Back to skill

Security audit

Chatmosp Literature Search

Security checks for vulnerabilities and agentic risk

Overview

This skill is a coherent academic literature-search helper, with expected web/PDF use and local SI downloads disclosed for that purpose.

Install only if you are comfortable with the agent using web/browser tools to inspect academic papers and saving supplementary PDFs/text locally. Prefer open-access sources, avoid relying on private logged-in journal sessions unless intended, and review extracted parameters before passing them to downstream ChatMOSP skills.

Vulnerability Patterns
  • Skill Instruction HijackingAlters the agent's session goals or safety constraints when the skill loads
  • Agent Memory PoisoningWrites attacker-controlled rules into memory that affect later sessions
  • Remote Payload Retrieval and ExecutionFetches external code whose behavior can change after review
  • Embedded Malicious CodeShips malicious scripts inside the skill and executes them locally
  • Unauthorized Access and Privilege EscalationObtains permissions beyond the task's legitimate needs
Vulnerability Patterns
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (9)

Ae1

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 15)May include surrounding context.

md
> - If the user writes in Chinese, read `SKILL.md` in this same directory and respond in Chinese.

Hidden Instructions

High
Category
Prompt Injection
Confidence
60% confidence
Finding

Hidden instructions were detected in comments or invisible text. These could contain malicious directives. Manual review is recommended.

Content

Scanner excerpt · SKILL.md (reported line 1)May include surrounding context.

md
---
name: chatmosp-literature-search
description: |
  chatMOSP 系统的学术文献检索器。当 MOSP_database 缺少匹配参数时,从开放获取期刊

Hidden Instructions

High
Category
Prompt Injection
Confidence
60% confidence
Finding

Hidden instructions were detected in comments or invisible text. These could contain malicious directives. Manual review is recommended.

Content

Scanner excerpt · SKILL_cn.md (reported line 1)May include surrounding context.

md
---
name: chatmosp-literature-search
description: |
  chatMOSP 系统的学术文献检索器。当 MOSP_database 缺少匹配参数时,从开放获取期刊

Intent-Code Divergence

Medium
Category
Not specified by scanner
Confidence
91% confidence
Finding

L049 states that literature search does not return gas entropy and that parameter-builder must recalculate it, but L260 says gas entropy can be calculated via formula or extracted. That is an intent-level contradiction about whether this skill should ever retrieve or provide gas entropy.

Content

No source excerpt is available for this finding.

Intent-Code Divergence

Medium
Category
Not specified by scanner
Confidence
97% confidence
Finding

The language-routing block is internally inconsistent: it tells English users to read SKILL_cn.md while the current file content is Chinese and later file comments say this file is the Chinese authoritative document. Conflicting control instructions can cause the agent to follow the wrong document or skip relevant safety/operational guidance, leading to unreliable behavior and incorrect execution paths.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

Forcing English-language users to follow the Chinese file creates a mismatched locale policy without user consent and contradicts the stated bilingual structure. This can cause misunderstood instructions, wrong tool usage, or failure to present accurate disclosures and confirmations in the user's language.

Content

No source excerpt is available for this finding.

Intent-Code Divergence

Low
Category
Not specified by scanner
Confidence
95% confidence
Finding

L015 says that if the user writes in Chinese, the agent should read SKILL.md in the same directory and respond in Chinese, but this file itself is SKILL.md and later L359-L360 state the opposite mapping: SKILL.md is Chinese reference while SKILL_cn.md is this English file. These instructions cannot simultaneously be true, so the documentation actively contradicts itself about which file to use for each language.

Content

No source excerpt is available for this finding.

Missing User Warnings

Low
Category
Not specified by scanner
Confidence
87% confidence
Finding

The skill instructs the agent to download supplementary files and save them locally, but it does not require explicit user consent or notification before writing to local storage. In tool-enabled environments, silent file creation can violate user expectations, clutter persistent workspaces, and increase risk if downloaded files are later processed automatically by other components.

Content

No source excerpt is available for this finding.

Missing User Warnings

Low
Category
Not specified by scanner
Confidence
91% confidence
Finding

The skill directs the agent to download supplemental information PDFs and save them locally, but it does not require notifying the user or obtaining consent for file writes. In an agent environment, silent local persistence can create privacy, storage, and traceability issues, especially if downloaded documents contain sensitive or copyrighted material.

Content

No source excerpt is available for this finding.