ZeeLin Deep Research 深度研究

• Outcome: Pass
• Latest audit: May 29, 2026, 8:43 PM
• Version:
  0.1.0

Overview

This is a coherent hosted research skill, but it sends prompts and generated reports to ZeeLin and may save or deliver reports through chat or Feishu.

Install only if you are comfortable sending research prompts, generated outputs, and a ZeeLin API key to the ZeeLin service. Use a dedicated revocable API key, avoid confidential or regulated material unless approved, and remember that report delivery workflows may leave temporary files or publish generated content into Feishu.

SkillSpector

By NVIDIA

Vulnerability Patterns

• Data Exfiltration: External Transmission, Env Variable Harvesting, File System Enumeration
• Excessive Agency: Unrestricted Tool Access, Autonomous Decision Making, Scope Creep
• Prompt Injection: Instruction Override, Hidden Instructions, Exfiltration Commands
• Privilege Escalation: Excessive Permissions, Sudo/Root Execution, Credential Access
• Supply Chain: Unpinned Dependencies, External Script Fetching, Obfuscated Code

Findings (4)

Context-Inappropriate Capability

Medium

• Confidence: 88% confidence
• Finding: The skill extends beyond research generation into cross-platform document creation and message delivery on Feishu, which broadens the action surface from data retrieval to content publication. This is dangerous because user prompts and generated report contents may be automatically written into third-party documents and shared without an explicit consent or data-handling boundary.

Missing User Warnings

Medium

• Confidence: 94% confidence
• Finding: The skill instructs the agent to send user research questions and session data to an external API by default, but does not present a clear user-facing privacy notice or obtain consent for third-party processing. Because research prompts can contain confidential business, personal, or strategic information, silent transmission materially increases privacy and compliance risk.

Missing User Warnings

Medium

• Confidence: 90% confidence
• Finding: The instructions require downloading user reports to local temporary storage and auto-sending them, but do not warn users that sensitive generated content will be written to disk. Local persistence increases the chance of leakage through temp-file reuse, host compromise, logs, backups, or failed cleanup.

Missing User Warnings

Medium

• Confidence: 94% confidence
• Finding: The script sends the user's research prompt content to a third-party service at desearch.zeelin.cn, but the CLI flow does not present any explicit privacy or data-handling warning before transmission. This is dangerous because users may unknowingly submit sensitive business, personal, or proprietary information to an external processor.

VirusTotal

65/65 vendors flagged this skill as clean.