中文 EN

扫描报告

扫描新文件

✓ 可信 — 风险评分 5/100
上次扫描：21 小时前 重新扫描
5 /100
self-evolving-agent
Build a goal-driven self-learning loop for OpenClaw and coding agents
This is a legitimate self-evolving-agent skill for AI agent capability improvement. All scripts serve documented purposes with no malicious patterns detected.
技能名称self-evolving-agent
分析耗时31.8s
引擎pi
✓
可以安装
This skill is safe to use. No security concerns identified.
资源类型声明权限推断权限状态证据
文件系统 NONE READ ✓ 一致 Scripts read/write workspace files in ~/.openclaw/ - documented in SKILL.md
命令执行 NONE READ ✓ 一致 subprocess.run in run-benchmark.py for 'codex exec' - benchmark tooling
网络访问 NONE NONE — No network requests found in code
8 项发现
🔗
中危 外部 URL 外部 URL
https://img.shields.io/badge/Language-English-0A7CFF?style=flat-square
README.md:4
🔗
中危 外部 URL 外部 URL
https://img.shields.io/badge/%E8%AF%AD%E8%A8%80-%E7%AE%80%E4%BD%93%E4%B8%AD%E6%96%87-16A34A?style=flat-square
README.md:5
🔗
中危 外部 URL 外部 URL
https://img.shields.io/badge/OpenClaw-Skill-111827?style=flat-square
README.md:7
🔗
中危 外部 URL 外部 URL
https://img.shields.io/github/actions/workflow/status/RangeKing/self-evolving-agent/ci.yml?branch=main&style=flat-square...
README.md:8
🔗
中危 外部 URL 外部 URL
https://img.shields.io/github/license/RangeKing/self-evolving-agent?style=flat-square
README.md:9
🔗
中危 外部 URL 外部 URL
https://img.shields.io/github/stars/RangeKing/self-evolving-agent?style=flat-square
README.md:10
🔗
中危 外部 URL 外部 URL
https://img.shields.io/badge/Benchmark-Model--in--the--Loop-7C3AED?style=flat-square
README.md:11
🔗
中危 外部 URL 外部 URL
https://img.shields.io/badge/Agent-Goal--Driven%20Learning-0F766E?style=flat-square
README.md:12

目录结构

39 文件 · 141.8 KB · 4641 行
Markdown 28f · 3327L Python 3f · 932L JSON 3f · 202L Shell 3f · 153L TypeScript 1f · 23L YAML 1f · 4L
├─ ▾ 📁 agents
│ └─ 📋 openai.yaml YAML 4L · 318 B
├─ ▾ 📁 assets
│ ├─ 📝 CAPABILITIES.md Markdown 324L · 8.7 KB
│ ├─ 📝 ERRORS.md Markdown 42L · 896 B
│ ├─ 📝 EVALUATIONS.md Markdown 41L · 846 B
│ ├─ 📝 FEATURE_REQUESTS.md Markdown 32L · 869 B
│ ├─ 📝 LEARNING_AGENDA.md Markdown 48L · 2.2 KB
│ ├─ 📝 LEARNINGS.md Markdown 48L · 1.2 KB
│ └─ 📝 TRAINING_UNITS.md Markdown 46L · 956 B
├─ ▾ 📁 benchmarks
│ ├─ ▾ 📁 schemas
│ │ └─ 📋 judge-output.schema.json JSON 56L · 993 B
│ └─ 📋 suite.json JSON 117L · 5.0 KB
├─ ▾ 📁 demos
│ ├─ 📝 demo-1-diagnosis.md Markdown 83L · 2.7 KB
│ ├─ 📝 demo-2-training-loop.md Markdown 86L · 2.6 KB
│ ├─ 📝 demo-3-promotion-and-transfer.md Markdown 79L · 2.6 KB
│ ├─ 📝 demo-4-agenda-review.md Markdown 62L · 2.4 KB
│ └─ 📝 demo-5-pre-task-risk-diagnosis.md Markdown 51L · 2.3 KB
├─ ▾ 📁 evals
│ └─ 📋 evals.json JSON 29L · 2.3 KB
├─ ▾ 📁 hooks
│ └─ ▾ 📁 openclaw
│ ├─ 📜 handler.ts TypeScript 23L · 939 B
│ └─ 📝 HOOK.md Markdown 28L · 983 B
├─ ▾ 📁 modules
│ ├─ 📝 capability-map.md Markdown 173L · 3.8 KB
│ ├─ 📝 curriculum.md Markdown 114L · 2.8 KB
│ ├─ 📝 diagnose.md Markdown 152L · 3.9 KB
│ ├─ 📝 evaluator.md Markdown 121L · 2.7 KB
│ ├─ 📝 learning-agenda.md Markdown 116L · 3.2 KB
│ ├─ 📝 promotion.md Markdown 83L · 1.9 KB
│ └─ 📝 reflection.md Markdown 108L · 2.4 KB
├─ ▾ 📁 scripts
│ ├─ 🔧 activator.sh Shell 15L · 375 B
│ ├─ 🔧 bootstrap-workspace.sh Shell 95L · 2.3 KB
│ ├─ 🔧 error-detector.sh Shell 43L · 757 B
│ ├─ 🐍 migrate-self-improving.py Python 196L · 6.4 KB
│ ├─ 🐍 run-benchmark.py Python 351L · 12.1 KB
│ └─ 🐍 run-evals.py Python 385L · 11.6 KB
├─ ▾ 📁 system
│ └─ 📝 coordinator.md Markdown 336L · 7.7 KB
├─ 📝 CHANGELOG.md Markdown 24L · 866 B
├─ 📝 CONTRIBUTING.md Markdown 76L · 2.3 KB
├─ 📝 install.md Markdown 176L · 5.0 KB
├─ 📝 README.md Markdown 317L · 12.5 KB
├─ 📝 README.zh-CN.md Markdown 319L · 12.2 KB
├─ 📝 SECURITY.md Markdown 29L · 881 B
└─ 📝 SKILL.md Markdown 213L · 7.4 KB

安全亮点

✓ Clean codebase with no obfuscation or base64-encoded payloads
✓ Subprocess usage is limited to legitimate benchmark tooling (codex exec)
✓ All scripts are well-documented with clear purposes
✓ No credential harvesting or environment variable iteration
✓ No sensitive path access (~/.ssh, ~/.aws, .env)
✓ No data exfiltration or C2 communication patterns
✓ SKILL.md accurately describes all functionality
✓ No external dependencies with unpinned versions (uses Python stdlib only)