<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom" xmlns:dc="http://purl.org/dc/elements/1.1/">
  <channel>
    <title>DEV Community: Steve</title>
    <description>The latest articles on DEV Community by Steve (@danby).</description>
    <link>https://dev.to/danby</link>
    <image>
      <url>https://media2.dev.to/dynamic/image/width=90,height=90,fit=cover,gravity=auto,format=auto/https:%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Fuser%2Fprofile_image%2F4137839%2Fbfdaa57c-492a-4ad7-90ff-3131d6e95873.jpg</url>
      <title>DEV Community: Steve</title>
      <link>https://dev.to/danby</link>
    </image>
    <atom:link rel="self" type="application/rss+xml" href="https://dev.to/feed/danby"/>
    <language>en</language>
    <item>
      <title>How to prove what your AI agent emailed</title>
      <dc:creator>Steve</dc:creator>
      <pubDate>Wed, 18 Nov 2026 16:15:53 +0000</pubDate>
      <link>https://dev.to/danby/how-to-prove-what-your-ai-agent-emailed-m02</link>
      <guid>https://dev.to/danby/how-to-prove-what-your-ai-agent-emailed-m02</guid>
      <description>&lt;p&gt;When an AI agent sends email for you, you usually have only its own account of what it sent. That is a weak record. The first time someone asks "did we really send that, to whom, and what was checked first?", your own logs are all you have, and nobody outside your company can check them.&lt;/p&gt;

&lt;p&gt;Here is a different approach: a signed receipt for each send, which anyone can check in a browser, without an account and without seeing your email.&lt;/p&gt;

&lt;h2&gt;
  
  
  What the receipt contains
&lt;/h2&gt;

&lt;p&gt;SentriSend is an outbound email security gateway. Each successful single send returns a field called &lt;code&gt;publicReceipt&lt;/code&gt;. It is a compact JWS signed with an Ed25519 key, and it records:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;a random receipt ID, the time it was issued, and the issuer (&lt;code&gt;https://sentrisend.com&lt;/code&gt;)&lt;/li&gt;
&lt;li&gt;the outcome, &lt;code&gt;provider_accepted&lt;/code&gt;, which means the email provider took the message, not that it reached an inbox&lt;/li&gt;
&lt;li&gt;the result of each check run before sending: recipient pre-flight, a global threat lookup, and policy&lt;/li&gt;
&lt;li&gt;two salted SHA-256 "commitments", one for the recipient and one for the subject and message&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;It does not contain the address, the subject, the message body or your API key. The commitments are fingerprints. They let you prove later that a receipt belongs to one particular send, without revealing what was sent.&lt;/p&gt;

&lt;p&gt;The public keys are at &lt;code&gt;https://sentrisend.com/api/v1/receipts/keys&lt;/code&gt;, so anyone can check a signature without asking SentriSend for permission.&lt;/p&gt;

&lt;h2&gt;
  
  
  Check a real one
&lt;/h2&gt;

&lt;p&gt;This receipt comes from a real send through the production service on 17 November 2026. It has no address, subject or message text in it:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight plaintext"&gt;&lt;code&gt;eyJhbGciOiJFZERTQSIsInR5cCI6InNlbnRyaXNlbmQtcmVjZWlwdCtqd3MiLCJraWQiOiJyZWNlaXB0LWQ3ZGY5ZDk3LTdkNjMtNDIxNC04ZGI3LTk3ZjE4MGRiN2Y4MiJ9.eyJ2ZXJzaW9uIjoxLCJpc3MiOiJodHRwczovL3NlbnRyaXNlbmQuY29tIiwidHlwZSI6InNlbnRyaXNlbmQuc2VuZC1yZWNlaXB0IiwicmVjZWlwdElkIjoiYTE5MmE3ODctZjhkYS00MTUxLWE3ZWEtYzI4NmE1Zjk4MjY2IiwiaXNzdWVkQXQiOiIyMDI2LTExLTE3VDIwOjMwOjQ4LjUwNFoiLCJvdXRjb21lIjoicHJvdmlkZXJfYWNjZXB0ZWQiLCJjaGVja3MiOnsicmVjaXBpZW50UHJlZmxpZ2h0IjoiYWxsb3dlZCIsImdsb2JhbFRocmVhdExvb2t1cCI6Im5vX21hdGNoIiwicG9saWN5IjoiYWxsb3dlZCJ9LCJyZWNpcGllbnRDb21taXRtZW50IjoiOWYxNzkzZWUyYWQwMzYwODQzNTMwOGM0ZTY1NTE4YzVhMTcxODE5Yzk0MjBlYjU2NGE5MTY4YWJiMmQ3ZDM2ZCIsIm1lc3NhZ2VDb21taXRtZW50IjoiMmI0NmYxNTk2ZjY0Y2Y5MTViNWQ5NGI4NjFiZGNmYjEyNDE3MTU2NWRmNWE4ZmZlNTIxZmNmY2I0MjIxOTVmMSJ9.UtbX-w-lCv1oYlxeccWdqGDvOfXxKv-8gFheY8Mnkdy4Hfw3wCidoqB8-5a1Ha7O8tTDz9Pww6LnIwZ2NRi2CQ
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;Paste it into the checker at &lt;a href="https://sentrisend.com/receipts" rel="noopener noreferrer"&gt;sentrisend.com/receipts&lt;/a&gt; and press &lt;strong&gt;Check signature&lt;/strong&gt;. The page reports whether it matches a published, non-revoked key, and shows the decoded fields.&lt;/p&gt;

