reddit · SideProject · 09-30 18:50 60 分 ← 返回列表

I built an email verifier that refuses to fake a "deliverable" answer

Side project turned product. The problem: when you verify an email, catch-all domains (and inboxes behind Proofpoint/Mimecast) accept everything, so most tools return "unknown" — or worse, proxy-hop until they get a "valid" that isn't true. What skink does differently: - Tries to resolve catch-alls using out-of-band signals + SMTP behavior fingerprinting - Every result ships a calibrated confidence score and reasons[], not just a status - When it genuinely can't tell, it says so, a

五维评分

付费意愿

需求真实度

个人可落地

竞争空间

新鲜度

AI 分析

痛点：邮件验证工具面对catch-all域名和Proofpoint/Mimecast时大量返回unknown或伪造valid，导致发件人浪费发送量、损害送达率，用户需要可信的置信度与解释。

目标用户：冷邮件/外贸/增长团队、邮件营销代理机构、销售运营，以及通过Claude/Cursor等AI代理批量验证名单的开发者。

证据：Side project turned product. ... When it genuinely can't tell, it says so, and you're not charged for it. ... MCP server, so you can verify a list from Claude or Cursor directly. ... if a result says "72% likely deliverable, reason: accepts-all + mailbox signal found", would you send to it or not?

变现：按成功验证条数/积分付费，unknown不收费；提供API和MCP订阅套餐，面向团队按月订阅+超额用量计费。

MVP：用Go或Python写最小SMTP探测+catch-all指纹识别，只输出置信度与reasons[]；用CSV上传/API/MCP接入，Stripe按有效结果计费。先聚焦Proofpoint/Mimecast和catch-all一个痛点，不做全套邮件营销功能。

竞品：ZeroBounce、NeverBounce、Hunter、MillionVerifier、Clearout、Bouncer、Verifalia

摘要：Reddit侧项目发布skink邮件验证器，主打不伪造deliverable：对catch-all及Proofpoint/Mimecast邮箱给出置信度和原因，无法判断时不收费，并提供MCP供Claude/Cursor调用。发帖人询问用户对“72% likely deliverable”是否愿意发送，说明真实付费验证市场存在，但竞品密集，差异化在透明度和AI代理集成。