Standardmäßig ist der Prompt ausgewählt, der zuerst die Quelle prüft. Sie können zu einem direkten Befehl wechseln oder eine lokale Kopie herunterladen.
Quelldateien prüfen
Lesen Sie SKILL.md und alle von SkillsMP angezeigten Begleitdateien, bevor Sie sich für eine Installation entscheiden.
Mit Codex oder Claude installieren Kopieren Sie diesen Prompt, fügen Sie ihn in Codex, Claude oder einen anderen Assistant ein und lassen Sie die Skill-Seite prüfen und installieren.
Ein direkter Befehl überspringt den Prüf-Prompt. Prüfen Sie die Quelle, bevor Sie ihn ausführen.
Help the user with SendGrid (Twilio) platform questions — from transactional email via the Email API and SMTP relay through Marketing Campaigns, Dynamic Templates, Event Webhooks, Inbound Parse, domain authentication, dedicated IPs, suppressions, email testing, and integrations.
Step 1 — Gather context
If references/learnings.md exists, read it first for accumulated knowledge.
Ask the user:
What area of SendGrid do you need help with?
A) Email API — sending transactional email via REST API or SMTP relay, SDKs
B) Marketing Campaigns — drag-and-drop editor, single sends, automations, signup forms
C) Dynamic Templates — Handlebars-based transactional templates, template versioning
What are you trying to accomplish? (describe your specific goal or question)
If the user's request already provides most of this context, skip directly to the relevant step. Lead with your best-effort answer using reasonable assumptions (stated explicitly), then ask only the most critical 1-2 clarifying questions at the end — don't gate your response behind gathering complete context.
Note: If the user needs a specialized skill, route them there with a brief explanation of why that skill is a better fit.
Step 2 — Route or answer directly
If the request maps to a specialized skill, route:
General email marketing strategy / best practices -> /sales-email-marketing
Otherwise, answer directly from platform knowledge using the reference below.
Step 3 — SendGrid platform reference
Read references/platform-guide.md for detailed module documentation, pricing, integrations, and data model.
You no longer need the platform guide details — focus on the user's specific situation.
Step 4 — Actionable guidance
Based on the user's specific question:
Sending transactional email via the API:
Create an API key in Settings > API Keys with "Mail Send" permission
Authenticate your sending domain in Settings > Sender Authentication (add the CNAME records to your DNS)
Install the SDK for your language (e.g., npm install @sendgrid/mail for Node.js)
Set the API key: sgMail.setApiKey(process.env.SENDGRID_API_KEY)
Compose and send: set to, from, subject, text/html (or templateId + dynamicTemplateData)
Handle the response — check for 202 Accepted (success) or error codes
Set up Event Webhooks to track delivery and engagement events
Setting up Marketing Campaigns automation:
Subscribe to a Marketing Campaigns plan (automations require Advanced plan)
Import or add contacts to a list — use CSV upload, signup forms, or the Contacts API
Navigate to Automations > Create Automation
Choose a trigger: welcome series (new contact added to list), date-based, or custom
Build the email sequence — add email steps with delays between them
Design each email using the drag-and-drop editor or code editor
Activate the automation and monitor entry/completion rates in the automation dashboard
Configuring domain authentication:
Go to Settings > Sender Authentication > Authenticate Your Domain
Select your DNS host and enter your sending domain
SendGrid generates CNAME records for SPF and DKIM — add these to your DNS provider
Click Verify in SendGrid — it checks for the DNS records
Set up branded tracking links (link branding) — add another CNAME for click/open tracking
If using a dedicated IP, configure reverse DNS to map the IP to your domain
Set up a DMARC record separately in your DNS (e.g., v=DMARC1; p=quarantine; rua=mailto:dmarc@yourdomain.com)
Setting up Event Webhooks:
Go to Settings > Mail Settings > Event Webhook
Enter your webhook URL (must be publicly accessible HTTPS endpoint)
Select the events you want to receive (processed, delivered, open, click, bounce, etc.)