&lt;p&gt;To tie a receipt to one particular send, you also need the original recipients, subject and message, plus the opening data returned with the send. The comparison runs in your browser. Even a trailing space changes the result, because the commitments are exact.&lt;/p&gt;

&lt;h2&gt;
  
  
  What a receipt does not show
&lt;/h2&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;Delivery.&lt;/strong&gt; &lt;code&gt;provider_accepted&lt;/code&gt; means the provider accepted the message, not that it was read.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;That the message was appropriate.&lt;/strong&gt; The checks are the ones SentriSend ran. They are not a judgement on the content.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Anonymity.&lt;/strong&gt; Someone who guesses your recipient list could test the guess against the opening data, so handle it with care.&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;The opening data (two random salts) comes back only in the original send response, and SentriSend doesn't store it. Keep it if you may want to prove a send later.&lt;/p&gt;

&lt;h2&gt;
  
  
  Try it
&lt;/h2&gt;

&lt;p&gt;A free Sandbox key needs no card, allows 100 requests a month, and sends only to its owner's verified address, so it is a safe way to try receipts on your own mailbox. Full field definitions are in the &lt;a href="https://sentrisend.com/openapi.json" rel="noopener noreferrer"&gt;OpenAPI description&lt;/a&gt;.&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;Site: &lt;a href="https://sentrisend.com" rel="noopener noreferrer"&gt;sentrisend.com&lt;/a&gt;
&lt;/li&gt;
&lt;li&gt;Docs and examples: &lt;a href="https://github.com/sentrisend/SentriSend" rel="noopener noreferrer"&gt;github.com/sentrisend/SentriSend&lt;/a&gt;
&lt;/li&gt;
&lt;li&gt;Hosted MCP server: &lt;code&gt;https://sentrisend.com/mcp&lt;/code&gt;
&lt;/li&gt;
&lt;/ul&gt;

</description>
      <category>ai</category>
      <category>security</category>
      <category>email</category>
      <category>api</category>
    </item>
    <item>
      <title>Why your Resend or SES account gets paused after a signup spike (and how to stop it)</title>
      <dc:creator>Steve</dc:creator>
      <pubDate>Tue, 27 Oct 2026 14:47:22 +0000</pubDate>
      <link>https://dev.to/danby/why-your-resend-or-ses-account-gets-paused-after-a-signup-spike-and-how-to-stop-it-24bo</link>
      <guid>https://dev.to/danby/why-your-resend-or-ses-account-gets-paused-after-a-signup-spike-and-how-to-stop-it-24bo</guid>
      <description>&lt;p&gt;Your email account can be paused after one bad day of signups, even if you never sent a single spam message. Here is why, and what to do about it before it happens.&lt;/p&gt;

&lt;p&gt;Picture a small app (this one is invented, but the pattern is common). It sends about 100 emails a day: confirmation links, password resets, receipts. One morning it is mentioned in a popular newsletter and 400 people sign up in an hour. Forty of them mistype their address or use one that never existed. Forty confirmation emails bounce within minutes. By lunchtime the developer gets a warning from their email provider, and by the evening sending is paused. Real customers are now not receiving their password resets.&lt;/p&gt;

