ClawHub

Daily Learning

Security checks across malware telemetry and agentic risk

Overview

This daily-learning skill is mostly coherent, but it tells recurring agents to inspect prior user context and turn inferred needs into durable learning plans without clear consent or limits.

Install only if you are comfortable with a recurring learning agent reading workspace guidance and recent conversation context to personalize future topics. Before use, restrict it to explicit learning requests or approved project files, disable recent-conversation mining unless intentionally wanted, and review any wiki-bound output before it goes into a shared knowledge base.

SkillSpector

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

Context-Inappropriate Capability

Medium
Confidence
94% confidence
Finding
The skill instructs the agent to inspect AGENTS.md and recent conversations to infer future user needs, which expands data collection beyond the immediate learning task. That creates a privacy and scope-creep risk because user context may be mined and repurposed without explicit consent or minimization, especially in an automated cron-style workflow.

Missing User Warnings

Medium
Confidence
95% confidence
Finding
The skill tells agents to analyze recent conversations and AGENTS.md for future planning without any privacy notice, consent check, or limitation. In practice, this can normalize silent reuse of user-provided context for secondary purposes, increasing the risk of overcollection, inappropriate inference, and persistence of sensitive information.

Ssd 3

Medium
Confidence
96% confidence
Finding
Mining user conversations and AGENTS.md to generate future learning topics repurposes contextual data beyond the immediate task and may feed it into durable planning artifacts. In a recurring learning workflow, that makes the behavior more dangerous because collection can become routine and cumulative, potentially capturing sensitive project or personal details over time.

VirusTotal

63/63 vendors flagged this skill as clean.

View on VirusTotal