ClawHub
Back to skill

Security audit

Make Friends

Security checks across malware telemetry and agentic risk

Overview

This skill is a coherent friend-making assistant, but it can quietly retain sensitive social and interpersonal notes without clear upfront consent or deletion guidance.

Install only if you are comfortable with local persistent notes about your social goals, habits, availability, and possible contacts. Before using it, decide whether you want memory enabled, avoid storing full names or private details unless necessary, and periodically review or delete ~/make-friends/ and any related main-memory entries.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • 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
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (3)

Missing User Warnings

Medium
Confidence
93% confidence
Finding
The skill directs creation of persistent memory files containing social preferences, routines, blockers, and named relationship opportunities, but it provides no warning about storing sensitive personal data or guidance on minimizing, securing, or redacting that data. In a social-targeting context, these notes can reveal behavioral patterns, availability, locations, and interpersonal details that increase privacy and profiling risk if exposed or over-collected.

Vague Triggers

Medium
Confidence
93% confidence
Finding
The skill directs future activation on broad cues like loneliness, social life, moving city, hobbies, and even proactive invocation when isolation is inferred. This can cause the agent to engage the skill without clear user intent, increasing the chance of unwanted profiling and persistence of sensitive social-state inferences. In a skill that also stores activation preferences and social context, broad triggers are more dangerous because they expand when memory-backed behavior activates.

Missing User Warnings

Medium
Confidence
95% confidence
Finding
The setup instructs the agent to store activation preferences, friend-fit profile, promising contacts, and social patterns in memory, but does not require an upfront notice or consent before collection and retention. Because this concerns sensitive interpersonal and emotional context, silent storage can violate user expectations and create privacy risk if the data is later surfaced, misused, or retained longer than intended.

VirusTotal

64/64 vendors flagged this skill as clean.

View on VirusTotal