Back to skill

Security audit

新兴产业研究

Security checks for vulnerabilities and agentic risk

Overview

The scanners mark this skill suspicious because it asks users to share an API key in chat and describes an industry research tool whose reported behavior also installs a runner and converts documents; the review is incomplete because a referenced local path was missing.

Consider avoiding installation until the API-key instructions are changed to use local protected credential storage and the runner’s source and revision can be reviewed. If you proceed, review each setup command before running it, especially commands that use sudo, and remember that the scan is partial: one referenced local path was missing, so the evidence does not cover everything.

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
  • 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
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Output HandlingUnvalidated Output Injection, Cross-Context Output, Unbounded Output
Findings (9)

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
99% confidence
Finding

The supplied code only converts an existing report.md into report.docx and report.pdf. It contains no industry research, competitive analysis, business-model assessment, or legal-risk mapping. Document conversion could support a reporting workflow, but no such analytical workflow is present in this chunk, making its demonstrated primary purpose materially different from the description.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
98% confidence
Finding

Installing a runner is a legitimate dependency of this research workflow, so installation alone is not a description-behavior mismatch. However, the instructions clone or update an unpinned repository and subsequently execute its Python code with access to local files and Cue credentials; a compromised repository or mirror could therefore introduce arbitrary code into the agent's environment.

Content

No source excerpt is available for this finding.

The setup workflow requests an API key in chat without warning about credential exposure.

High
Category
Output Handling
Confidence
99% confidence
Finding

The onboarding instructions explicitly ask users to paste a secret API key into the agent conversation. This unnecessarily exposes the credential to conversation storage, logging, and any parties or integrations with transcript access, potentially enabling unauthorized Cue usage or access within the key's permissions.

Content

No source excerpt is available for this finding.

External Script Fetching

High
Category
Supply Chain
Confidence
90% confidence
Finding

Remote code is downloaded and executed. This bypasses code review and could introduce malicious code.

Content

Scanner excerpt · SKILL.md (reported line 230)May include surrounding context.

md
CUE_KEY=$(python3 -c "import json;print(json.load(open('$HOME/.cue/config.json'))['api_key'])" 2>/dev/null || echo "$CUE_API_KEY")
echo "=== 1/4 API Key ===" && [ -n "$CUE_KEY" ] && echo "已配置" || echo "未配置!"
echo "=== 2/4 Cue 服务 ===" && curl -sS --max-time 10 "https://cuecue.cn/api/health" -H "Authorization: Bearer $CUE_KEY"
echo "=== 3/4 搭子 ===" && curl -sS --max-time 10 "https://cuecue.cn/api/playbook" -H "Authorization: Bearer $CUE_KEY" | python3 -c "import sys,json;scenes=json.load(sys.stdin).get('data',{}).get('scenes',[]);buddy=[b for s in scenes if s.get('secondary_category')=='行业研究' for b in s.get('buddies',[]) if b.get('title')=='新兴产业研究'];print(f'可用:{len(buddy)}个') if buddy else print('暂不可用')"
echo "=== 4/4 剩余积分 ===" && python3 -c "
import json, urllib.request, os
p = os.path.expanduser('~/.cue/config.json')

Onboarding instructs the agent to solicit a secret API key through chat.

Medium
Category
Data Exfiltration
Confidence
99% confidence
Finding

This is the same credential-handling issue identified by SQP-2, not a separate exposure. Requiring an API key to pass through chat expands its exposure beyond the local authentication mechanism; the assertion that the agent cannot view the key does not protect a key explicitly submitted to it.

Content

No source excerpt is available for this finding.

Mandatory verbatim Chinese prompts impose an output language without user opt-in.

Medium
Category
Output Handling
Confidence
91% confidence
Finding

The skill requires the Chinese-language question and options to be used verbatim and prohibits changing their wording. This mandates Chinese output even for users communicating in another language, without offering a language choice or documenting a justified locale restriction.

Content

No source excerpt is available for this finding.

Sudo/Root Execution

Medium
Category
Privilege Escalation
Confidence
70% confidence
Finding

Commands invoke sudo or root privileges. Verify this elevated access is necessary and justified.

Content

Scanner excerpt · SKILL.md (reported line 207)May include surrounding context.

md
| 目标格式 | 依赖 | macOS | Ubuntu |
|----------|------|-------|--------|
| Word (.docx) | pandoc | `brew install pandoc` | `sudo apt install pandoc` |
| PDF (.pdf) | pandoc + LaTeX | `brew install --cask basictex` | `sudo apt install texlive-xetex` |

## 架构说明

Sudo/Root Execution

Medium
Category
Privilege Escalation
Confidence
70% confidence
Finding

Commands invoke sudo or root privileges. Verify this elevated access is necessary and justified.

Content

Scanner excerpt · SKILL.md (reported line 208)May include surrounding context.

md
| 目标格式 | 依赖 | macOS | Ubuntu |
|----------|------|-------|--------|
| Word (.docx) | pandoc | `brew install pandoc` | `sudo apt install pandoc` |
| PDF (.pdf) | pandoc + LaTeX | `brew install --cask basictex` | `sudo apt install texlive-xetex` |

## 架构说明

The research workflow mandates referral and social-media promotion unrelated to industry analysis.

Low
Category
Excessive Agency
Confidence
90% confidence
Finding

The manifest describes industry research and compliance-risk analysis, but L113–L124 require a post-delivery prompt promoting user referrals, purchases, and Xiaohongshu sharing when credits fall below 100. L126–L131 additionally construct a promotional sharing link using the actual report ID; referral recruitment and social-media promotion are not necessary to produce or deliver the requested research, although the workflow does not automatically publish anything.

Content

No source excerpt is available for this finding.