ClawHub
Back to skill

Security audit

deep-research

Security checks across malware telemetry and agentic risk

Overview

This research skill appears mostly purpose-aligned, but it needs review because it can load executable code from an environment-selected path and may run on broad research-like prompts.

Install only if you are comfortable with prompts being sent to an external AI service and with reviewing the runtime environment. Do not set GOOGLE_GENAI_SDK_PATH unless it points to trusted code, and use explicit requests when invoking the skill.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • 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)

Context-Inappropriate Capability

Medium
Confidence
92% confidence
Finding
The script dynamically imports executable code from a path supplied by the GOOGLE_GENAI_SDK_PATH environment variable, with only minimal path normalization and no trust validation. If an attacker can influence the environment or deployment configuration, they can cause arbitrary local JavaScript to be loaded and executed with the privileges of the skill process, which exceeds what is necessary for a research/report-generation helper.

Vague Triggers

Medium
Confidence
87% confidence
Finding
The trigger list includes broad everyday phrases such as '研究一下', '详细分析', and 'deep dive', which can match many normal user requests and cause the skill to be invoked unexpectedly. In a skill that can launch long-running autonomous research and consume an API-backed external service, accidental invocation can waste resources, expose user prompts to a third-party model, and create confusing or privacy-impacting behavior.

VirusTotal

63/63 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.