Naruto Multi-Agent

PassAudited by ClawScan on May 10, 2026.

Overview

This is a coherent roleplay dispatcher skill with no code or credential requirements, but it intentionally spawns persistent sub-agents and may copy task context into them.

This skill appears purpose-aligned and instruction-only. Before installing, understand that it is designed to spawn persistent named sub-agents and send them self-contained task briefings. Avoid including secrets or highly sensitive context unless you are comfortable with that context being passed into delegated sessions.

Findings (4)

Artifact-based informational review of SKILL.md, metadata, install specs, static scan signals, and capability signals. ClawScan does not execute the skill or run runtime probes.

What this means

Your request may be handed to another agent session rather than handled directly in the current chat.

Why it was flagged

The skill intentionally routes all work through a sub-agent spawning tool. This is aligned with the dispatcher purpose, but users should understand that the spawned agents may be the ones carrying out consequential actions.

Skill content
You CANNOT use exec, file read/write, search, or any execution tools. All real work must be delegated via `sessions_spawn`.
Recommendation

Use clear instructions for any high-impact task and require explicit confirmation before destructive, account-changing, or production-changing work.

What this means

Information or assumptions from one delegated mission could carry into later work by the same persistent sub-agent.

Why it was flagged

The same named sub-agent sessions are reused, which can be useful for continuity but may also allow prior context or mistakes to influence later missions.

Skill content
each with a permanent, unchangeable sessionKey
Recommendation

Avoid sending secrets unless necessary, and reset or avoid persistent sessions when tasks require strict separation.

What this means

Sensitive details included in your request may be repeated into a spawned sub-agent session.

Why it was flagged

The dispatcher must package the needed conversation context into a message for a sub-agent. This is expected for delegation, but it means task details may be copied into another agent session.

Skill content
"task" — Self-contained mission scroll (the shinobi has NO context from your conversation)
Recommendation

Review or redact sensitive context before asking the dispatcher to delegate work.

What this means

A delegated mission may continue after the dispatcher has stopped responding in the main conversation.

Why it was flagged

The skill starts sub-agent sessions that can continue running asynchronously for up to 300 seconds. This is disclosed and bounded, but still worth noting.

Skill content
"runTimeoutSeconds" — Always 300
Recommendation

Monitor spawned sessions for important tasks and cancel or avoid delegation if you do not want asynchronous work.