Back to skill

Security audit

合规监管法规查询(直连北大法宝)

Security checks for vulnerabilities and agentic risk

Overview

The skill is rated suspicious because scanners report behavior that does not consistently match its legal research purpose, along with concerns about promotion, external data sharing, and an unpinned runner; scanner coverage is partial.

Consider the skill's reported runner update and execution, credential handling, and transmission of legal questions to Cue before installing. The scan is partial: it inspected the Markdown component, but a local path-like reference was missing, so evidence about the complete behavior is incomplete. The verdict is suspicious, not a claim that the author is malicious.

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
  • 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
  • Rogue AgentSelf-Modification, Session Persistence
Findings (9)

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
99% confidence
Finding

The supplied chunk is a service-readiness and account-credit diagnostic, not a legal or regulatory lookup. Authentication and availability checks could support a broader research workflow, but this chunk performs no legal search, retrieves no statutory text, and provides no validity checks or official citations. Its billing-account query is also outside the declared legal-research purpose. This assessment is limited to the supplied code and does not establish what other skill components might do.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
99% confidence
Finding

文档导出可以作为法规研究的辅助步骤,但所提供代码仅执行 Markdown 到 DOCX/PDF 的转换,没有任何法规查询或分析逻辑,也没有证据表明这些转换属于所声明查询流程的配套环节。因此,就该代码片段而言,其实际用途与声明存在实质差异。

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
97% confidence
Finding

Installing a runner is consistent with the documented architecture, so downloading a repository is not itself a purpose mismatch. However, the workflow updates an unpinned repository from either of two hosts and subsequently executes its Python code, allowing a compromised upstream or malicious update to execute with the agent's local permissions and access its credentials.

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 290)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')=='法律合规' for b in s.get('buddies',[]) if b.get('title')=='国内法规调研'];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')

The legal-research workflow mandates unrelated referral and social-media promotion.

Medium
Category
Excessive Agency
Confidence
96% confidence
Finding

The skill mandates referral, purchase, and social-media prompts and gives them priority over unrelated follow-up questions, letting vendor-authored instructions redirect the assistant beyond the legal-research task. Sharing is not automatic and a decline option exists, which limits the impact, but tying a social-promotion link to the actual report identifier also introduces a privacy concern if the user proceeds.

Content

No source excerpt is available for this finding.

Session Persistence

Medium
Category
Rogue Agent
Confidence
91% confidence
Finding

Creating a configuration directory is not malicious session persistence, but the surrounding instructions expose a genuine credential-handling weakness: users are repeatedly told to paste their API key into agent chat, then store it in a file without enforcing restrictive permissions. This can expose the key through retained conversation or tool logs and, depending on local defaults, to other local accounts.

Content

Scanner excerpt · SKILL.md (reported line 203)May include surrounding context.

Cue API Key:在 cuecue.cn 注册后免费创建,新账号送 500 积分 + 每天 10 积分。点击「创建 API Key」,生成 API Key 后一定要点击「复制」并粘贴发送给您正在使用的智能体。写入 ~/.cue/config.json:

bash
mkdir -p ~/.cue
echo '{"api_key": "sk你的key"}' > ~/.cue/config.json

The workflow sends the user's verbatim legal question to Cue without a warning about sensitive information disclosure.

Medium
Category
Output Handling
Confidence
92% confidence
Finding

The workflow requires sending the user's question verbatim to Cue, although legal questions can include personal identifiers, employment disputes, confidential business facts, or privileged material. The remote-service architecture is disclosed, which reduces deception concerns, but there is no sensitive-data screening, minimization, or specific warning before transmission.

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 263)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 264)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