Skill flagged — suspicious patterns detected

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

QQ Email with AI

v1.0.0

QQ 邮箱智能管理工具。支持收发邮件、搜索筛选、附件处理,以及 AI 智能整理功能(自动摘要、分类、优先级排序、待办提取)。当用户需要操作 QQ 邮箱、查收邮件、发送邮件、整理收件箱或处理邮件相关任务时使用此技能。

0· 171·0 current·0 all-time
bysilas@aohoyo
MIT-0
Download zip
LicenseMIT-0 · Free to use, modify, and redistribute. No attribution required.
Security Scan
VirusTotalVirusTotal
Benign
View report →
OpenClawOpenClaw
Suspicious
high confidence
Purpose & Capability
The name/description (QQ 邮箱管理 + AI 整理) matches the included scripts: IMAP/SMTP client, send/read/manage emails, and AI summarization/classification/prioritization/todo extraction. This functionality legitimately requires an email address, an IMAP/SMTP auth code and (optionally) an AI API key — but the registry metadata does not declare those env vars (see environment_proportionality).
!
Instruction Scope
Runtime instructions and scripts will fetch full email bodies, attachments, and metadata and (when AI features are used) call a third‑party Generation API (dashscope / qwen-plus) with email content. That means potentially sensitive email contents are transmitted off‑host. The SKILL.md shows commands for downloading attachments and references scripts (e.g., scripts/download_attachments.py) that are referenced but not present in the provided manifest excerpt — a discrepancy to verify. The instructions otherwise stay within the stated email-management scope (no unrelated system reads), but they do not clearly call out the privacy/telemetry implications in the main SKILL.md (only in reference docs).
Install Mechanism
This is instruction-only with bundled Python scripts and no install spec that downloads arbitrary code at install time. No brew/npm/external archive installs were specified. Risk from install mechanism is low, but the package includes executable scripts which will run locally.
!
Credentials
The code requires QQ_EMAIL and QQ_EMAIL_AUTH_CODE (IMAP/SMTP auth) and also references DASHSCOPE_API_KEY for AI calls. Those are sensitive credentials and the registry metadata lists no required env vars / primary credential — an inconsistency. DASHSCOPE usage is proportionate to the AI features, but users must understand that email text/attachments will be sent to an external AI service if that key is set/AI mode enabled. Requiring an auth code for mailbox access is expected; requiring an external AI key is optional for AI features but should be declared.
Persistence & Privilege
Skill is user‑invocable and not always-enabled; it does not request permanent platform privileges or modify other skills. It does open IMAP/SMTP sessions to user accounts (expected for an email client) but does not attempt to persist itself beyond its own scripts.
What to consider before installing
Key points to check before installing/using: - Credentials: The registry metadata did not declare required env vars, but the code and SKILL.md expect QQ_EMAIL and QQ_EMAIL_AUTH_CODE (IMAP/SMTP authorization code). AI features require DASHSCOPE_API_KEY. Only set these in an environment you control; treat the QQ auth code like a password and rotate it if exposed. - Privacy: If you enable AI features (or set DASHSCOPE_API_KEY), email subjects/bodies (and possibly extracted attachments) are sent to the DashScope/通义千问 API. Only enable this if you trust that provider and are comfortable with email contents leaving your device. - Missing/verify files: SKILL.md references a download_attachments script that is not present in the provided manifest excerpt — verify the full package contains all referenced scripts (especially any that write files or upload data). Review scripts that handle attachments and sending to ensure they behave as you expect. - Minimal use: If you only need basic send/read/manage features, run scripts without setting the DASHSCOPE_API_KEY or use the provided '--no-ai' / non‑AI options to avoid external API calls. - Audit the code: Because the skill runs locally and connects to your mailbox, inspect scripts (especially send_email, download/attachment handling, and AI call sites) yourself or run them in a restricted/test account first. Consider running in an isolated environment or on a throwaway QQ account until you are confident. - If in doubt: classify this skill as suspicious due to metadata omissions and external AI calls. Ask the publisher to update registry metadata to declare the required env vars and to explicitly state in SKILL.md when/where email contents are sent to third parties.

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

latestvk971kyc9phrx2qmpg05j0syr6x8364fc

License

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

Comments