Back to skill

Security audit

涉外法律

Security checks for vulnerabilities and agentic risk

Overview

The skill’s legal-research purpose is coherent, but it asks the agent to install and run mutable remote code and use a local Cue API key, so users should review it carefully before enabling it.

Install only if you trust the Cue runner repository and are comfortable with an agent creating or updating ~/.cue/cue-skills, running code from that checkout, and using the Cue API key stored in ~/.cue/config.json. Review the repository or pin it yourself before use, and confirm credit-consuming research runs only for intended subjects.

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
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (3)

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
87% confidence
Finding

The '何时用' section describes the skill scope as '涉外法律:信披与监管、企业尽调、全网检索', which is a very broad domain description rather than a specific invocation boundary. It does not define concrete trigger phrases, exclusions, or negative examples, so the skill could be invoked for many ordinary research requests beyond its intended scope.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

The skill instructs the agent to perform local filesystem writes and networked git operations that clone or update code into ~/.cue/cue-skills, then later execute a Python script from that repository. This creates a supply-chain and local-environment modification risk without prominent upfront warning or tighter trust controls, especially because the repository source can change over time and a fallback mirror is also accepted.

Content

No source excerpt is available for this finding.

Missing User Warnings

Low
Category
Not specified by scanner
Confidence
82% confidence
Finding

The skill states that the runner automatically reads an API key from ~/.cue/config.json, which normalizes access to a locally stored secret without an explicit privacy warning or guidance on least-privilege handling. While it does not directly instruct exfiltration, it encourages secret consumption from local storage and could desensitize users to credential access by agent-driven workflows.

Content

No source excerpt is available for this finding.