&lt;h2&gt;
  
  
  The limits are published, and they are low
&lt;/h2&gt;

&lt;p&gt;Email providers watch two numbers: the share of your messages that bounce (the address does not exist or cannot receive mail), and the share that recipients mark as spam.&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;Resend&lt;/strong&gt; asks senders to stay &lt;a href="https://resend.com/docs/knowledge-base/account-quotas-and-limits" rel="noopener noreferrer"&gt;under a 4% bounce rate and under a 0.08% spam rate&lt;/a&gt;. Going over either can lead to a temporary sending pause.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Amazon SES&lt;/strong&gt; &lt;a href="https://docs.aws.amazon.com/ses/latest/dg/reputationdashboardmessages.html" rel="noopener noreferrer"&gt;automatically places an account under review at a 5% bounce rate or a 0.1% complaint rate&lt;/a&gt;, and might pause sending at 10% bounces or 0.5% complaints.&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;These figures come from each provider's own documentation as of October 2026. Check the current pages before you rely on them.&lt;/p&gt;

&lt;h2&gt;
  
  
  Why small senders get caught
&lt;/h2&gt;

&lt;p&gt;A percentage is fair over the long run but jumpy over a short one. Providers look at your recent sending, so what matters is how much good mail surrounds a spike. A busy sender with thousands of normal emails in that window barely notices five bad addresses. A small app that sends 100 emails on a quiet day sees the same five as a 5% bounce rate. The addresses are no worse; there is just less good mail around them to dilute the spike.&lt;/p&gt;

&lt;p&gt;Signup forms make it worse. They are the one place where strangers type addresses by hand, and where people deliberately enter fake ones to get past a gate. A typo like gmial.com, a throwaway address, or a made-up domain will often bounce.&lt;/p&gt;

&lt;p&gt;The other trap is timing. A bounce problem usually arrives as a spike, not a slow drift, and a spike can cross a threshold before you have looked at a dashboard.&lt;/p&gt;

&lt;h2&gt;
  
  
  What actually helps, in order
&lt;/h2&gt;

&lt;p&gt;Each step catches a different kind of bad address. Do the cheap ones first.&lt;/p&gt;

&lt;ol&gt;
&lt;li&gt;
&lt;strong&gt;Check the shape.&lt;/strong&gt; One @, something on each side, no spaces. Keep it simple; clever patterns reject real addresses.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Check the domain can receive mail.&lt;/strong&gt; A domain with no mail servers will always bounce. In Node this is a few lines:
&lt;/li&gt;
&lt;/ol&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight javascript"&gt;&lt;code&gt;&lt;span class="k"&gt;import&lt;/span&gt; &lt;span class="p"&gt;{&lt;/span&gt; &lt;span class="nx"&gt;promises&lt;/span&gt; &lt;span class="nx"&gt;as&lt;/span&gt; &lt;span class="nx"&gt;dns&lt;/span&gt; &lt;span class="p"&gt;}&lt;/span&gt; &lt;span class="k"&gt;from&lt;/span&gt; &lt;span class="dl"&gt;"&lt;/span&gt;&lt;span class="s2"&gt;node:dns&lt;/span&gt;&lt;span class="dl"&gt;"&lt;/span&gt;&lt;span class="p"&gt;;&lt;/span&gt;

