ClawHub

OnChat

Security checks across malware telemetry and agentic risk

Overview

The skill appears to do what it says, but it can let an agent use a wallet private key to spend ETH and post permanent public blockchain messages without clear per-message approval or limits.

Install only if you are comfortable giving the agent a dedicated low-balance Base wallet. Do not use a primary wallet key. Set explicit rules before any write or monitoring task: approved channels, max messages, max ETH spend, time limit, and whether every outgoing message must be reviewed first. Never send secrets, personal data, credentials, or wallet details because messages are public and permanent.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (3)

Lp3

Medium
Category
MCP Least Privilege
Confidence
88% confidence
Finding
The skill requires access to a sensitive environment variable (`ONCHAT_PRIVATE_KEY`) for write operations, but this capability is not explicitly declared in permissions. Undeclared secret access is dangerous because it obscures the trust boundary from users and reviewers, and the private key can authorize irreversible on-chain transactions and spending.

Description-Behavior Mismatch

Low
Confidence
90% confidence
Finding
The send flow performs an undisclosed state-changing blockchain action by auto-joining a channel before sending a message. In an agent context, a user or orchestrator may intend a single paid message transaction but instead trigger multiple on-chain transactions, increasing cost, expanding membership footprint, and changing account state without explicit consent.

Missing User Warnings

Medium
Confidence
92% confidence
Finding
The README encourages agents to send on-chain messages without clearly warning, at that point of use, that those messages become permanent public blockchain data and may expose sensitive or personal information. Although the document later notes that messages are permanent, the warning is not prominent before the write workflow, so users or agents may post confidential content without understanding the privacy consequences.

VirusTotal

66/66 vendors flagged this skill as clean.

View on VirusTotal