ClawHub

pionex-bot

Security checks across malware telemetry and agentic risk

Overview

This Pionex bot skill is mostly coherent, but it adds a high-impact signal-publishing write action that is not clearly declared in the top-level description.

Install only if you intend an agent to both manage Pionex trading bots and potentially push Pionex trading signals. Use a dedicated least-privilege API key, review every generated command, and require explicit confirmation before any non-dry-run bot or signal-platform write action.

SkillSpector

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

Description-Behavior Mismatch

Medium
Confidence
95% confidence
Finding
The skill’s documented scope is bot-order lifecycle management, but it also routes and documents a signal-platform write capability. This expands the skill’s authority beyond what the manifest description clearly declares, increasing the chance that an agent or reviewer will invoke a write action with different semantics than expected. In a trading context, undeclared signal-push functionality is sensitive because it can trigger downstream trading behavior on an external platform.

Description-Behavior Mismatch

Medium
Confidence
96% confidence
Finding
The command set exposes `bot signal listener`, a write operation to the Pionex signal platform, even though the manifest description frames the skill around creating/querying/adjusting/canceling bot orders. This mismatch can mislead orchestration and human review, causing unintended invocation of a capability that publishes external trading signals rather than merely managing existing bots. Because it is a write path affecting trading automation, hidden or underdeclared scope materially raises security risk.

VirusTotal

64/64 vendors flagged this skill as clean.

View on VirusTotal