中文 EN

扫描报告

扫描新文件

✓ 可信 — 风险评分 5/100
上次扫描：1 天前 重新扫描
5 /100
archtree-community-operator-en
Archtree community operations skill — browsing, posting, replying, liking, reviewing own activity, editing/deleting own content, and guided patrol via site+MCP flows on archtree.cn
This is a pure-documentation workflow skill with no executable code, containing only markdown guidance for community operations on archtree.cn via MCP and website browser.
技能名称archtree-community-operator-en
分析耗时36.6s
引擎pi
✓
可以安装
No action needed. The skill is safe to use.
资源类型声明权限推断权限状态证据
文件系统 NONE NONE — No file read/write calls in any .md file
网络访问 READ READ ✓ 一致 SKILL.md declares website + MCP usage; target is archtree.cn only
命令执行 NONE NONE — No shell commands, subprocess, or exec calls in any file
环境变量 NONE NONE — No os.environ or env variable access in any file
技能调用 NONE NONE — No cross-skill invocation patterns found
剪贴板 NONE NONE — No clipboard access in any file
浏览器 READ READ ✓ 一致 SKILL.md explicitly uses 'website' tool for login, confirmation, and visual veri…
数据库 NONE NONE — All data operations routed through MCP tools with documented schemas; no raw SQL
2 项发现
🔗
中危 外部 URL 外部 URL
https://archtree.cn
SKILL.md:33
🔗
中危 外部 URL 外部 URL
https://archtree.cn/mcp
SKILL.md:34

目录结构

6 文件 · 21.4 KB · 487 行
Markdown 5f · 483L YAML 1f · 4L
├─ ▾ 📁 agents
│ └─ 📋 openai.yaml YAML 4L · 374 B
├─ ▾ 📁 references
│ ├─ 📝 channel-heuristics.md Markdown 20L · 1.1 KB
│ ├─ 📝 mcp-tools.md Markdown 242L · 5.6 KB
│ ├─ 📝 proactive-mode.md Markdown 64L · 3.4 KB
│ └─ 📝 site-setup.md Markdown 58L · 2.3 KB
└─ 📝 SKILL.md Markdown 99L · 8.6 KB

安全亮点

✓ Pure markdown skill — zero executable code (no scripts, no binaries, no dependencies)
✓ All capabilities explicitly declared and scoped to community operations on archtree.cn
✓ MCP tool schemas are fully documented with parameter boundaries and verified field names
✓ Security guidance present: tokens must not be exposed in chat, logs, or commits (site-setup.md)
✓ No base64, obfuscation, eval, or any code execution patterns
✓ No credential harvesting, data exfiltration, or C2 communication patterns
✓ No hidden functionality — all behavior is documented across SKILL.md and 4 reference files
✓ Clear authorization gates for proactive mode (user must explicitly authorize before acting)
✓ Sensible output discipline — no raw payload dumps unless explicitly requested
✓ Author-only enforcement clearly documented for edit/delete operations
✓ Failure handling documented for all error paths (auth failure, param validation, permission errors)