PCEC EvoMap Bounty

• Outcome: Review
• Latest audit: May 28, 2026 at 9:37 AM UTC
• Version:
  1.0.0

Overview

The skill is not overtly malicious, but it can drive real EvoMap bounty actions under a fixed node identity without clear user approval boundaries.

Install only if you control the EvoMap node and are comfortable with the agent making bounty-related network requests. Replace the hard-coded node identity with your own authorized configuration, inspect any assets before publication, and manually approve every claim, publish, and complete action.

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 (2)

Vague Triggers

Medium

• Confidence: 92% confidence
• Finding: The trigger phrases are very broad ('PCEC 执行', 'Bounty 任务') and are not scoped to explicit user intent, safe contexts, or read-only behavior. That increases the chance the skill activates during unrelated conversations and initiates workflows that lead to external task fetching, claiming, and completion actions against a live service.

Missing User Warnings

Medium

• Confidence: 96% confidence
• Finding: The skill explicitly describes an automated workflow to fetch, claim, solve, publish, and complete external bounty tasks, but provides no warning, consent gate, or approval checkpoint for state-changing network actions. In context, this is especially risky because the documented actions can alter external system state, consume reputation/resources, and submit assets on behalf of a node.

VirusTotal

66/66 vendors flagged this skill as clean.