メニュー
Monitor and protect your email sender reputation. Use when deliverability drops, investigating spam placement, checking blocklists, or recovering from reputation damage.
Codex または Claude でインストール この Prompt をコピーして Codex、Claude、または他のアシスタントに貼り付けると、Skill ページを確認してインストールできます。
SOC 職業分類に基づく
Diagnose email problems and plan email infrastructure. Use when emails land in spam, bounce rates spike, deliverability drops, open rates decline, providers throttle sending, you need to set up email for a new product, or you're not sure which email skill to use.
Choose and configure an agent-native inbox provider for AI agents that send and receive email. Use when giving an AI agent its own email address, setting up two-way agent email communication, comparing agent inbox platforms, or adding guardrails to autonomous email workflows.
Build dashboards, alerts, and monitoring systems for email sending operations. Use when setting up deliverability monitoring, configuring alert thresholds, checking blocklists, building email metrics dashboards, or responding to deliverability incidents.
Classify inbound email replies by intent and route them appropriately. Use when building reply handling for outreach, support, or transactional email - detecting interested, not interested, OOO, bounce, unsubscribe, forwarded, and question replies.
Maintain email conversation context across messages using threading headers. Use when building thread reconstruction, linking replies to conversations, detecting thread hijacking, stripping quoted content, or providing thread context to AI agents.
Receive, parse, and process incoming email via provider webhooks. Use when setting up inbound email handling, parsing MIME messages, extracting content from replies, detecting threads, filtering spam on inbound, or routing incoming messages.
| name | sender-reputation |
| description | Monitor and protect your email sender reputation. Use when deliverability drops, investigating spam placement, checking blocklists, or recovering from reputation damage. |
| license | MIT |
Understand how mailbox providers score your domain, monitor your reputation, and recover when it degrades.
domain-authentication - SPF, DKIM, DMARC setup (prerequisite for good reputation)email-warmup - ramping volume on new domains/IPs without damaging reputationbounce-handling - processing bounces that directly affect reputation signalssuppression-lists - managing the lists that protect you from repeat damagerate-limiting - volume controls that prevent reputation-damaging spikesMailbox providers maintain a score for every domain (and IP) that sends email to their users. This score determines whether your messages land in the inbox, the spam folder, or get rejected outright. The score is not a single number you can look up - each provider calculates its own, using its own signals and weights.
The core signals are the same everywhere:
| Signal | Weight | Threshold |
|---|---|---|
| Spam complaint rate | Highest | Under 0.1% (hard limit 0.3%) |
| Bounce rate | High | Under 2% (ideally under 1%) |
| Spam trap hits | High | Zero tolerance for pristine traps |
| Engagement (opens, replies, clicks) | High | Provider-specific, relative to your history |
| Volume patterns | Medium | Consistent, no sudden spikes |
| Authentication (SPF, DKIM, DMARC) | Medium | All three passing and aligned |
| List age and hygiene | Medium | No purchased/scraped lists |
| Content patterns | Lower | No spam-trigger phrases or deceptive content |
Two things to internalize about reputation:
Until roughly 2020, IP reputation was king. You could warm up a new dedicated IP, build its reputation, and deliverability was primarily an IP-level concern. Shared IPs meant shared risk.
That has shifted. Modern mailbox providers - Gmail especially - now weight domain reputation more heavily than IP reputation. Here is why that matters:
Domain reputation:
IP reputation:
Practical implications:
mail.example.com instead of example.com) gives you a fresh domain reputation, but providers are increasingly sophisticated at linking subdomains to parent domains.This is the single most damaging signal. When a recipient clicks "Report Spam" or "Mark as Junk," that is a direct negative vote against your domain.
Thresholds:
Complaint rate is calculated as complaints divided by delivered messages (not sent messages). A 0.1% rate means 1 complaint per 1,000 delivered emails.
What drives complaints up:
Every hard bounce - a permanent delivery failure to an address that does not exist or permanently rejects mail - is a negative signal. Providers interpret high bounce rates as evidence that you are not maintaining your lists.
