ClawHub
Back to skill

Security audit

cue-litigation-drafting

Security checks for vulnerabilities and agentic risk

Overview

This drafting skill needs review because it sends sensitive case details to Cue without adequately explaining how its privacy promises apply to server processing and hosted reports.

Before submitting confidential case materials, confirm Cue's processing and retention terms and who can access hosted reports. The privacy notice does not establish local-only handling or automatic deletion of generated reports. Use redacted information until these controls are clear, and review the downloaded runner before allowing execution.

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
  • Output HandlingUnvalidated Output Injection, Cross-Context Output, Unbounded Output
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (2)

Confidential case information is sent to Cue without a clear third-party data disclosure warning.

High
Category
Output Handling
Confidence
97% confidence
Finding

The prescribed runner sends user-provided case facts and drafting requirements to Cue, while setup and execution are instructed to occur silently. Although the skill identifies Cue as the service provider and mentions server-side drafting, it lacks a clear, user-facing disclosure and authorization step before confidential litigation information leaves the local environment; the privacy assurance may further obscure that transfer. This is particularly consequential for privileged communications, personal information, and nonpublic evidence.

Content

No source excerpt is available for this finding.

The default no-storage privacy assurance conflicts with delivery of a server-accessible copy of the litigation report.

Medium
Category
Excessive Agency
Confidence
89% confidence
Finding

The skill promises default non-storage and local-only temporary handling, yet requires delivery of a server-accessible report link for viewing and forwarding. Although non-storage of source files does not necessarily preclude storage of generated reports, litigation reports can reproduce equally sensitive client information, making the unqualified privacy assurance misleading. The file does not establish whether share links require authentication, expire, or can be revoked, so public accessibility is not proven.

Content

No source excerpt is available for this finding.

Static analysis

No suspicious patterns detected.