ClawHub

Discover Artblocks Projects

Security checks across malware telemetry and agentic risk

Overview

This Art Blocks browsing skill is mostly read-only, but it points agents toward purchase-transaction and broad query tools without clear safeguards.

Install only if you are comfortable with the external artblocks-mcp server and with agents querying Art Blocks wallet/profile data. Treat any purchase-transaction follow-up as out of scope for a simple discovery skill unless you explicitly ask for it and review the transaction details yourself.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • 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
Findings (3)

Description-Behavior Mismatch

Medium
Confidence
95% confidence
Finding
The skill is presented as a discovery/browsing capability, but the documentation directs the agent to use additional undeclared operations such as GraphQL tools and downstream tools including transaction-related flows. This creates capability drift between the declared interface and the documented behavior, which can cause an orchestrator or reviewer to underestimate what the skill may induce an agent to do.

Context-Inappropriate Capability

High
Confidence
98% confidence
Finding
Referencing `build_purchase_transaction` inside a skill framed as read-only discovery introduces an unexpected path from information retrieval to transaction preparation. Even if no signing occurs here, normalizing or encouraging transaction-building in a browsing skill increases the chance of unsafe agent behavior, confused-deputy actions, or user surprise about financial operations.

Missing User Warnings

Medium
Confidence
90% confidence
Finding
The wallet and profile features explicitly state that holdings and eligibility are aggregated across all linked wallets, but the skill provides no user-facing privacy warning or consent guidance. This can expose a broader financial/profile picture than a user expects, especially when querying by username, increasing privacy and profiling risk.

VirusTotal

65/65 vendors flagged this skill as clean.

View on VirusTotal