&lt;span class="k"&gt;async&lt;/span&gt; &lt;span class="kd"&gt;function&lt;/span&gt; &lt;span class="nf"&gt;domainHasMailServer&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="nx"&gt;email&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt; &lt;span class="p"&gt;{&lt;/span&gt;
  &lt;span class="kd"&gt;const&lt;/span&gt; &lt;span class="nx"&gt;domain&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="nx"&gt;email&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nf"&gt;split&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="dl"&gt;"&lt;/span&gt;&lt;span class="s2"&gt;@&lt;/span&gt;&lt;span class="dl"&gt;"&lt;/span&gt;&lt;span class="p"&gt;)[&lt;/span&gt;&lt;span class="mi"&gt;1&lt;/span&gt;&lt;span class="p"&gt;];&lt;/span&gt;
  &lt;span class="k"&gt;try&lt;/span&gt; &lt;span class="p"&gt;{&lt;/span&gt;
    &lt;span class="kd"&gt;const&lt;/span&gt; &lt;span class="nx"&gt;records&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="k"&gt;await&lt;/span&gt; &lt;span class="nx"&gt;dns&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nf"&gt;resolveMx&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="nx"&gt;domain&lt;/span&gt;&lt;span class="p"&gt;);&lt;/span&gt;
    &lt;span class="k"&gt;return&lt;/span&gt; &lt;span class="nx"&gt;records&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nx"&gt;length&lt;/span&gt; &lt;span class="o"&gt;&amp;gt;&lt;/span&gt; &lt;span class="mi"&gt;0&lt;/span&gt;&lt;span class="p"&gt;;&lt;/span&gt;
  &lt;span class="p"&gt;}&lt;/span&gt; &lt;span class="k"&gt;catch&lt;/span&gt; &lt;span class="p"&gt;{&lt;/span&gt;
    &lt;span class="k"&gt;return&lt;/span&gt; &lt;span class="kc"&gt;false&lt;/span&gt;&lt;span class="p"&gt;;&lt;/span&gt; &lt;span class="c1"&gt;// no MX records or no such domain (simplified: also false on temporary DNS errors)&lt;/span&gt;
  &lt;span class="p"&gt;}&lt;/span&gt;
&lt;span class="p"&gt;}&lt;/span&gt;
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;Some domains legitimately accept mail without MX records, so treat a failure as a warning for a signup form rather than a hard block if you want to be generous.&lt;/p&gt;

&lt;ol&gt;
&lt;li&gt;
&lt;strong&gt;Block disposable domains.&lt;/strong&gt; Throwaway inboxes often vanish after the first email. Use a maintained list of disposable domains and refresh it regularly.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Confirm before you send anything else.&lt;/strong&gt; Send only the confirmation link, and send nothing more until it is clicked. An unconfirmed address then costs you one email, not twenty.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Stop sending to hard bounces.&lt;/strong&gt; When your provider reports an address as permanently undeliverable, never send to it again.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Watch your own bounce rate.&lt;/strong&gt; Check it daily and set an alert at about half of your provider's limit, so a spike warns you before it pauses you.&lt;/li&gt;
&lt;/ol&gt;

&lt;h2&gt;
  
  
  Where SentriSend fits
&lt;/h2&gt;

&lt;p&gt;I built &lt;a href="https://sentrisend.com" rel="noopener noreferrer"&gt;SentriSend&lt;/a&gt; because I wanted steps 2 to 6 handled in one place. You can use it only to check addresses before you send with any provider, or you can send through it.&lt;/p&gt;

&lt;p&gt;It checks an address before you send: disposable-domain screening and live DNS and mail-server checks. Sending through it adds a protection layer per API key: if hard bounces start rising, it warns, then slows, then pauses sending for that key, to help stop a spike before it reaches your provider's limit. Every accepted send gets a signed receipt and an audit record that leaves out the recipient, subject and body.&lt;/p&gt;

&lt;p&gt;What it does not do:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;It does not guarantee inbox placement, and it does not replace your provider's own account health.&lt;/li&gt;
&lt;li&gt;It only knows about mail sent through it. Bounces from elsewhere on your account are invisible to it.&lt;/li&gt;
&lt;li&gt;The bounce protection has automated tests but I have not yet run it against a real provider's suspension limits. I will say here when I have.&lt;/li&gt;
&lt;li&gt;It is new, from a small company, with no customer case studies yet.&lt;/li&gt;
&lt;/ul&gt;

&lt;h2&gt;
  
  
  Try it
&lt;/h2&gt;

&lt;p&gt;The free plan gives you 100 requests a month and no card is needed; you verify your email address and get a key. Free keys can send only to your own address, which is enough to see how the checks and receipts work. Pro is $29 a month for 50,000 requests and sending to any recipient. Details are on the &lt;a href="https://sentrisend.com/pricing" rel="noopener noreferrer"&gt;pricing page&lt;/a&gt;.&lt;/p&gt;

&lt;p&gt;If you try it and something is unclear or broken, email me at &lt;a href="mailto:steve@sentrisend.com"&gt;steve@sentrisend.com&lt;/a&gt;. I read every message and I would rather hear about a problem than lose you quietly.&lt;/p&gt;

&lt;p&gt;Whatever you use, do steps 1 to 6. A paused email account on a launch day is a worse problem than a few lines of validation.&lt;/p&gt;

