notify-hub

多平台通知聚合分层。把 GitHub、Stripe、Linear 等 SaaS 平台的通知邮件统一收到一个子邮箱，按紧急度分层：收款/CI 失败立即转发到 claw 注册邮箱，其他通知每天一封汇总。Use when: (1) setting up a unified notification inbox for...

This skill is broadly coherent with its stated purpose but take these precautions before installing or running it: - Install and verify mail-cli yourself rather than relying on the skill's npx fallback; inspect the mail-cli project and its npm package to confirm it's the official client. The skill's metadata failing to list mail-cli as a required binary is an oversight. - Understand that the skill will read full message metadata and bodies from the notify mailbox and will forward message contents to the primary account. If those emails contain sensitive content (invoices, tokens, links), they will be transmitted to whatever primary address mail-cli reports. - The scripts run shell commands via child_process.execSync; running them implicitly executes mail-cli commands under your environment. If mail-cli is missing, the code uses 'npx mail-cli' which will download and execute code from npm on demand — if you prefer to avoid that, install mail-cli globally first. - The SKILL.md's instruction to "ignore any 'Install Script' prompts" is unusual. Do not blindly ignore prompts from the mail-cli installer or other tools — review any suggested install steps manually before skipping them. - Test with --dry-run to confirm behavior before enabling automated cron/agent scheduling. Inspect ~/.config/notify-hub/config.json and the temp log files to confirm routes and contents. If you want to proceed, manually install and configure mail-cli from a trusted source, confirm the notify profile works, run router.js with --dry-run, and review the digests produced by summarize.js before enabling automatic scheduling.

Type: OpenClaw Skill Name: notify-hub Version: 1.0.5 The skill is classified as suspicious due to a potential Remote Code Execution (RCE) vulnerability in `scripts/router.js` and `scripts/summarize.js`. These scripts use `execSync` to execute shell commands via `mail-cli` and include email subjects—which are untrusted external inputs—directly in the command string. The sanitization is insufficient (only escaping double quotes), allowing an attacker to execute arbitrary commands by sending an email with a subject containing shell metacharacters (e.g., backticks or command substitutions). While the skill's stated purpose of notification aggregation appears legitimate, this implementation flaw poses a significant security risk.