ClawHub

ekalavya-self-improvement

Security checks across malware telemetry and agentic risk

Overview

This instruction-only skill is coherent with its stated goal of helping an assistant keep working on already approved tasks, but users should set clear boundaries for edits, commits, and durable rule changes.

Install this only if you want the assistant to continue executing already approved work with fewer status updates. Before using it, tell the assistant whether it may commit changes, when it must ask before editing files, and whether creating or modifying durable skills, rules, or project documents needs explicit approval.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • 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 (2)

Vague Triggers

Medium
Confidence
89% confidence
Finding
The activation criteria are broad and subjective, using cues like 'at risk of drifting', 'keep moving expectations', and 'simple UI/product shaping' without a tight trigger boundary. That can cause the skill to activate in many normal conversations and bias the agent toward execution-oriented behavior when the user may not have explicitly consented to file changes or autonomous action.

Missing User Warnings

Medium
Confidence
95% confidence
Finding
The skill explicitly instructs the agent to 'Edit files or run the next concrete action immediately' and to 'Commit when the change is meaningful' while also emphasizing 'silent execution mode' and infrequent updates. In security terms, this reduces transparency around impactful actions and can normalize autonomous modifications to user assets without contemporaneous notice, increasing the chance of unauthorized or surprising changes.

VirusTotal

64/64 vendors flagged this skill as clean.

View on VirusTotal