Skill flagged — suspicious patterns detected

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

auto-workflow

v2.0.0

自动化工作流引擎 - 将重复性任务自动化。 支持:文件处理、数据转换、定时任务、API 调用、多步骤工作流。 触发词:"自动化"、"工作流"、"批量处理"、"定时任务"、"workflow"、"automate"。 自动执行:预设工作流或自定义流程。

0· 78·0 current·0 all-time
MIT-0
Download zip
LicenseMIT-0 · Free to use, modify, and redistribute. No attribution required.
Security Scan
VirusTotalVirusTotal
Pending
View report →
OpenClawOpenClaw
Suspicious
medium confidence
Purpose & Capability
The name/description match the included workflow engine code (file ops, HTTP, archiving, scheduling/cron-style examples). However the docs/examples mention cloud uploads (upload.cos/upload.cdn) and automatic scheduled execution while the package declares no credentials, no cloud integrations, and no install/service to run scheduled jobs — suggesting incomplete or inconsistent capability claims.
!
Instruction Scope
SKILL.md and examples instruct running arbitrary workflows that include shell.exec, file.delete, download, and archive operations on arbitrary paths. The runtime instructions do not constrain these actions or declare the need for credentials for cloud uploads. The engine exposes shell.exec (likely runs arbitrary shell commands) and unrestricted file operations — expected for a workflow engine but high-risk if workflows are untrusted or run with elevated privileges.
Install Mechanism
There is no install spec (instruction-only), which reduces installer risk, but a Python script (scripts/workflow-engine.py) is included without an install/binary wrapper. SKILL.md refers to a CLI 'auto-workflow' command but the package provides no installer to create that command — a small coherence gap (manual setup required). No external downloads were used.
!
Credentials
The skill declares no required environment variables or credentials, yet documentation and examples reference uploading to cloud/CDN buckets. Those actions normally require credentials (API keys, cloud access). The absence of declared env vars or guidance on supplying credentials is a mismatch and could lead users to embed secrets directly in workflows or run incomplete features.
Persistence & Privilege
always:false (no forced presence) which is appropriate. The skill can be invoked autonomously (platform default), which combined with its ability to run shell commands and modify arbitrary files increases blast radius — but autonomous invocation alone is not sufficient to mark it malicious.
What to consider before installing
This package appears to be a real workflow engine, but there are several red flags to consider before installing or running it: - Review the included Python script carefully (scripts/workflow-engine.py). It supports shell.exec and unrestricted file operations; do not run workflows you don't fully trust. - The docs reference uploading to cloud/CDN and scheduled automatic runs, but no credentials or installer are declared. Ask the author how to provide credentials safely (use env vars, not hard-coded secrets) and how scheduling is implemented (background service, cron integration, or manual invocation). - Because the engine can execute shell commands and delete/move files, run it in a low-privilege, isolated environment (container or dedicated VM) during evaluation. - If you plan to use cloud upload steps, require the skill to explicitly declare which environment variables/keys it needs and prefer scoped, minimal-permission credentials. - If you need this in production, request an install script or packaging that creates the CLI and a secure mechanism for background scheduling, and ask for provenance (source repo/homepage) — the skill metadata lacks a homepage and origin. If you want, I can: (a) highlight the parts of the script that implement shell execution and network/upload behavior, (b) suggest a minimal safe workflow policy, or (c) propose an environment variable list the skill should require for cloud uploads.

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

latestvk9730k7h1pbbn32nrj6d10kv1183mgg4

License

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

Comments