</description>
      <category>email</category>
      <category>webdev</category>
      <category>node</category>
      <category>saas</category>
    </item>
    <item>
      <title>One bad tenant can sink your whole email platform: what I learned building a reputation circuit breaker</title>
      <dc:creator>Steve</dc:creator>
      <pubDate>Tue, 20 Oct 2026 17:51:20 +0000</pubDate>
      <link>https://dev.to/danby/one-bad-tenant-can-sink-your-whole-email-platform-what-i-learned-building-a-reputation-circuit-5067</link>
      <guid>https://dev.to/danby/one-bad-tenant-can-sink-your-whole-email-platform-what-i-learned-building-a-reputation-circuit-5067</guid>
      <description>&lt;p&gt;I'm a solo founder in the UK building SentriSend, a security gateway that sits in front of transactional email (AWS SES / Resend). This week I shipped two features. Building them changed how I think about multi-tenant email, so here's what I learned.&lt;/p&gt;

&lt;h2&gt;
  
  
  The problem nobody sees: subscription bombing
&lt;/h2&gt;

&lt;p&gt;Bots push a victim's address into hundreds of signup forms at once. The victim's inbox floods, and every site involved picks up complaints and spam-trap hits.&lt;/p&gt;

&lt;p&gt;The nasty part is that &lt;strong&gt;each individual site only sees one or two "signups"&lt;/strong&gt;. From where any single app is standing, nothing looks wrong.&lt;/p&gt;

&lt;h2&gt;
  
  
  Two layers of defence
&lt;/h2&gt;

&lt;p&gt;&lt;strong&gt;1. Per-sender limits.&lt;/strong&gt; Any one sender is stopped from repeatedly emailing the same address in a short window. Blocked requests get an HTTP 429 with a &lt;code&gt;Retry-After&lt;/code&gt; header, and no quota is charged.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;2. Cross-sender detection.&lt;/strong&gt; A gateway sits in front of many senders, so it can see something no single app can: lots of &lt;em&gt;unrelated&lt;/em&gt; senders suddenly emailing the same address. That's a strong bombing signal. It triggers an alert first, then a temporary pause.&lt;/p&gt;

&lt;p&gt;In a live production test, six identical sends went to one address. Five were accepted, and the sixth came back &lt;code&gt;429 recipient_rate_limited&lt;/code&gt;.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Privacy:&lt;/strong&gt; no plain addresses are stored. Recipients become keyed, pseudonymous identifiers, which are cleared automatically within hours.&lt;/p&gt;

&lt;h2&gt;
  
  
  The lesson: one bad tenant can sink everyone
&lt;/h2&gt;

&lt;p&gt;While planning the next feature, I realised something uncomfortable. &lt;strong&gt;Every customer sends through my provider account.&lt;/strong&gt; One customer with a scraped or stale list could push bounces and complaints high enough to get the whole account suspended, and every other customer would go down with them.&lt;/p&gt;

&lt;p&gt;Worse, I had no feedback loop at all. Provider "accepted" isn't the same as delivered, and I wasn't receiving bounce or complaint reports.&lt;/p&gt;

&lt;h2&gt;
  
  
  The fix: a reputation circuit breaker
&lt;/h2&gt;

&lt;ol&gt;
&lt;li&gt;
&lt;strong&gt;Listen.&lt;/strong&gt; A signed webhook receives delivery, bounce and complaint events. Each event is matched back to the API key that sent it, using hashed message IDs, and duplicates are counted once.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Measure.&lt;/strong&gt; Each key's hard-bounce and complaint rates are tracked over a rolling window, with minimum sample sizes so a key isn't judged on three emails.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Act.&lt;/strong&gt; Warn, then slow, then pause, all well before the provider's published limits. A paused key only returns through a manual review and a probation period. If the reputation check itself fails, sending stops rather than quietly carrying on.&lt;/li&gt;
&lt;/ol&gt;

&lt;p&gt;There's also an account-wide alert, plus an alarm if webhook reports stop arriving, because silence shouldn't be mistaken for good health.&lt;/p&gt;

&lt;h2&gt;
  
  
  Takeaways if you run multi-tenant email
&lt;/h2&gt;

