100Critical
This skill is vulnerable to prompt
Behavior Analysis
Description verified — behavior matches claim
Claims to do
SKILL.md — 腾讯微云管理 Skills 定义: > **使用方法**：本文档定义了所有可用的腾讯微云管理 Skills。AI Agent 或开发者可根据此文档调用 Python 脚本完成云存储操作。 > > **认证方式**（二选一）： > ```bash > # Method 1: QR code login (recommended) > python weiyun_skills/login.py --method qrcode > > # Method 2: Copy cookies from browser > python weiyun_skills/login.py --method cookies --cookies "uin=o012345678; skey=@abcdef1234; ..." > ``` > > **调用方式**： > ```bash > # CLI > python weiyun_skills/main.py <command> [args] [options] > > # Python SDK > from weiyun_skills.client import WeiyunClient > client = WeiyunClient() > client.<skill_name>(**params) > ``` > > **统一返回格式**： > ```json > { "success": true, "data": { ... }, "message": "ok" } > ```
Actually does
This skill executes Python scripts (`weiyun_skills/login.py`, `weiyun_skills/main.py`) to interact with Tencent Weiyun cloud storage. It performs authentication via QR code scanning or user-provided cookies, storing session data in `cookies.json`. The skill manages files (list, upload, download, delete, move, copy, rename, create folder, search), shares (create, cancel, list), and space (info, recycle bin) by contacting Weiyun's services.
Skill Info
Namewscats/weiyun-skills
Registryclawhub
Version1.0.0
PURLpkg:clawhub/wscats/weiyun-skills@1.0.0
Stars3
Downloads11,329
Source
SHA-2567d5a86514ed2...
SHA-1149dde9230ff...
MD5a5b7845457a4...
TLSHcfa28520dc86...
ssdeep384:iXJZTiiL...
Size22,976 bytes
Install
This skill has been flagged as potentially malicious. Review the findings below before installing.
ClawHubDocs
openclaw skills install wscats/weiyun-skillsAssessments (2)
AI Agent Traps ↗Prompt Injection Detectedcriticalprompt injection
DeBERTa classifier detected prompt injection (confidence: 1.00)
100% confidence
Command line exposure of authentication cookiescriticalcredential theft
The skill's CLI examples demonstrate passing sensitive authentication cookies directly as a command-line argument. This exposes credentials in plain text in process lists, shell history, and potentially logs, making them vulnerable to theft.
100% confidenceline 25
python weiyun_skills/login.py --method cookies --cookies "uin=o012345678; skey=@abcdef1234; ..."
Arbitrary local file read/write via skill parametershighdata exfiltration
The `upload_file` and `download_file` skills allow specifying arbitrary `local_path` values. This capability could be abused to exfiltrate sensitive files from the agent's host or to write malicious files to critical system locations.
100% confidenceline 168
upload_file: `local_path` | `string` | ✅ | - | 本地文件路径 (line 168) download_file: `local_path` | `string` | ✅ | - | 本地保存路径 (line 197)
Permanent data deletion capabilitymediumimpact
The `delete_file` skill allows permanent deletion of files by setting `permanent=True`, and `clear_recycle_bin` permanently deletes all items. While `clear_recycle_bin` requires a `confirm=True` flag, an agent could be prompted to provide it, leading to irreversible data loss.
90% confidenceline 226
delete_file: `permanent` | `boolean` | ❌ | `false` | 是否永久删除（跳过回收站） (line 226) clear_recycle_bin: `confirm` | `boolean` | ✅ | - | 必须为 `true` 才执行 (line 504)
Potential path traversal/injection in file operationsmediumcommand execution
Multiple file management skills accept `remote_path` and `local_path` as string inputs. Without explicit sanitization or validation, these parameters could be vulnerable to path traversal attacks (e.g., `../../`) or command injection if the underlying Python implementation directly concatenates them into shell commands.
80% confidenceline 139
`remote_path` in `list_files` (line 139), `upload_file`, `download_file`, `delete_file`, `move_file`, `copy_file`, `rename_file`, `create_folder`, `create_share`.
Badge
Markdown
[](https://mondoo.com/ai-agent-security/skills/clawhub/wscats/weiyun-skills)HTML
<a href="https://mondoo.com/ai-agent-security/skills/clawhub/wscats/weiyun-skills"><img src="https://mondoo.com/ai-agent-security/api/badge/clawhub/wscats/weiyun-skills.svg" alt="Mondoo Skill Check" /></a>Image URL
https://mondoo.com/ai-agent-security/api/badge/clawhub/wscats/weiyun-skills.svgSecure your AI agents
Skills can read files, run commands, and access credentials. Mondoo helps organizations manage the security risks of AI agent skills across their entire fleet.
- Continuous scanning of skills across all registries
- Policy enforcement before skills reach your agents
- Integration with your existing security workflow