Naruto Multi-Agent CN

• Analyzed: May 14, 2026, 3:45 AM
• Findings: 1
• Version: 1.5.0
• Hash: b6f1e444653f7ec8310eb43aeeafce44e7f8195a175394667236ba5f3e2c628b
• Source: palm
• External report: View on VirusTotal

Findings (1)

The skill bundle itself is designed as a multi-agent dispatcher, explicitly forbidding the main agent from performing risky actions like `exec`, file I/O, or search (SKILL.md). However, it instructs the agent to take user input and pass it directly into the `task` parameter of `sessions_spawn` for delegation to sub-agents. This creates a significant prompt injection vulnerability, as a malicious user could craft harmful instructions that would then be executed by the downstream sub-agents, making it a risky capability without clear malicious intent within this skill.