Enable the webhook
Implement signature verification on your endpoint using SendGrid's public key
Process incoming JSON arrays of event objects — each has event, email, timestamp, and custom args
Handle retries — SendGrid retries failed deliveries for up to 24 hours
Setting up Inbound Parse:
Add an MX record for your receiving domain pointing to mx.sendgrid.net
Go to Settings > Inbound Parse > Add Host & URL
Enter your receiving domain and the webhook URL to receive parsed email
Choose raw (MIME) or parsed (structured fields) format
Implement your webhook endpoint to process the multipart POST data (from, to, subject, body, attachments)
Test by sending an email to any address at your configured domain
Gotchas
Best-effort from research — verify details against current SendGrid documentation.
Email API and Marketing Campaigns are separate subscriptions with separate billing. Many users expect one plan covers everything. If you need to send both transactional email (order confirmations, password resets) and marketing email (newsletters, promotions), you need two subscriptions. Budget accordingly and understand which features belong to which plan.
The Free Trial is 60 days only with 100 emails/day — not a permanent free tier. Unlike some competitors, SendGrid's free offering expires. After 60 days you must upgrade to a paid plan or lose sending capability. Plan your evaluation and migration timeline around this limit.
Dedicated IPs require warmup or you will hit deliverability problems immediately. New dedicated IPs have no sending reputation. If you send high volume on a cold IP, mailbox providers will throttle or reject your email. Use SendGrid's automated warmup (enabled by default) or create a manual warmup schedule ramping from hundreds to your target volume over 4-6 weeks.
Suppression groups are not enabled by default — without them, unsubscribes are global. If you do not configure suppression groups (also called "unsubscribe groups"), any unsubscribe removes the contact from ALL your email. Set up suppression groups early (e.g., "Marketing", "Product Updates", "Billing Alerts") so recipients can opt out of specific categories without losing all communication.
Event Webhook payloads are batched and can be delayed — do not rely on them for real-time logic. SendGrid batches webhook events and may deliver them with a delay of seconds to minutes. If you need real-time delivery confirmation, poll the API or use the mail/send response code (202 = accepted for processing). Design your webhook consumers to handle out-of-order and duplicate events idempotently.
sg_event_id changes on retries — do NOT use it for deduplication. Despite the docs saying sg_event_id is "a unique ID to this event that you can use for deduplication purposes," the ID changes on every retry attempt. Same email, same event, same timestamp — but a different sg_event_id. Build a composite dedup key instead: ${sg_message_id}:${event}:${email}. Store processed keys in Redis (with 48-hour TTL to cover SendGrid's 24-hour retry window) or a database, and check before processing each event.
Shared IP reputation can tank your deliverability with no warning. On free/Essentials plans, you share IPs with other senders. If bad actors on the same pool trigger Spamhaus blocklists or Microsoft blocks, your email gets caught too — even with 99% sender score and perfect authentication. SendGrid's official position is that "blocklists are a natural part of sending through shared IP Pools." If you're seeing unexplained deliverability drops, check your IP against blocklists (mxtoolbox.com) and consider upgrading to a dedicated IP (Pro plan).
The Node.js SDK doesn't pool HTTP connections — bulk sends cause socket exhaustion. If you fire many sgMail.send() calls concurrently, you'll hit EMFILE errors (file descriptor limit). The SDK creates a new connection per request. Use sequential sending (for...of with await), a concurrency limiter (p-limit set to 2-10), or personalizations (up to 1000 recipients per single request) to reduce connection count.
Sending all messages in a single SDK array call can delay delivery by hours. When you pass an array to sgMail.send(), SendGrid queues them server-side. For time-sensitive sends, use personalizations (same content, different recipients in one request) or send individual calls with concurrency control rather than one massive batch.
Self-improving: If you discover something not covered here, append it to references/learnings.md with today's date.
