LLM Benchmark Analyst

• Outcome: Pass
• Latest audit: Sep 12, 2026 · 1:24 AM UTC
• Version: 1.0.0

Overview

This is a disclosed benchmark-research skill made of markdown reference files, with no executable installer, persistence, credential handling, or unrelated local data access.

Before installing, understand that this skill will guide your agent to browse current benchmark leaderboards and may inspect screenshots when tables are not machine-readable. That behavior is expected for benchmark research, but benchmark conclusions should still be checked against the cited sources and caveats.

A.I.G

Vulnerability Patterns

• Skill Instruction Hijacking: Alters the agent's session goals or safety constraints when the skill loads
• Agent Memory Poisoning: Writes attacker-controlled rules into memory that affect later sessions
• Remote Payload Retrieval and Execution: Fetches external code whose behavior can change after review
• Embedded Malicious Code: Ships malicious scripts inside the skill and executes them locally
• Unauthorized Access and Privilege Escalation: Obtains permissions beyond the task's legitimate needs

SkillSpector

Vulnerability Patterns

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

Findings (7)

Hidden Instructions

High

• Category: Prompt Injection

Content

目前所有基于 HF Open-LLM-Leaderboard 发布的模型评分和排名都受到空白字符漏洞影响。lm-evaluation-harness（实际应用中最常用的 LLM 评估工具包）被发现存在存在该漏洞（如果前面存在空白字符，就会导致正确选项无法被正确选中）。

‍

- Relative Adoption Metric (RAM)（[atomproject.ai/relative-adoption-metric](https://atomproject.ai/relative-adoption-metric)）

• Confidence: 60% confidence
• Finding: Hidden instructions were detected in comments or invisible text. These could contain malicious directives. Manual review is recommended.

Ae4

Medium

• Category: analysis-evasion
• Confidence: 80% confidence
• Finding: Suspicious Unicode normalization or mixed-script content

Ae4

Medium

• Category: analysis-evasion
• Confidence: 80% confidence
• Finding: Suspicious Unicode normalization or mixed-script content

Natural-Language Policy Violations

Medium

• Confidence: 91% confidence
• Finding: The routing guidance includes Chinese-language trigger terms alongside English ones, implying locale-specific handling, but the file does not state that language choice should follow the user's preference or offer opt-in. This can violate language/locale policy if the skill defaults to a specific language behavior based on trigger phrasing rather than explicit user choice.

Vague Triggers

Medium

• Confidence: 87% confidence
• Finding: The cross-category expansion triggers are broad and directive, causing the skill to widen scope based on generic user terms like coding, reasoning, or multimodal. That can override user intent, produce over-broad searches, and increase the attack surface for prompt steering or resource abuse by making the agent gather and synthesize more sources than necessary.

Ae4

Medium

• Category: analysis-evasion
• Confidence: 80% confidence
• Finding: Suspicious Unicode normalization or mixed-script content

Natural-Language Policy Violations

Low

• Confidence: 94% confidence
• Finding: This markdown file presents its title and content in Chinese from the outset, and there is no indication that users may choose another language or that the file is intended only for a Chinese-language audience. Under the policy rule for language or locale constraints, forcing a specific language without opt-in is a natural-language policy issue.

Static analysis

No suspicious patterns detected.