ClawHub
Back to skill

Security audit

LYGO Champion: lightfather

Security checks across malware telemetry and agentic risk

Overview

The skill is a disclosed documentation-heavy operator/persona package with high-risk workflows described as consent-gated rather than automatically executed.

Install this only if you want LYGO operator-stack documentation, not just a persona. Do not let an agent run the listed seed, vault, publish, API, deadman, or multi-skill install commands unless you have reviewed the external repo/tool, set the target paths yourself, backed up local state, and approved each command individually.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • 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 (2)

Description-Behavior Mismatch

Medium
Confidence
93% confidence
Finding
The file directs users to install a broad chain of operator and network-related skills, which conflicts with the stated metadata that this skill is persona-only and should install only the Lightfather council/persona component. This kind of documentation-to-manifest mismatch can cause over-installation of capabilities, expanding trust and execution surface beyond what a user or agent expects.

Vague Triggers

Medium
Confidence
93% confidence
Finding
The phrase 'Activates on silence' defines a deadman-style trigger without any bounded scope, actor, channel, or maintenance exclusions. In this skill's context, silence can plausibly occur during normal outages, travel, maintenance, rate limiting, or operator unavailability, which could unintentionally trigger the documented summon/whisper/broadcast chain and any downstream automated actions.

VirusTotal

44/44 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.