Skill flagged — suspicious patterns detected

ClawHub Security flagged this skill as suspicious. Review the scan results before using.

Memo

v1.2.0

这个技能应在用户需要记录工作事项、查询历史记录、生成工作统计报告或管理待办事项时使用。支持口语化输入,数据持久化存储在本地 JSON 文件中,实现长期记忆。

0· 104·0 current·0 all-time
MIT-0
Download zip
LicenseMIT-0 · Free to use, modify, and redistribute. No attribution required.
Security Scan
VirusTotalVirusTotal
Benign
View report →
OpenClawOpenClaw
Suspicious
medium confidence
!
Purpose & Capability
Name/description claim a local work-record assistant — that matches the included main.py which reads/writes a local records.json. However SKILL.md points to a specific data file path 'D:/华为云盘/records.json' and describes creating/updating external '自动化任务' (reminders). The main.py shown uses a records.json co-located with the script (BASE_DIR) and contains no code for creating/updating external automation tasks or interacting with external services. This mismatch between what the skill instructs and what the code actually does is incoherent.
!
Instruction Scope
SKILL.md instructs the agent to read/write a specific absolute path (D:/华为云盘/records.json), to always read records.json before answering, and to create/update/delete '自动化' reminders with rrule and fixed cwds 'D:\华为云盘'. The included main.py operates on a local records.json in the skill directory and (in the visible portion) does not implement automation creation/deletion or any external scheduler API calls. Because SKILL.md gives the agent authority to access a user-specific absolute path and to manage automation tasks, this discrepancy is a privacy/behavior risk unless clarified. Also SKILL.md contains some operational directives (e.g., 'must read records.json before answering') that, if followed against other paths, could cause the agent to access user files unintentionally.
Install Mechanism
No install spec; this is an instruction + code bundle that performs local file I/O. No downloads or external package installs are declared.
Credentials
The skill requests no environment variables or external credentials (proportional). However SKILL.md hardcodes use of a user-specific path (D:/华为云盘) and a working directory cwds=D:\华为云盘 for automation; hardcoded user-paths can lead to unexpected access to user data. Also the skill metadata/headers include a contact phone number (not harmful but unusual).
Persistence & Privilege
always is false and the skill does not request elevated platform privileges. It persists data locally (records.json) which is expected for a memo tool; nothing in the package requests forced/global persistence or modifies other skills.
What to consider before installing
This skill appears to implement a local records.json-based memo/todo helper, but there are important inconsistencies you should resolve before installing or enabling it autonomously: - File-path mismatch: SKILL.md tells the agent to use D:/华为云盘/records.json, while main.py uses a records.json in the skill directory (BASE_DIR). Confirm which path will actually be read/written. If the agent follows SKILL.md it could access your D: drive/cloud-sync folder — verify you are comfortable with that. - Automation claims vs implementation: SKILL.md describes creating/updating/deleting '自动化' reminders with rrule and a fixed cwds. The visible main.py does not implement these actions. Ask the developer which component manages automations; absence of implementation could mean the agent will be instructed to call external tools or manually perform actions (which may touch other systems). - Inspect the full main.py: the provided main.py was truncated in the listing. Review the complete file for any network calls, hidden endpoints, or code that would send data externally. The visible portions perform only local JSON read/write and Markdown export — which is expected — but you must confirm there's no hidden exfiltration in the remainder. - Privacy: records.json may contain sensitive workplace data. Decide where the file should live (skill directory vs your cloud drive) and restrict its permissions. Consider encrypting or excluding highly sensitive items. - Version/metadata inconsistencies: SKILL.md, skill.yaml, and registry version numbers/authors differ slightly; this suggests the package may have been edited. Prefer a version from a known source or ask the author to clarify. If you want to proceed safely: (1) ask the author to clarify and fix the path/automation inconsistencies, (2) test the skill in an isolated environment with a disposable records.json, and (3) review the complete main.py for any network or subprocess usage before granting it access to real data.

Like a lobster shell, security has layers — review code before you run it.

latestvk97fdnx0pdj7xnfmd8z86yxbjh83a7ca

License

MIT-0
Free to use, modify, and redistribute. No attribution required.

Comments