This report was generated in Chinese. Some content may be in Chinese.
Low Risk — Risk Score 18/100
Last scan:4 hr ago Rescan
18 /100
agi-evolution-model
基于双环架构的AGI进化模型,通过意向性分析、人格层映射、元认知检测和错误智慧库实现持续自我演进
AGI进化模型技能,功能为认知架构与人格系统,无恶意行为;CLI工具声明充分,但环境变量暴露面偏宽且有eval()风险,构成轻微安全隐患。
Skill Nameagi-evolution-model
Duration76.4s
Enginepi
ClawHub AGI数字伙伴 v1.0.2 by kiwifruit13
📥 209
ClawHub Verdict Suspicious dynamic_code_executionllm_suspiciousvt_suspicious
Safe to install
建议对action_env()增加更严格的凭证关键词过滤(如OPENAI_KEY、GITHUB_TOKEN、AWS_SECRET等),并限制calculator工具的eval()使用白名单表达式或迁移至ast.literal_eval。

Findings 3 items

Severity Finding Location
Low
环境变量凭证过滤不足 Sensitive Access
action_env()函数读取全部os.environ,但敏感词过滤仅覆盖PASSWORD/TOKEN/SECRET/KEY/AUTH,无法识别OPENAI_KEY、GITHUB_TOKEN、AWS_SECRET_ACCESS_KEY、STRIPE_SECRET等常见凭证格式
env_vars = dict(os.environ)
sensitive_keys = ['PASSWORD', 'TOKEN', 'SECRET', 'KEY', 'AUTH']
→ 扩展敏感词列表:['PASSWORD','TOKEN','SECRET','KEY','AUTH','OPENAI','GITHUB','AWS','STRIPE','SLACK','DATABASE_URL'],或使用正则匹配常见凭证模式
scripts/cli_system_info.py:306
Low
calculator工具使用eval()存在代码执行风险 RCE
perception_node.py的calculator工具对用户输入的表达式直接使用Python eval()执行,理论上可执行任意Python代码
result = eval(expression)
→ 使用ast.literal_eval()限制为安全表达式,或实现表达式白名单(数字、运算符、括号)
scripts/perception_node.py:530
Info
CLI工具箱安全策略声明宽松 Priv Escalation
cli-tools-guide.md明确说明'采用宽松的安全策略,赋予智能体较大的操作权限',虽配有危险命令黑名单,但无白名单机制
⚠️ **注意**:CLI工具采用宽松的安全策略,赋予智能体较大的操作权限,建议在受控环境中使用。
→ 考虑补充明确的权限范围声明,说明工具仅在./agi_memory目录下操作,避免模糊性
references/cli-tools-guide.md:35
ResourceDeclaredInferredStatusEvidence
Filesystem WRITE WRITE ✓ Aligned cli_file_operations.py 完整读写删除操作,SKILL.md已声明
Shell WRITE WRITE ✓ Aligned cli_executor.py:103 shell=True,SKILL.md已声明CLI工具箱
Environment READ READ ✓ Aligned cli_system_info.py:306 os.environ,SKILL.md在CLI工具箱中声明
Network NONE NONE 全代码库无网络请求调用
Database NONE NONE 全代码库无数据库操作
4 Critical 16 findings
💀
Critical Dangerous Command 危险 Shell 命令
rm -rf /
references/cli-tools-guide.md:326
💀
Critical Dangerous Command 危险 Shell 命令
wget xxx | sh
references/cli-tools-guide.md:332
💀
Critical Dangerous Command 危险 Shell 命令
curl xxx | sh
references/cli-tools-guide.md:332
💀
Critical Dangerous Command 危险 Shell 命令
rm -rf ~
references/troubleshooting.md:136
🔗
Medium External URL 外部 URL
https://en.wikipedia.org/wiki/Tf%E2%80%93idf
references/cognitive-insight-v2-implementation.md:1220
🔗
Medium External URL 外部 URL
https://en.wikipedia.org/wiki/Cache_replacement_policies#LRU
references/cognitive-insight-v2-implementation.md:1221
🔗
Medium External URL 外部 URL
https://en.wikipedia.org/wiki/Moving_average#Exponential_moving_average
references/cognitive-insight-v2-implementation.md:1222
🔗
Medium External URL 外部 URL
https://www.gnu.org/licenses/agpl-3.0.html
references/cognitive-insight-v2-implementation.md:1223
🔗
Medium External URL 外部 URL
https://storage.example.com/reports/report.pdf?signature=...
references/tool_use_spec.md:625
🔗
Medium External URL 外部 URL
https://docs.example.com/migration/weather-v1-to-v2
references/tool_use_spec.md:956
🔗
Medium External URL 外部 URL
https://platform.openai.com/docs/guides/function-calling
references/tool_use_spec.md:2137
🔗
Medium External URL 外部 URL
https://docs.anthropic.com/en/docs/tool-use
references/tool_use_spec.md:2138
🔗
Medium External URL 外部 URL
https://modelcontextprotocol.io/
references/tool_use_spec.md:2139
🔗
Medium External URL 外部 URL
https://json-schema.org/
references/tool_use_spec.md:2140
🔗
Medium External URL 外部 URL
https://html.spec.whatwg.org/multipage/server-sent-events.html
references/tool_use_spec.md:2141
🔗
Medium External URL 外部 URL
https://www.python.org/downloads/
references/troubleshooting.md:117