&lt;ul&gt;
&lt;li&gt;If tenants share your sending account, &lt;strong&gt;their reputation is your reputation&lt;/strong&gt;. Isolate them.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;"Sent" isn't "delivered".&lt;/strong&gt; Wire up provider webhooks early.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Fail safely.&lt;/strong&gt; Decide deliberately which checks fail open and which fail closed.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Write the privacy policy from the code&lt;/strong&gt;, not from memory. Mine caught two mismatches before publishing.&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;There's a free developer sandbox at sentrisend.com. If you run multi-tenant email, I'd love to hear how you stop one tenant hurting the rest.&lt;/p&gt;

</description>
      <category>email</category>
      <category>security</category>
      <category>saas</category>
      <category>webdev</category>
    </item>
    <item>
      <title>How AWS SES bounce rates can put your account at risk, and how to reduce bad signups before sending</title>
      <dc:creator>Steve</dc:creator>
      <pubDate>Tue, 22 Sep 2026 14:50:14 +0000</pubDate>
      <link>https://dev.to/danby/why-aws-ses-puts-saas-accounts-on-probation-and-how-to-stop-the-5-bounce-cliff-55h6</link>
      <guid>https://dev.to/danby/why-aws-ses-puts-saas-accounts-on-probation-and-how-to-stop-the-5-bounce-cliff-55h6</guid>
      <description>&lt;p&gt;If you send transactional email through Amazon SES, your bounce rate matters more than most developers realise until it becomes a problem. This post explains how bad signups push it up, what AWS says about thresholds, and one way to reduce the problem before you send. I'm the author of SentriSend, the API mentioned below, so treat this as a builder's perspective rather than a neutral review.&lt;/p&gt;

&lt;h2&gt;
  
  
  The thresholds
&lt;/h2&gt;

&lt;p&gt;AWS publishes bounce-rate guidance for SES. As I understand it, AWS recommends keeping the rate low (around 2% or under), may place an account under review above 5%, and may pause sending at 10% or more. These numbers can change, so check the current &lt;a href="https://docs.aws.amazon.com/ses/" rel="noopener noreferrer"&gt;SES documentation&lt;/a&gt; rather than relying on this post. Other providers, including SendGrid, have their own rules.&lt;/p&gt;

&lt;p&gt;If sending is paused, everything that depends on your email stops: verification codes, password resets and receipts.&lt;/p&gt;

&lt;h2&gt;
  
  
  Clean DNS doesn't prevent bounces
&lt;/h2&gt;

&lt;p&gt;SPF, DKIM and DMARC matter for deliverability, but they don't stop bounces caused by the addresses you send to. Three common sources:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;Automated signups with disposable or made-up addresses&lt;/li&gt;
&lt;li&gt;Typos in real users' addresses&lt;/li&gt;
&lt;li&gt;Domains that don't exist or can't receive mail&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;For a small sender, a modest number of bad addresses can move the percentage a lot, because the total volume is low.&lt;/p&gt;

&lt;h2&gt;
  
  
  Handling it after the fact isn't enough
&lt;/h2&gt;

&lt;p&gt;Many guides cover SNS/SQS bounce notifications and suppression lists. Those are worth setting up, but they run after the message has been sent and the bounce has been recorded. To reduce bounces in the first place, you need to check the address before you send.&lt;/p&gt;

&lt;h2&gt;
  
  
  A pre-send check
&lt;/h2&gt;

