Customer.io Event Forwarding
Forward real product behaviour to Customer.io so campaigns fire on what users do.
What it adds
A server-side pipeline that sends a defined catalogue of product events and profile attributes to Customer.io under a stable identity.
What your agent is told to do
5
What your agent is told to do
5-
1
Write down the event catalogue before writing any code: a small fixed set of names in one tense and one convention, each with a documented property list. An open-ended stream of ad-hoc names is unusable for anyone building a campaign.
-
2
Emit events from the server after the underlying change has committed, through the app's existing background-job system. Do not emit from the browser, where events are lost to ad blockers, duplicated by reloads, and trivially forged.
-
3
Key every event and every profile on the app's own immutable user identifier rather than the email address, so changing an email moves the profile instead of creating a second one.
-
4
Separate the two payload shapes deliberately: durable profile attributes such as plan, workspace, and lifecycle stage belong on the profile, and per-occurrence detail belongs on the event. Do not accumulate one-off transactional values as profile attributes.
-
5
Where a contact-and-list sync brief is also applied to this app, that feature owns identity and subscription state. This one owns behaviour only and must never write subscription status, or the two will fight over the same contact.
Edge cases it handles
8
Edge cases it handles
8- Anonymous activity captured before signup must be attached to the real profile exactly once. A merge that replays those events as new ones double-counts every campaign trigger built on them.
- Retried and delayed jobs arrive out of order. Send each event's own occurrence timestamp rather than letting the provider stamp receipt time, or a campaign reacts to something that happened hours ago as if it were new.
- Data the app is not permitted to share — payment details, message bodies, identity documents, anything under a processing restriction — must be excluded by an allowlist of properties, never by a blocklist that the next new field escapes.
- Renaming an event breaks every campaign already listening for it. Treat names as a public contract: deprecate rather than rename, and emit both during any transition.
- Rate limits and provider outages must apply backoff and hold events in the queue. Dropping events on a rate-limit response leaves campaigns permanently missing a step for those users.
- A user deleted in the app must be deleted or suppressed at the provider too, or campaigns keep mailing someone who left.
- A burst from a bulk import, a backfill, or a runaway loop must be throttled and capped, so one job cannot exhaust the quota that every other event depends on.
- Development, staging, and seeded test activity must never reach the production workspace.
Definition of done
9
Definition of done
9- A documented event catalogue exists and every forwarded event conforms to it.
- Events are emitted server-side after commit, through the existing background-job system.
- Profiles are keyed on a stable internal identifier that survives an email change.
- Anonymous pre-signup activity is attached to the account once, with no duplicated events.
- Event properties are assembled from an allowlist and contain no restricted personal data.
- Rate limiting and provider downtime delay events rather than discarding them.
- Account deletion in the app propagates to the provider.
- The feature matches the existing design system.
- No existing functionality is broken.
Related features
Mailgun Transactional Email
Mailgun Transactional Email
Send transactional email through Mailgun without losing bounce and complaint state.
What it does
A Mailgun delivery path bound to the correct region and sending domain, with durable suppression and event handling.
How it works
- 1 Put Mailgun behind the app's existing mail layer as one adapter. It is an alternative to the other transactional email integrations, not an addition; whichever adapter is in place, the template contract and suppression handling above it are unchanged.
- 2 Bind the region and sending domain in server-side configuration and validate at startup that they are consistent. A credential issued in one region will simply not work against another, and the failure looks like an authentication problem rather than a configuration one.
- 3 Check the suppression state for an address before a message is queued, not after the provider rejects it, so bouncing and complaining addresses stop consuming send attempts.
Copy the prompt
No account needed
Add this feature to my app:
https://addthisfeature.com/x/mailgun-transactional-email
Firebase Cloud Messaging
Firebase Cloud Messaging
Deliver push through Firebase with registration tokens that stay valid as devices change.
What it does
A Firebase-backed push channel with server-side token lifecycle management, consistent payload handling across platforms, and safe handling of the routes a notification opens.
How it works
- 1 Store one record per device registration, tied to the account signed in at the time, the platform it came from, and when it was last confirmed working, and let one account hold many of them.
- 2 Send from the app's existing background-job system with the provider credentials held server-side, and treat every send as a job that can be retried rather than work done inline during a request.
- 3 Decide, per notification type, whether the payload is meant to be displayed by the platform or handled by the app, and keep that decision the same on every platform. The same event behaving as a visible alert on one platform and a silent update on another produces bugs nobody can reproduce.
Copy the prompt
No account needed
Add this feature to my app:
https://addthisfeature.com/x/firebase-cloud-messaging
Slack Channel Alerts
Slack Channel Alerts
Send the events a team actually needs into a Slack channel without flooding it.
What it does
Operator-configured routing of selected app events into Slack channels, with grouping, throttling, and reliable delivery.
How it works
- 1 List the events genuinely worth interrupting a team for — failed payments, high-value signups, error thresholds, approvals waiting — and make each individually selectable per channel. An all-or-nothing firehose gets muted within a week and then nobody sees the one that mattered.
- 2 Run the connection through the provider's authorisation flow, then confirm the workspace and channel at setup by posting a test message the operator has to acknowledge. Store the resulting credential server-side only.
- 3 Write each message so it stands alone: what happened, to whom, when, and a link back into the app. Optional fields that are absent must be omitted cleanly rather than rendering as empty labels.
Copy the prompt
No account needed
Add this feature to my app:
https://addthisfeature.com/x/slack-channel-alerts
How it works
-
1
Copy the link
Grab the Markdown instruction URL for this feature.
-
2
Give it to your AI
Paste it into Claude Code, Cursor, v0, Lovable — whatever you build with.
-
3
It inspects, then implements
Your agent reads your existing app first, then adds the feature to fit it.
Works with your stack
These instructions are written to adapt. They tell the agent to detect your framework, match your existing design system, and reuse what you already have — rather than assuming a particular stack.
Need it tighter than that? Customize the feature and tell it exactly what you're running.