polymarket-predictradar-daily-anomalies-skills

• Outcome: Pass
• Latest audit: May 29, 2026 at 8:34 PM UTC
• Version:
  1.0.0

Overview

This is a disclosed Polymarket anomaly-reporting skill, but it can save wallet-linked public-market findings locally and should be used intentionally.

Install only if you want Polymarket anomaly reports that include full public wallet addresses. Run it from a directory where saved JSON reports are acceptable, delete or redact old reports if needed, verify the external polymarket-data-layer dependency, and treat labels like Whale Wars or Insider Watch as research leads rather than trading advice.

SkillSpector

By NVIDIA

Vulnerability Patterns

• Trigger Abuse: Overly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
• MCP Tool Poisoning: Hidden Instructions, Unicode Deception, Parameter Description Injection
• Prompt Injection: Instruction Override, Hidden Instructions, Exfiltration Commands
• Data Exfiltration: External Transmission, Env Variable Harvesting, File System Enumeration
• Privilege Escalation: Excessive Permissions, Sudo/Root Execution, Credential Access

Findings (4)

Description-Behavior Mismatch

Medium

• Confidence: 93% confidence
• Finding: The script persists raw wallet addresses, trade sizes, timestamps, and market-level anomaly labels to a local JSON file, creating a durable intelligence dataset rather than only generating an ephemeral narrative report. Even if wallet addresses are public on-chain, aggregating and exporting them in a structured daily file materially increases privacy, profiling, and misuse risk beyond the stated skill purpose.

Vague Triggers

Medium

• Confidence: 89% confidence
• Finding: The trigger set is broad enough to match common user requests like 'daily report', 'what happened today', or 'any anomalies', which can cause the skill to auto-activate unexpectedly. That increases the chance of unintentional external queries, report generation, and disclosure of sensitive trading-related data such as full wallet addresses when the user did not explicitly request this skill.

Natural-Language Policy Violations

Medium

• Confidence: 85% confidence
• Finding: Forcing a Chinese interpretation in output without user opt-in is an unsafe output-policy choice because it can override user intent, degrade accuracy, and miscommunicate sensitive market information. In a financial-analysis context, unwanted translation or reinterpretation can materially change meaning and cause users to act on misunderstood signals.

Missing User Warnings

Medium

• Confidence: 91% confidence
• Finding: The file write occurs automatically and stores potentially sensitive analytical output, including wallet-linked anomaly assessments, without user awareness, consent, or any guardrail around destination or retention. This can expose data to other local users, backup systems, logs, or downstream tooling that consumes the generated file, increasing unintended disclosure risk.

VirusTotal

54/54 vendors flagged this skill as clean.