扫描报告
5 /100
vmware-aiops
AI-powered VMware/vSphere/ESXi VM lifecycle management skill — power, clone, deploy, migrate, guest exec, cluster management, and alarm acknowledgment
Documentation-only skill that wraps the legitimate vmware-aiops CLI/PyPI package for VMware VM lifecycle management. No malicious behavior, hidden scripts, or credential exfiltration detected.
可以安装
This skill is safe to use. Install from PyPI (`uv tool install vmware-aiops`), review source at github.com/zw008/VMware-AIops, and ensure `.env` credentials are properly protected with `chmod 600`.
安全发现 2 项
| 严重性 | 安全发现 | 位置 |
|---|---|---|
| 提示 | Documentation-only repository | SKILL.md:1 |
| 提示 | Audit log path discrepancy | references/setup-guide.md:1 |
| 资源类型 | 声明权限 | 推断权限 | 状态 | 证据 |
|---|---|---|---|---|
| 文件系统 | NONE | NONE | — | No file operations in SKILL.md; actual filesystem access depends on installed vm… |
| 网络访问 | NONE | READ | ✓ 一致 | Webhooks for notifications only to user-configured endpoints (Slack/Discord); di… |
| 命令执行 | WRITE | WRITE | ✓ 一致 | Bash tool declared in allowed-tools; used for CLI commands (uv install, vmware-a… |
| 环境变量 | READ | READ | ✓ 一致 | Reads VMWARE_AIOPS_CONFIG, VSPHERE_USER, VSPHERE_PASSWORD from environment |
| 技能调用 | NONE | READ | ✓ 一致 | Skill routing to companion skills declared in SKILL.md (vmware-monitor, vmware-s… |
目录结构
5 文件 · 27.3 KB · 569 行 Markdown 4f · 520L
JSON 1f · 49L
├─
▾
evals
│ └─
evals.json
JSON
├─
▾
references
│ ├─
capabilities.md
Markdown
│ ├─
cli-reference.md
Markdown
│ └─
setup-guide.md
Markdown
└─
SKILL.md
Markdown
依赖分析 2 项
| 包名 | 版本 | 来源 | 已知漏洞 | 备注 |
|---|---|---|---|---|
vmware-aiops | unpinned | PyPI | 否 | Recommend pinning version: uv tool install vmware-aiops==X.Y.Z |
pyVmomi | implicit | vmware-aiops dep | 否 | Official VMware Python SDK |
安全亮点
✓ Open source (MIT license) with publicly reviewable source at github.com/zw008/VMware-AIops
✓ Security best practices documented: chmod 600 for .env files, least-privilege service account recommendation
✓ Prompt injection protection: vSphere content wrapped in boundary markers before LLM output
✓ Double confirmation required for all destructive operations (delete, power-off, etc.)
✓ TTL auto-cleanup prevents orphaned VMs
✓ Webhook notifications disabled by default, only send to user-controlled endpoints
✓ SSL/TLS verification enabled by default with clear warning against disabling in production
✓ Comprehensive audit trail with before/after state logging
✓ Input validation for VM names, CPU, memory, and disk parameters
✓ Builds on established libraries (pyVmomi, Click, APScheduler)
✓ No base64 encoding, no direct IP connections, no eval() usage detected
✓ No access to sensitive paths like ~/.ssh, ~/.aws, or .env outside the designated ~/.vmware-aiops/