Skill flagged — suspicious patterns detected
ClawHub Security flagged this skill as suspicious. Review the scan results before using.
content-factory
v1.3.0Create complete WeChat Official Account viral articles from a user-provided title by researching high-view YouTube videos, confirming topic/outline with user...
⭐ 0· 46·0 current·0 all-time
MIT-0
Download zip
LicenseMIT-0 · Free to use, modify, and redistribute. No attribution required.
Security Scan
Capability signals
These labels describe what authority the skill may exercise. They are separate from suspicious or malicious moderation verdicts.
OpenClaw
Suspicious
medium confidencePurpose & Capability
The declared goal—research YouTube and produce WeChat-ready articles (MD+HTML), generate covers, and optionally publish—is consistent with the bundled scripts (yt_dlp_*, generate_cover_photo.py, wechat_publish.py). However the registry metadata claims no required env vars/credentials while the package docs and scripts clearly expect a GLM_API_KEY (Zhipu/GLM-Image) and optionally WECHAT_APP_ID/WECHAT_APP_SECRET. That mismatch between declared requirements and actual code is an incoherence.
Instruction Scope
SKILL.md mandates running local binaries and scripts and explicitly instructs the agent to read credential files in global host paths (e.g. /root/.openclaw/credentials/tavily.json, /root/.openclaw/credentials/brave.json, ~/.claude/skills, /root/.openclaw/workspace/scripts). It also requires checking yt-dlp and executing local smart_search.py and yt_dlp scripts. These instructions cause the skill to access system-wide files and other workspace tools beyond the skill folder and could expose unrelated credentials or data.
Install Mechanism
There is no install spec (instruction-only), which reduces direct disk-write risk, but the package includes many executable Python scripts that the skill expects to run. There are no remote download/install steps in the manifest, which is positive, but executing bundled scripts still runs code on the host and must be reviewed.
Credentials
Registry metadata lists no required env vars, yet multiple docs and scripts require a GLM_API_KEY for cover generation and optionally WECHAT_APP_ID/WECHAT_APP_SECRET for publishing. SKILL.md also instructs reading other global credential files (Tavily/Brave) from /root/.openclaw/credentials, which are unrelated to the published metadata and give the skill potential access to unrelated service keys. This is disproportionate and undocumented.
Persistence & Privilege
The skill doesn't set always:true and doesn't claim to persist system-wide changes, but its runtime instructions explicitly reference and read workspace/global credential paths belonging to the agent environment (/root/.openclaw and ~/.claude). Accessing other skills' or platform-level credential files increases privilege/impact even without always:true.
What to consider before installing
Before installing or running this skill, consider the following: 1) Do not trust the package solely because the registry lists no env vars — the code and docs expect GLM_API_KEY and optionally WeChat credentials. Ask the publisher to update the registry metadata to list required env vars and explain why each is needed. 2) Inspect the included Python scripts (especially yt_dlp_*, generate_cover_photo.py, wechat_publish.py, and any file that reads /root or ~/.claude paths) to confirm what files they read and what network endpoints they contact (e.g., open.bigmodel.cn). 3) Avoid storing global or high-value credentials (platform keys, production WeChat AppSecret, or other skills' keys) in the locations the skill will read; instead use a disposable/test API key if you want to try it. 4) Run the skill in an isolated environment (container, VM, or restricted user) and audit outbound network activity before allowing it access to real secrets. 5) If you need automated publishing, consider providing only a dedicated WeChat account with limited permissions. 6) Ask the author to remove or document reads of global credential paths (Tavily/Brave paths and /root/.openclaw/workspace scripts) or to make them opt-in and configurable (skill-local .env only). If the author can (a) declare the needed env vars in registry metadata, (b) confine credential access to the skill's own .env or credential file, and (c) document all external endpoints the scripts call, the package would be materially safer. If you cannot perform these checks, treat the skill as untrusted and do not run it with real/prod credentials.Like a lobster shell, security has layers — review code before you run it.
latestvk97dah5p9szenk2v01b1e8gn6584g0dp
License
MIT-0
Free to use, modify, and redistribute. No attribution required.
