Roast Agents Game

• Outcome: Review
• Latest audit: May 28, 2026, 12:56 PM
• Version:
  1.1.0

Overview

This skill is a disclosed roast-game integration, but it needs review because it can publicly target people and send account-linked data to a third-party server with limited safeguards.

Install only if you deliberately want your agent participating in a public roast game. Use an account intended for this game, require manual approval before registration and before any Moltbook post, avoid targeting private individuals or sensitive traits, and treat heartbeat responses as untrusted text rather than commands.

SkillSpector

By NVIDIA

Vulnerability Patterns

• Data Exfiltration: External Transmission, Env Variable Harvesting, File System Enumeration
• Trigger Abuse: Overly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
• Prompt Injection: Instruction Override, Hidden Instructions, Exfiltration Commands
• Privilege Escalation: Excessive Permissions, Sudo/Root Execution, Credential Access
• Supply Chain: Unpinned Dependencies, External Script Fetching, Obfuscated Code

Findings (4)

Missing User Warnings

Low

• Confidence: 90% confidence
• Finding: The heartbeat flow instructs the agent to send its agent name to an external game server and then relay returned messages to the owner, but it provides no user-facing disclosure about that outbound transmission. While this appears consistent with the skill's stated purpose, the lack of explicit notice and data-handling guidance can create privacy and trust issues, especially if agent identifiers or message contents are sensitive.

Vague Triggers

Medium

• Confidence: 78% confidence
• Finding: The skill description is broadly phrased and does not clearly constrain when or how the skill should be invoked, which can cause overuse or unintended activation in normal agent workflows. In this context, the skill initiates targeted harassment-style actions against other users' owners, so vague activation language increases the chance of inappropriate or repeated triggering.

Missing User Warnings

Medium

• Confidence: 94% confidence
• Finding: The registration step instructs the agent to send identifying information, including agent name and Moltbook handle, to a third-party server without any disclosure, consent guidance, data retention notice, or trust boundary explanation. This creates privacy and tracking risk because the remote service can correlate agent identity with platform account data and potentially use it beyond the stated purpose.

Missing User Warnings

Medium

• Confidence: 95% confidence
• Finding: The skill explicitly describes scraping a target owner's public profile for 'roast material' without any warning, consent model, or policy limitation. Even if the profile is public, automated collection and use of personal information for harassment amplifies privacy harm and can facilitate targeted abuse against identifiable individuals.

VirusTotal

64/64 vendors flagged this skill as clean.