&lt;p&gt;SentriSend is an API that validates an address with one call before you send. It checks that:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;the address is well formed&lt;/li&gt;
&lt;li&gt;the domain isn't on a bundled list of about 133,000 known disposable domains (the list is built into the service at build time)&lt;/li&gt;
&lt;li&gt;the domain exists and can receive mail, using live DNS/MX lookups&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;If the DNS result can't be determined, the address is allowed with MEDIUM risk, so a slow lookup doesn't block a genuine user.&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight shell"&gt;&lt;code&gt;curl &lt;span class="nt"&gt;-X&lt;/span&gt; POST https://sentrisend.com/api/v1/validate &lt;span class="se"&gt;\&lt;/span&gt;
  &lt;span class="nt"&gt;-H&lt;/span&gt; &lt;span class="s2"&gt;"Authorization: Bearer YOUR_SENTRISEND_KEY"&lt;/span&gt; &lt;span class="se"&gt;\&lt;/span&gt;
  &lt;span class="nt"&gt;-H&lt;/span&gt; &lt;span class="s2"&gt;"Content-Type: application/json"&lt;/span&gt; &lt;span class="se"&gt;\&lt;/span&gt;
  &lt;span class="nt"&gt;-d&lt;/span&gt; &lt;span class="s1"&gt;'{"email": "user@example.com"}'&lt;/span&gt;
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;An allowed address returns HTTP 200:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight json"&gt;&lt;code&gt;&lt;span class="p"&gt;{&lt;/span&gt;&lt;span class="w"&gt; &lt;/span&gt;&lt;span class="nl"&gt;"allowed"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;&lt;span class="w"&gt; &lt;/span&gt;&lt;span class="kc"&gt;true&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;&lt;span class="w"&gt; &lt;/span&gt;&lt;span class="nl"&gt;"disposable"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;&lt;span class="w"&gt; &lt;/span&gt;&lt;span class="kc"&gt;false&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;&lt;span class="w"&gt; &lt;/span&gt;&lt;span class="nl"&gt;"risk_score"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;&lt;span class="w"&gt; &lt;/span&gt;&lt;span class="s2"&gt;"LOW"&lt;/span&gt;&lt;span class="w"&gt; &lt;/span&gt;&lt;span class="p"&gt;}&lt;/span&gt;&lt;span class="w"&gt;
&lt;/span&gt;&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;A disposable domain returns HTTP 400:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight json"&gt;&lt;code&gt;&lt;span class="p"&gt;{&lt;/span&gt;&lt;span class="w"&gt; &lt;/span&gt;&lt;span class="nl"&gt;"allowed"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;&lt;span class="w"&gt; &lt;/span&gt;&lt;span class="kc"&gt;false&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;&lt;span class="w"&gt; &lt;/span&gt;&lt;span class="nl"&gt;"disposable"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;&lt;span class="w"&gt; &lt;/span&gt;&lt;span class="kc"&gt;true&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;&lt;span class="w"&gt; &lt;/span&gt;&lt;span class="nl"&gt;"reason"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;&lt;span class="w"&gt; &lt;/span&gt;&lt;span class="s2"&gt;"DISPOSABLE_DOMAIN"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;&lt;span class="w"&gt; &lt;/span&gt;&lt;span class="nl"&gt;"risk_score"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;&lt;span class="w"&gt; &lt;/span&gt;&lt;span class="s2"&gt;"HIGH"&lt;/span&gt;&lt;span class="w"&gt; &lt;/span&gt;&lt;span class="p"&gt;}&lt;/span&gt;&lt;span class="w"&gt;
&lt;/span&gt;&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;Your application reads the result and decides whether to continue sending through SES, Resend or whatever you use.&lt;/p&gt;

&lt;h2&gt;
  
  
  What it doesn't do
&lt;/h2&gt;

&lt;ul&gt;
&lt;li&gt;It doesn't confirm that an individual mailbox exists. A real domain with no such mailbox can still bounce.&lt;/li&gt;
&lt;li&gt;It doesn't detect bots. It validates the address, so pair it with CAPTCHA or rate limiting on your forms.&lt;/li&gt;
&lt;li&gt;It doesn't replace bounce handling. Keep your SNS/SQS suppression list.&lt;/li&gt;
&lt;/ul&gt;

&lt;h2&gt;
  
  
  Privacy
&lt;/h2&gt;

&lt;p&gt;In SentriSend's send audit records, the recipient, subject and message bodies are redacted. Customer account and billing data is stored separately.&lt;/p&gt;

&lt;h2&gt;
  
  
  Try it
&lt;/h2&gt;

&lt;p&gt;The free tier is 100 requests a month, with no card needed. There's also a live simulator on the site, and the docs and a working example are public.&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fypk8j89ta47aenlbccks.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fypk8j89ta47aenlbccks.png" alt=" " width="800" height="418"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;Site: &lt;a href="https://sentrisend.com" rel="noopener noreferrer"&gt;https://sentrisend.com&lt;/a&gt;
&lt;/li&gt;
&lt;li&gt;GitHub: &lt;a href="https://github.com/sentrisend/SentriSend" rel="noopener noreferrer"&gt;https://github.com/sentrisend/SentriSend&lt;/a&gt;
&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;If you've dealt with bad signups hitting your bounce rate, I'd like to hear how you handle it.&lt;/p&gt;

</description>
      <category>security</category>
      <category>aws</category>
      <category>webdev</category>
      <category>saas</category>
    </item>
  </channel>
</rss>
