ClawHub
Back to skill

Security audit

Promitheus — Sentience Engine

Security checks across malware telemetry and agentic risk

Overview

The skill is coherent with its stated goal of persistent agent mood, but users should understand that it intentionally writes local state and can feed that state back into future prompts.

Install only if you want generated emotional state to persist across sessions and appear in future agent context. Review the npm plugin before enabling it, keep STATE.md visible and easy to clear, and avoid putting sensitive information into event summaries.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • 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
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
Findings (1)

Missing User Warnings

Medium
Confidence
95% confidence
Finding
The skill explicitly instructs use of a tool that writes a persistent STATE.md file and auto-injects that content into prompts, but it does not warn the user that local files will be modified or that prompt context will be altered automatically. Persistent prompt-injected state can influence later model behavior, carry forward sensitive or untrusted content across sessions, and create a stealthy channel for prompt manipulation if the file contents are ever tampered with.

VirusTotal

65/65 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.