Skill flagged — suspicious patterns detected
ClawHub Security flagged this skill as suspicious. Review the scan results before using.
股票列表自动管理
v1.0.0从图片自动识别股票并加入股票池。触发条件：用户发送股票截图/图片并要求加入股票池。功能：(1) 使用RapidOCR识别图片中的股票代码、名称、涨幅、价格 (2) 自动去重 (3) FIFO原则管理30只股票上限 (4) 保存到manual_stock_list.json。
⭐ 0· 36·0 current·0 all-time
by@wy859
MIT-0
Download zip
LicenseMIT-0 · Free to use, modify, and redistribute. No attribution required.
Security Scan
OpenClaw
Suspicious
medium confidencePurpose & Capability
Name/description (parse stock screenshots and maintain a 30-item FIFO list) aligns with the included scripts: both scripts use RapidOCR/OpenCV to read images, parse 6-digit codes, de-duplicate, enforce FIFO, and save JSON.
Instruction Scope
SKILL.md and scripts instruct the agent to read images from an absolute path (/Users/wy/.openclaw/media/inbound/{uuid}.jpg) and write to a hard-coded path (/Users/wy/.openclaw/workspace-changniu/stone_quant/manual_stock_list.json). That direct access to a user home path tied to username 'wy' is brittle and could read or overwrite unexpected files if the runtime differs. The instructions also assume the RapidOCR runtime and OpenCV are available but do not declare or install them.
Install Mechanism
No install spec provided. The code imports third-party packages (rapidocr_onnxruntime, cv2) but the skill does not declare required binaries or dependency installation steps. This is a mismatch (the code will fail unless the environment already has these packages).
Credentials
The skill requests no credentials or env vars, which is proportionate. However it writes to and reads from absolute paths under a specific user's home — this is not secret-leakage but is an access-scope concern (it will modify files in that path). No network exfiltration or external endpoints are present in the code.
Persistence & Privilege
always is false and the skill does not attempt to modify other skills or global agent configuration. It only writes its own JSON file in a workspace path; that is expected for this functionality.
What to consider before installing
What to check before installing:
- Confirm the hard-coded paths are intended: both scripts use /Users/wy/.openclaw/… which is a specific user directory. If your environment uses a different user or a container, the skill may fail or overwrite files unexpectedly. Ask the author to make the media and workspace paths configurable.
- Ensure dependencies (rapidocr_onnxruntime and OpenCV / python-opencv) are available in your runtime. The package/dependency installation is not declared; the skill will fail without them.
- Review the files the skill writes to (manual_stock_list.json) and back them up if needed. Test in a sandbox or with dummy files first.
- Confirm privacy expectations: the skill will read inbound images and persist recognized stock codes and names locally. If inbound images contain other sensitive content, be aware it will be processed.
- If you want to reduce risk, request the author to: (1) replace absolute paths with configurable environment variables or relative workspace paths, (2) add an install spec or dependency list, and (3) include explicit error handling/logging and a dry-run mode. If these changes are made, re-evaluate; as-is the issues look like sloppy engineering rather than overt malicious behavior, but they warrant caution.Like a lobster shell, security has layers — review code before you run it.
latestvk977myhe4f1qtnt62eyj41wy1d84rj59
License
MIT-0
Free to use, modify, and redistribute. No attribution required.