ClawHub

Adaptive Reasoning Plus

Security checks across malware telemetry and agentic risk

Overview

This is an instruction-only reasoning skill, but it contains contradictory guidance that could make an agent treat file deletion as a direct low-risk action.

Review before installing in an agent that can edit or delete files. The skill has no executable payload, but its instructions should be corrected so destructive or irreversible operations require target verification, explicit confirmation, and elevated reasoning.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • 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
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (2)

Vague Triggers

Medium
Confidence
89% confidence
Finding
The activation triggers are broad phrases like '需要深度分析', '复杂问题', and '多步推理', which can match many ordinary prompts and cause the skill to activate unexpectedly. In an agent setting, overbroad activation can alter response behavior, increase hidden reasoning paths, and bypass more specific safety or routing logic that should have handled the request.

Missing User Warnings

Medium
Confidence
95% confidence
Finding
The validation examples classify '这个文件删了' as a low-complexity task to execute directly, normalizing destructive action without requiring confirmation, scope validation, or impact warning. In an agent environment with tool access, this can lead to accidental deletion of data or systems through ambiguous or underspecified user commands.

VirusTotal

64/64 vendors flagged this skill as clean.

View on VirusTotal