File Tree

55 files · 935.5 KB · 28170 lines
Python 30f · 15979L Markdown 24f · 12068L JSON 1f · 123L
├─ 📁 assets
│ └─ 📋 personality_template.json JSON 123L · 4.1 KB
├─ 📁 references
│ ├─ 📝 architecture.md Markdown 1327L · 50.0 KB
│ ├─ 📝 async-migration-progress.md Markdown 261L · 6.8 KB
│ ├─ 📝 c_extension_usage.md Markdown 128L · 3.0 KB
│ ├─ 📝 capability_boundaries.md Markdown 118L · 4.5 KB
│ ├─ 📝 cli-tools-guide.md Markdown 388L · 10.9 KB
│ ├─ 📝 cognitive-architecture-insight-module.md Markdown 330L · 11.9 KB
│ ├─ 📝 cognitive-insight-quick-reference.md Markdown 227L · 7.0 KB
│ ├─ 📝 cognitive-insight-v2-implementation.md Markdown 1227L · 32.7 KB
│ ├─ 📝 error_wisdom_spec.md Markdown 1167L · 31.3 KB
│ ├─ 📝 information-flow-main-loop.md Markdown 358L · 20.4 KB
│ ├─ 📝 information-flow-overview.md Markdown 301L · 12.4 KB
│ ├─ 📝 information-flow-secondary-loop.md Markdown 467L · 20.4 KB
│ ├─ 📝 init_dialogue_optimized_guide.md Markdown 371L · 11.1 KB
│ ├─ 📝 intelligence-agent-response-rules.md Markdown 103L · 3.2 KB
│ ├─ 📝 intentionality_architecture.md Markdown 563L · 16.2 KB
│ ├─ 📝 maslow_needs.md Markdown 160L · 4.4 KB
│ ├─ 📝 metacognition-check-component.md Markdown 610L · 30.2 KB
│ ├─ 📝 metacognition-enhancement-guide.md Markdown 492L · 12.6 KB
│ ├─ 📝 personality_mapping.md Markdown 161L · 5.6 KB
│ ├─ 📝 stratified-storage-design.md Markdown 323L · 10.6 KB
│ ├─ 📝 tool_use_spec.md Markdown 2156L · 56.3 KB
│ ├─ 📝 troubleshooting.md Markdown 207L · 5.0 KB
│ └─ 📝 usage-examples.md Markdown 270L · 7.3 KB
├─ 📁 scripts
│ ├─ 🐍 advice_pool.py Python 469L · 14.9 KB
│ ├─ 🐍 cli_executor.py Python 197L · 6.3 KB
│ ├─ 🐍 cli_file_operations.py Python 337L · 11.0 KB
│ ├─ 🐍 cli_process_manager.py Python 391L · 13.5 KB
│ ├─ 🐍 cli_system_info.py Python 380L · 13.2 KB
│ ├─ 🐍 cognitive_error_analyzer.py Python 755L · 26.8 KB
│ ├─ 🐍 cognitive_insight.py Python 584L · 21.3 KB
│ ├─ 🐍 concept_extraction_extension.py Python 761L · 25.3 KB
│ ├─ 🐍 error_wisdom_manager.py Python 905L · 31.4 KB
│ ├─ 🐍 error_wisdom_prevention.py Python 618L · 20.2 KB
│ ├─ 🐍 error_wisdom_rule_generator.py Python 761L · 25.3 KB
│ ├─ 🐍 error_wisdom_timeliness.py Python 591L · 21.2 KB
│ ├─ 🐍 history_manager.py Python 229L · 7.4 KB
│ ├─ 🐍 init_dialogue_optimized.py Python 464L · 16.4 KB
│ ├─ 🐍 intentionality_analyzer.py Python 366L · 12.8 KB
│ ├─ 🐍 intentionality_classifier.py Python 388L · 13.8 KB
│ ├─ 🐍 intentionality_collector.py Python 308L · 9.7 KB
│ ├─ 🐍 intentionality_daemon.py Python 406L · 14.3 KB
│ ├─ 🐍 intentionality_regulator.py Python 438L · 15.1 KB
│ ├─ 🐍 intentionality_trigger.py Python 592L · 19.9 KB
│ ├─ 🐍 memory_store_async.py Python 642L · 20.6 KB
│ ├─ 🐍 memory_store_pure.py Python 699L · 21.8 KB
│ ├─ 🐍 metacognition_history.py Python 502L · 16.3 KB
│ ├─ 🐍 objectivity_evaluator.py Python 585L · 21.2 KB
│ ├─ 🐍 perception_node.py Python 1041L · 34.8 KB
│ ├─ 🐍 personality_core_pure.py Python 144L · 4.4 KB
│ ├─ 🐍 personality_customizer.py Python 675L · 24.0 KB
│ ├─ 🐍 personality_layer_pure.py Python 951L · 32.2 KB
│ ├─ 🐍 strategy_selector.py Python 468L · 15.5 KB
│ └─ 🐍 transcendence_keeper.py Python 332L · 11.7 KB
└─ 📝 SKILL.md Markdown 353L · 15.4 KB

Dependencies 1 items

PackageVersionSourceKnown VulnsNotes
aiofiles >=23.0.0 pip No 异步文件操作,仅用于Phase 0异步化,无已知漏洞

Security Positives

✓ 无网络外传行为:全代码库未发现任何HTTP请求、POST发送或C2通信模式
✓ 无凭证窃取行为:代码不包含遍历敏感路径(~/.ssh、~/.aws、.env)或凭证外传逻辑
✓ 危险命令黑名单有效:cli_executor.py内置危险命令检测,可阻止rm -rf /、wget|sh等高危操作
✓ 文档-行为基本一致:SKILL.md声明了CLI工具箱全部能力,代码无未声明的阴影功能
✓ IOC为文档引用非实际执行:预扫描标记的4个CRITICAL IOCs均出自references/*.md说明文档,非实际可执行代码
✓ 标准库依赖安全:仅依赖aiofiles,无三方依赖引入供应链风险