Thresholds:
Soft bounces (temporary failures like a full mailbox or server timeout) are less damaging individually but indicate problems if they persist. A recipient that soft-bounces repeatedly (3-5 times across multiple sends) should be suppressed. Providers track whether you continue sending to addresses that keep bouncing.
Spam traps are email addresses operated by mailbox providers, blocklist operators, and anti-spam organizations specifically to catch senders with bad practices. There are three types:
Pristine traps are addresses created solely to catch spammers. They have never belonged to a real person and have never opted in to anything. Hitting one means you acquired the address through scraping, purchasing lists, or guessing. The consequences are severe - immediate blocklisting is common.
Recycled traps are abandoned email addresses that providers have repurposed. They were once real addresses, returned bounces for a period (typically 6-12 months), and were then reactivated as traps. Hitting one means you are not cleaning your list of addresses that have been bouncing. The impact is less severe than pristine traps but still degrades reputation over time.
Typo traps capture emails sent to common misspellings of major domains (e.g., user@gmial.com, user@yaho.com). These indicate you are not validating addresses at the point of collection.
You will never know which specific addresses are spam traps. The only defense is list hygiene: validate addresses at collection, remove addresses that bounce, and never purchase or scrape email lists.
Mailbox providers track how recipients interact with your messages. Positive engagement (opens, clicks, replies) signals that your mail is wanted. Negative engagement (ignoring, deleting without reading, moving to spam) signals the opposite.
Gmail is particularly engagement-heavy in its scoring. A domain that sends to a large list with low open rates will see worse inbox placement over time, even if no one is actively complaining. The mail is just not wanted enough.
Engagement signals that help reputation:
Engagement signals that hurt reputation:
Disengaged recipients are a slow poison. They do not cause immediate damage like complaints or bounces, but they steadily erode your engagement ratios. A production email system should track recipient fatigue - scoring each contact based on send frequency, bounce history, complaint history, and engagement decay - and stop sending to contacts that show sustained disengagement.
Mailbox providers watch for sudden changes in sending volume. A domain that normally sends 500 emails per day and suddenly sends 50,000 looks like a compromised account or a new spam operation.
What triggers volume-based filtering:
Consistent, predictable volume is what providers want to see. Ramp gradually. If you need to increase volume, do it over weeks, not days. See the email-warmup skill for specific ramp schedules.
You cannot manage what you cannot measure. Set up monitoring on all three major provider tools and at least one third-party blocklist checker.
The primary tool for monitoring Gmail deliverability. Free, requires domain verification.
Setup:
What it shows (as of the 2025 v2 update):
Important change: In September 2025, Google retired the domain and IP reputation dashboards from Postmaster Tools, shifting to compliance-driven metrics. You no longer see a "High/Medium/Low/Bad" reputation label. Instead, focus on keeping your spam rate below the threshold lines and authentication rates at 100%.
Microsoft's equivalent for Outlook.com, Hotmail, and Live.com recipients.
Setup:
What it shows:
Limitations: SNDS is IP-based, not domain-based. You need to know your sending IPs. If you use a shared-IP provider, you may not be able to access SNDS data directly - ask your provider for it. Requires a minimum of 100 messages per day to an IP for data to appear.
Yahoo's monitoring tool for Yahoo Mail and AOL recipients.
Setup:
What it shows:
Notes: Data populates within 24-48 hours for domains meeting Yahoo's minimum daily volume. No API access available - you must check manually. Register for their CFL to receive individual complaint notifications.
In addition to provider-specific tools, use these:
Blocklist checkers:
Sender Score:
Feedback loops: Register for feedback loops (FBLs) with major providers. When a recipient marks your email as spam, the FBL sends you a notification so you can suppress that address immediately. Most ESPs handle this automatically, but verify it is configured.
