ClawHub

Agent Session State

Security checks across malware telemetry and agentic risk

Overview

This memory skill is mostly transparent and purpose-aligned, but it can persist raw long-session user exchanges in a shared workspace buffer with weak scoping and retention controls.

Install only if you are comfortable with the agent writing conversation-derived state into workspace memory files. Before using it in multi-user or sensitive environments, consider editing the working-buffer behavior to be per-channel, redact secrets and personal data before writing, require user opt-in for raw exchange buffering, and define a clear deletion schedule.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
Findings (2)

Missing User Warnings

Medium
Confidence
93% confidence
Finding
The skill directs the agent to persist conversation-derived state into workspace files but does not clearly warn users that their messages, decisions, and facts may be written to disk. In a multi-user or sensitive environment, this can undermine user expectations, increase privacy risk, and lead to retention of data that users did not realize would become durable artifacts.

Ssd 3

Medium
Confidence
96% confidence
Finding
The instruction to log every exchange into a natural-language working buffer creates a broad retention path for user-provided content, including potentially sensitive material that may not be recognized in real time. Because the buffer captures raw user messages plus model summaries, it increases the chance of storing secrets, personal data, or regulated information beyond what is necessary for continuity.

VirusTotal

65/65 vendors flagged this skill as clean.

View on VirusTotal