confidence
Finding

This line instructs autonomous claim severity routing with 'Green' claims being auto-approved under a monetary threshold. In insurance, even low-dollar claims can involve fraud indicators, coverage nuances, consumer protection obligations, and reserve implications, so a fully automated approval path materially increases the risk of improper claim decisions and exploitable payout logic.

Content

Scanner excerpt · SKILL.md (reported line 19)May include surrounding context.

md
### 2. Claims Processing Pipeline
- FNOL intake automation (voice + digital, structured extraction)
- Severity triage: Green (auto-approve <$2K) → Yellow (adjuster review $2K-$25K) → Red (SIU referral >$25K or fraud indicators)
- Subrogation identification triggers
- Reserve estimation formulas by claim type
- Settlement authority matrix

Autonomous Decision Making

Medium
Category
Excessive Agency
Confidence
88% confidence
Finding

The underwriting prompt directs the agent to recommend 'Auto-approve / Refer to senior / Decline' based on risk scoring, including use of credit and occupational factors, without any fairness, explainability, or human-review safeguards. In this context, autonomous or quasi-autonomous underwriting recommendations can lead to discriminatory or non-compliant adverse decisions if operators over-rely on the model.

Content

Scanner excerpt · SKILL.md (reported line 76)May include surrounding context.

  1. Score each factor against the risk matrix (1-10 scale)
  2. Calculate composite risk score (weighted by line of business)
  3. Flag any referral triggers (prior losses >3 in 5yr, credit <600, high-hazard occupation)
  4. Recommend: Auto-approve / Refer to senior / Decline with reason
  5. Generate underwriting memo with supporting data
text

Static analysis

No suspicious patterns detected.