Blocklists (also called blacklists or DNSBLs) are databases of IPs and domains known to send spam. Mailbox providers and corporate mail servers query these lists in real time when deciding whether to accept a message.
| Blocklist | Type | Impact | What gets you listed |
|---|---|---|---|
| Spamhaus SBL | IP | Very high | Sending spam, hosting spam operations |
| Spamhaus DBL | Domain | Very high | Spam domains in URLs, From addresses, or sending infrastructure |
| Spamhaus XBL | IP | High | Compromised/infected machines sending spam |
| Spamhaus PBL | IP | Medium | End-user IP ranges that should not send mail directly |
| Barracuda BRBL | IP | High | High spam volume from the IP |
| Spamcop | IP | Medium | User-reported spam from the IP |
| SORBS | IP | Medium | Various spam and abuse signals |
Spamhaus is the most consequential. A Spamhaus listing will affect delivery to a significant portion of the internet because most large mail servers query Spamhaus.
Check your sending domain and IPs against major blocklists:
# Check Spamhaus (replace with your IP)
dig +short 4.3.2.1.zen.spamhaus.org
# Non-empty response means you are listed
# Check Spamhaus DBL (domain)
dig +short example.com.dbl.spamhaus.org
# Check Barracuda
dig +short 4.3.2.1.b.barracudacentral.org
Or use a web-based tool like MXToolbox that checks multiple lists at once.
Spamhaus:
Barracuda:
General rules for any blocklist:
Recovery is harder than prevention. A domain with damaged reputation needs a disciplined, multi-week process to rebuild trust with mailbox providers.
Before you fix anything, understand the scope:
dig for SPF, DKIM, and DMARC recordsdomain-authentication skill.| Damage level | Typical recovery time |
|---|---|
| Minor (slightly elevated spam rate, no blocklisting) | 2-4 weeks |
| Moderate (spam rate over 0.3%, some filtering) | 4-8 weeks |
| Severe (blocklisted, widespread spam-folder placement) | 8-12+ weeks |
| Domain burned (persistent blocklisting, extensive complaints) | Consider a new subdomain with fresh warmup |
The cheapest reputation management is prevention. These practices keep your reputation healthy so you never need the recovery playbook.
email-warmup skill)List-Unsubscribe and List-Unsubscribe-Post headers in all non-transactional emailUse different subdomains for different types of email:
| Stream | Subdomain example | Why |
|---|---|---|
| Transactional | mail.example.com | Protects critical email (receipts, auth) from marketing reputation |
| Marketing | news.example.com | Isolates marketing risk from transactional delivery |
| Cold outreach | outreach.example.com | Highest risk - keep completely separate |
Each subdomain builds its own reputation. A problem with your marketing sends will not drag down your transactional delivery. Each subdomain also gets its own SPF record, solving the 10-lookup limit problem.
Ignoring disengaged recipients. The most common reputation killer is not bounces or complaints - it is continuing to send to people who never open your emails. Engagement ratios decline slowly, inbox placement drops gradually, and by the time you notice, your domain reputation has been degraded for weeks.
Reacting to blocklisting by switching IPs. Your domain reputation follows you. A new IP does not fix a domain reputation problem. Fix the root cause first.
Not monitoring until something breaks. By the time you notice deliverability problems in your open rates, the reputation damage happened days or weeks ago. Set up Google Postmaster Tools and SNDS before you need them.
Treating all bounces the same. Hard bounces (address does not exist) and soft bounces (temporary failure) require different handling. Hard bounces need immediate permanent suppression. Soft bounces need retry logic with escalation to suppression after repeated failures.
Sending the same volume every day including weekends. Some senders configure automated sends to run identically every day. Weekend sending patterns that match weekday patterns can look unusual to providers. Match your volume to natural business patterns.
No separation between transactional and marketing mail. When marketing sends get filtered, your password reset emails and order confirmations go with them. Use separate subdomains.
Assuming your ESP handles everything. Your ESP manages the technical sending infrastructure, but reputation is your domain's reputation. You are responsible for list hygiene, complaint rates, and engagement management. The ESP cannot fix bad sending practices.
Over-sending to new subscribers. A new subscriber who gets 5 emails in their first week is more likely to complain than one who gets a welcome email and then joins the normal cadence. Space out early touches.