Step 5 — Related skills
/sales-email-marketing — Email marketing strategy and best practices (platform-agnostic)
/sales-email-tracking — Email open and click tracking strategy
/sales-integration — Connect SendGrid to other tools via Zapier, Make, or API
/sales-funnel — Funnel strategy and conversion optimization
/sales-do — Not sure which skill to use? The router matches any sales objective to the right skill. Install: npx skills add sales-skills/sales --skill sales-do
Examples
Example 1: Setting up transactional email for a SaaS app
User says: "I need to send password reset and welcome emails from my Node.js app using SendGrid."
Skill does:
Creates a restricted API key with only "Mail Send" permission, then authenticates the domain (adds CNAME records for SPF/DKIM)
Shows how to install @sendgrid/mail and configure the API key via environment variable
Creates two Dynamic Templates in the SendGrid UI — one for password reset (with {{reset_link}}) and one for welcome (with {{first_name}}, {{login_url}})
Provides code examples using sgMail.send() with templateId and dynamicTemplateData, and recommends Event Webhooks to track delivery failures and alert on bounces
Result: User has production-ready transactional email with templates, proper authentication, and delivery monitoring
Example 2: Migrating marketing email from Mailchimp to SendGrid
User says: "We're moving our newsletter from Mailchimp to SendGrid Marketing Campaigns. What's the process?"
Skill does:
Explains the dual-subscription model — Marketing Campaigns is separate from Email API, so they need the right plan (Basic or Advanced)
Walks through exporting contacts from Mailchimp (CSV) and importing into SendGrid with field mapping — only import subscribed contacts, and recreate suppression groups first
Guides recreation of email templates using SendGrid's drag-and-drop editor or code editor
Covers DNS changes (remove Mailchimp's records, add SendGrid's CNAMEs for SPF/DKIM) and a warmup strategy: start with most engaged contacts, expand over 2-4 weeks, especially on a new dedicated IP (Advanced plan)
Result: User has a migration plan covering contacts, templates, DNS, suppressions, and warmup to maintain deliverability
Example 3: Setting up Event Webhooks for delivery tracking
User says: "I want to track which emails bounce and get delivered so I can update our CRM."
Skill does:
Walks through implementing a webhook endpoint that receives JSON event batches, with signature verification using SendGrid's public key to secure it
Designs the CRM update logic: mark contacts as "delivered" on delivery events, flag as "invalid" on hard bounces, queue for retry on deferrals
Warns about batched/delayed delivery and duplicate events — recommends idempotent processing using sg_message_id as a dedup key; also subscribe to open/click events for engagement tracking
Result: User has a webhook pipeline that keeps their CRM in sync with SendGrid delivery and engagement events
Troubleshooting
Emails going to spam after switching to a dedicated IP
Symptom: Deliverability drops and emails start landing in spam shortly after upgrading to a Pro or Advanced plan with a dedicated IP
Cause: The dedicated IP is new and has no sending reputation. Mailbox providers (Gmail, Yahoo, Outlook) are cautious about unknown IPs and will throttle or spam-folder email from them.
Solution: Enable automated IP warmup in Settings > IP Addresses (it should be on by default for new IPs). If you turned it off, re-enable it. During warmup, SendGrid splits your traffic between the new dedicated IP and the shared IP pool, gradually shifting volume to the dedicated IP over weeks. Do not send your full volume on a cold IP. If deliverability is already damaged, pause high-volume sends, re-enable warmup, and start with your most engaged recipients. Monitor mailbox-provider stats to see which providers are rejecting. See /sales-deliverability for a comprehensive warmup guide.
Event Webhook not receiving events
Symptom: You configured an Event Webhook URL but your endpoint is not receiving any event data
Cause: The webhook URL is not publicly accessible, HTTPS certificate is invalid, the endpoint is returning non-2xx status codes, or the webhook is not enabled.
