ClawHub
Back to skill

Security audit

Multi-Agent Communication

Security checks across malware telemetry and agentic risk

Overview

This instruction-only skill coherently documents multi-agent spawning and messaging, with disclosed but important caveats around invisible background agents and copied attachments.

Install only if you want agent-to-agent orchestration. Prefer visible session mode for sensitive work, avoid allow-all communication settings, keep child-agent limits and timeouts conservative, and do not pass sensitive attachments to child agents unless the copy and retention behavior is acceptable.

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 (2)

Missing User Warnings

Medium
Confidence
91% confidence
Finding
The skill explicitly states that child agents launched in mode="run" are not visible to the user by default, but it does not pair this with a clear warning about hidden execution, delegated actions, or the kinds of side effects those agents may produce. In a multi-agent orchestration skill, invisible background execution reduces user awareness and auditability, making unintended or unauthorized actions harder to detect.

Missing User Warnings

Medium
Confidence
95% confidence
Finding
The design note says attachments are copied to child agents and remain accessible even after the parent deletes them, but it does not warn users about retention, duplication, or privacy consequences. In a distributed agent system, silent snapshotting materially increases data exposure because sensitive files may persist across additional sessions and agent boundaries beyond the user's expectation.

VirusTotal

65/65 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.