AddThisFeature

WhatsApp Message Delivery

Reach people on WhatsApp with approved templates, consent checks, and delivery status.

involved Sales & Marketing Tools

What it adds

A WhatsApp delivery path that respects the messaging window, uses pre-approved templates, and tracks consent and delivery state.

What your agent is told to do

5
  1. 1

    Model the messaging window explicitly. Outside the window only a pre-approved template may be sent; inside it a free-form reply is allowed. Every send must decide which case it is in before it constructs anything.

  2. 2

    Store approved templates and their parameters in the app, with the language variants each one has been approved for, and validate the parameter set against that record before dispatch rather than discovering a mismatch from a rejection.

  3. 3

    Record consent per recipient with a timestamp and the source of that consent, and check it at dispatch time. A recipient who has withdrawn consent must be unreachable on this channel from that moment.

  4. 4

    Send from the app's background-job system and keep a message row with a stable identifier and a delivery state updated only from verified provider callbacks.

  5. 5

    This brief owns WhatsApp only. If the app also has an SMS path, the choice of which channel a given notification uses belongs to the notification routing layer, not to either delivery integration, so the two do not each decide to send.

Edge cases it handles

8
  • Sending free-form content when the conversation window has closed will simply be rejected. Determine the window state from the last inbound message before composing, and fall back to the appropriate approved template.
  • Consent and channel eligibility are separate questions. A recipient may have consented and still not be reachable on WhatsApp at that number, and both must be checked before the message is queued.
  • A template can be edited, re-approved, renamed, or approved in one language and not another. Pin the version and language a message was built against, and fail loudly on a mismatch instead of sending a message with the wrong or missing parameters.
  • Delivery callbacks arrive on a public endpoint, can be replayed, and can arrive out of order. Verify their authenticity, apply them idempotently, and never let an older event overwrite a newer state.
  • A recipient who cannot be reached on WhatsApp needs a defined fallback — another channel or a visible failure — decided by the routing layer rather than silently dropped by the integration.
  • Retries after a timeout must not produce a second conversation-opening message, which is both confusing and separately billable. Use a stable per-message key.
  • Template messages are billed by conversation category. Enforce a per-account ceiling so a retry storm cannot run up an unbounded cost.
  • When the provider is unavailable, the triggering action must complete, the message must remain queued, and nothing in the app should claim the recipient has been contacted.

Definition of done

8
  • Every send resolves the messaging window and chooses a template or free-form path accordingly.
  • Templates, their parameters, and approved languages are stored and validated before dispatch.
  • Consent is recorded with a timestamp and source and re-checked at dispatch time.
  • Delivery callbacks are authenticated, idempotent, and never regress a message to an older state.
  • Retries do not open a second conversation or send a duplicate message.
  • Unreachable recipients produce a defined outcome handled by the routing layer, not a silent drop.
  • The feature matches the existing design system.
  • No existing functionality is broken.

Related features

How it works

  1. 1

    Copy the link

    Grab the Markdown instruction URL for this feature.

  2. 2

    Give it to your AI

    Paste it into Claude Code, Cursor, v0, Lovable — whatever you build with.

  3. 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.