Solution: Verify the webhook is enabled in Settings > Mail Settings > Event Webhook. Ensure your URL is publicly reachable (not localhost or behind a firewall). Check that your HTTPS certificate is valid and not self-signed. Test the endpoint manually with a POST of sample JSON. Check that your endpoint returns 2xx status codes within 3 seconds — SendGrid treats timeouts and errors as failures and will retry, but eventually stops after 24 hours. Use a tool like ngrok for local development testing. Check SendGrid's Event Webhook activity log for delivery attempts and error codes.
Gmail/Yahoo blocking emails with DMARC alignment error
Symptom: Emails deferred or rejected with 421 4.7.32 From: header isn't aligned with either the authenticated SPF or DKIM organizational domain — especially common with scheduled emails
Cause: The From address domain doesn't match the domain you authenticated in SendGrid. For example, sending from a gmail.com address through SendGrid, or sending from user@example.com when only notifications.example.com is authenticated. Gmail and Yahoo enforce DMARC alignment strictly since Feb 2024.
Solution: Go to Settings > Sender Authentication > Authenticate Your Domain and authenticate the exact domain you send from. Add the CNAME records SendGrid generates to your DNS. Make sure your From address matches — if you authenticated example.com, send from anything@example.com. Add a DMARC record to your DNS: v=DMARC1; p=none; rua=mailto:dmarc@yourdomain.com (start with p=none to monitor). If you're on a shared IP plan and still seeing throttling after fixing authentication, the shared IP may have poor reputation — check it at mxtoolbox.com and consider a dedicated IP (Pro plan).
Symptom: Duplicate webhook events processed despite checking sg_event_id — same email, same event type, same timestamp, but different sg_event_id values
Cause: sg_event_id changes on every retry attempt. Despite the docs calling it a deduplication key, it is unique per delivery attempt, not per logical event.
Solution: Build a composite dedup key from fields that ARE stable across retries: ${sg_message_id}:${event}:${email}. Store processed keys in Redis with a 48-hour TTL (covers SendGrid's 24-hour retry window with margin) or a database table with a unique constraint. Check before processing each event. For high-volume systems, use a Bloom filter or Redis SET with EXPIRE for fast lookups.
EMFILE error / socket exhaustion when sending bulk email
Symptom: Error: connect EMFILE when sending hundreds or thousands of emails — typically the first few batches succeed and then later ones fail
Cause: Each sgMail.send() call opens a new HTTP connection. Node.js has a default file descriptor limit (~1024). Concurrent requests exhaust available sockets before earlier connections close. The SendGrid Node SDK does not pool connections.
Solution: Three approaches (pick one): (1) Send sequentially: for (const msg of messages) { await sgMail.send(msg); } — slowest but simplest. (2) Limit concurrency: use p-limit set to 2-10 concurrent requests: const limit = pLimit(5); await Promise.all(msgs.map(m => limit(() => sgMail.send(m))));. (3) Use personalizations: if recipients get the same content with different substitution data, send up to 1000 recipients in a single API call using the personalizations array — this is the most efficient approach and uses just one HTTP connection.
Marketing contacts not appearing in segments
Symptom: Contacts were imported or added via the API but do not appear in segments or lists as expected
Cause: Contact ingestion in Marketing Campaigns is asynchronous — it can take minutes to hours for contacts to be fully indexed and available in segments. Alternatively, the import may have partially failed due to invalid fields or duplicate handling.
Solution: Check the import status in Marketing > Contacts > Imports for errors or warnings. Allow up to 2 hours for large imports to fully index. Verify that custom field names in the import match the custom fields defined in your SendGrid account (mismatches cause fields to be silently dropped). Check that contacts have the expected field values by searching for a specific contact in the UI. For API-added contacts (PUT /v3/marketing/contacts), check the response for the job_id and poll GET /v3/marketing/contacts/imports/{id} for status. If contacts appear in the list but not the segment, review the segment query conditions — ensure the field names and operators match the actual contact data.
