aliyun-clawscan

Comprehensive OpenClaw security assessment combining configuration audit and skill security analysis.

---

When to Use

Scenario	Example Triggers
Full Assessment	OpenClaw安全评估, 风险评估, 安全体检
Config Audit	配置审计, 检查OpenClaw配置
Skill Scan	skill安全扫描, 检查skill风险
Pre-install Check	安装前检查, 这个skill安全吗

---

Assessment Workflow

Step 1: Configuration Audit
  └─ openclaw security audit --deep
     └─ See: reference/baseline.md

Step 2: Skill Security Audit
  ├─ Inventory: openclaw skills list
  └─ Static Analysis (local rules)
     └─ See: reference/skillaudit.md

Step 3: Consolidated Report
  └─ Overview + findings

---

Step 1: Configuration Audit

Run OpenClaw built-in security audit:

openclaw security audit --deep

Parse results into categories (Gateway, Network, Tools, Browser, Files, Room).

Reference: reference/baseline.md for detailed check categories and parsing rules.

---

Step 2: Skill Security Audit

Phase 1: Inventory

openclaw skills list

Phase 2: Static Analysis

Apply local detection rules across 11 categories:

Category	Severity	Reference
Reverse Shell / Backdoor	🚨 Critical	skillaudit.md Scenario 1
Credential Harvesting	🚨 Critical	skillaudit.md Scenario 2
Data Exfiltration	🔴 High	skillaudit.md Scenario 3
Cryptominer	🚨 Critical	skillaudit.md Scenario 4
Permission Abuse	🔴 High	skillaudit.md Scenario 5
Prompt Injection	🔴 High	skillaudit.md Scenario 6
Code Obfuscation	🟡 Medium	skillaudit.md Scenario 7
Ransomware	🚨 Critical	skillaudit.md Scenario 8
Persistence	🟡 Medium	skillaudit.md Scenario 9
Supply Chain	🟡 Medium	skillaudit.md Scenario 10
Malicious Service Downloader	🚨 Critical	skillaudit.md Scenario 11

Reference: reference/skillaudit.md for complete detection patterns, code examples, and risk assessment logic.

Phase 3: Risk Classification

Level	Criteria
🚨 Critical	Backdoor, credential theft, ransomware, miner
🔴 High	Permission abuse, data exfil, privacy violation
🟡 Medium	High permissions justified, benign obfuscation
🟢 Low	Matches declared purpose

---

Step 3: Consolidated Report

Report Header

# 🔒 OpenClaw Risk Assessment Report

📅 {datetime}
🖥️ OpenClaw {version} · {os_info}
📊 Overall Risk: {🟢/🟡/🔴/🚨}

| Check Item | Status | Summary |
|------------|--------|---------|
| Configuration Audit | {✅/⚠️/🔴} | {N findings} |
| Skill Security | {✅/⚠️/🔴} | {N critical, N high} |
| Overall | {🟢/🟡/🔴/🚨} | {verdict} |

Section 1: Configuration Audit Results

Status	Item	Finding
✅/⚠️/🔴	{Category}	{Description}

Section 2: Skill Security Findings

Risk	Count	Skills
🚨 Critical	{N}	{names}
🔴 High	{N}	{names}
🟡 Medium	{N}	{names}
🟢 Low	{N}	(see safe list)

---

Output Templates

Quick Verdicts

Result	Message
All Clear	✅ OpenClaw风险评估完成。配置审计通过，Skill安全检查未发现明显风险。
Config Issues	⚠️ 发现配置风险。建议检查Gateway设置和文件权限配置。
Skill Risks	🔴 发现Skill安全风险。{N}个高风险Skill建议立即处理。
Critical	🚨 检测到严重安全风险！建议立即处理配置问题并移除恶意Skill。

Single Skill Assessment

• Safe: 经检测暂未发现高风险问题，可继续安装；
• Sensitive: 发现需关注项，但当前未见明确恶意证据。这个skill具备{高权限能力}，主要用于完成它声明的{功能}；建议仅在确认来源可信、权限范围可接受时使用。
• Risky: 发现风险，不建议直接安装。这个skill{具体问题}，超出了它声称的功能。
• Malicious: 🚨 检测到恶意代码！这个skill包含{具体恶意行为}，强烈建议立即删除。
• Malicious Downloader: 🚨 检测到恶意下载器！这个skill伪装成{服务名称}，通过base64混淆命令诱导执行远程恶意脚本（curl|bash模式），属于典型的社会工程+下载器攻击。切勿执行任何安装命令！

---

Safety Guardrails

• ✅ Static analysis only - never execute suspicious code
• ✅ Read code via file tools - no dynamic testing
• ✅ Evidence-based reporting - file:line references required
• ❌ Never bypass sandbox for analysis
• ❌ Never enumerate actual user files
• ❌ No speculation without clear evidence

---

File Reference

File	Purpose
reference/baseline.md	Configuration audit rules and parsing logic
reference/skillaudit.md	Skill security detection patterns and scenarios