ClawHub
Back to skill

Security audit

Agent Config

Security checks across malware telemetry and agentic risk

Overview

This skill is not malicious, but it gives an agent broad authority to persistently rewrite its own behavior and related logs beyond the most clearly advertised files.

Install only if you intentionally want an agent to help maintain its own configuration. Require explicit approval before edits to AGENTS.md, SOUL.md, TOOLS.md, MEMORY.md, HEARTBEAT.md, BOOTSTRAP.md, daily logs, templates, skills, or decision/failure logs, and periodically review those files for sensitive information or unwanted behavior changes.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • 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
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (4)

Context-Inappropriate Capability

Medium
Confidence
94% confidence
Finding
The skill's stated purpose is modifying core context files, but it also instructs the agent to write change logs to an unrelated absolute path outside that declared scope. This expands the skill's write surface and creates an avenue for unintended persistence or tampering in external workspace files if the skill is triggered broadly or abused.

Context-Inappropriate Capability

Medium
Confidence
92% confidence
Finding
Directing the agent to write failure logs to a separate absolute-path learnings file extends behavior beyond the advertised core context files and introduces another persistence location outside the primary scope. Such side-effect writes can accumulate sensitive operational details or be abused to modify unrelated agent behavior artifacts over time.

Vague Triggers

High
Confidence
96% confidence
Finding
The activation criteria are extremely broad, covering almost any request related to changing behavior, rules, memory, delegation, safety, or prompt patterns. A skill that can rewrite foundational agent context with such loose triggers materially increases the risk of accidental or adversarial invocation, enabling prompt-level persistence or policy tampering.

Missing User Warnings

Medium
Confidence
88% confidence
Finding
The file documents a rule to perform 'silent logging' of events to daily logs without announcing or asking permission. Because these context files can include personal, operational, and user-related information, normalizing covert record creation increases privacy risk and can lead to collection of sensitive data without meaningful user awareness or consent.

VirusTotal

64/64 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.