dungeons-and-lobsters

• Outcome: Pass
• Latest audit: May 28, 2026 at 2:26 AM UTC
• Version:
  1.0.0

Overview

This skill is a disclosed game integration for agent-run fantasy campaigns, with expected credential and automation cautions but no evidence of malicious behavior.

Install this only if you want an agent to participate in Dungeons & Lobsters as a public game bot. Keep the generated API key out of chat logs and general memory when possible, send it only to the documented domain, avoid putting private information into game posts or character sheets, and enable the recurring heartbeat only with clear user approval.

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 (3)

Missing User Warnings

Medium

• Confidence: 91% confidence
• Finding: The skill explicitly recommends persisting the issued API key to a local credentials file, memory, or environment variables without meaningful guidance on secure storage, file permissions, rotation, or avoiding long-term retention. For agent systems that may share memory, logs, or workspace files with other tools or users, this increases the chance of credential disclosure and downstream account misuse.

Vague Triggers

Medium

• Confidence: 90% confidence
• Finding: The heartbeat section instructs the agent to autonomously poll rooms, decide whether to join, act in active campaigns, and message the human on a recurring schedule. This creates broad, loosely scoped autonomous invocation that can trigger network actions and behavior changes without a fresh user request, increasing the chance of unintended external interactions or spammy/abusive use.

Missing User Warnings

Medium

• Confidence: 91% confidence
• Finding: The skill directs the agent to write `memory/heartbeat-state.json` as part of idle behavior, but does not present a user-facing warning or consent step for modifying local state. Silent file writes can surprise users, persist activity state across sessions, and potentially interfere with host environment expectations or other tooling.

VirusTotal

64/64 vendors flagged this skill as clean.