ClawHub

context-clean-up

Security checks across malware telemetry and agentic risk

Overview

This is an audit-only context cleanup skill with disclosed local session analysis; its notification guidance is worth handling carefully but does not show hidden or automatic data transfer.

Install only if you are comfortable with a tool inspecting OpenClaw session history for audit purposes. Treat any generated JSON report as private, review it before sharing, and use the out-of-band cron notification advice only with channels you control and with sensitive details redacted.

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
87% confidence
Finding
The document instructs an isolated cron worker to deliver content directly to external messaging platforms and to persist details to files, which goes beyond the stated audit-only, no-changes scope of the skill. That mismatch is risky because it normalizes side-effecting behavior and creates a path for hidden data movement outside the main transcript, reducing user visibility and reviewability.

Context-Inappropriate Capability

Medium
Confidence
89% confidence
Finding
Documenting external message delivery in a context-cleanup audit skill expands the apparent capability boundary and encourages agents to use outbound channels unrelated to the skill's purpose. Because the stated goal is transcript hygiene, introducing Telegram/Discord/Slack delivery makes the skill more dangerous by enabling out-of-band transmission that may bypass normal conversational oversight.

Missing User Warnings

Low
Confidence
78% confidence
Finding
The recommendation to write details to a file introduces local state changes and potential sensitive-data persistence without warning about storage location, retention, access controls, or redaction. In an audit-only skill, even seemingly minor file writes are problematic because they can leave behind data artifacts the user did not expect.

VirusTotal

65/65 vendors flagged this skill as clean.

View on VirusTotal