RCS の開始方法

AWS RCS エージェントが正常に作成され、テスト登録が送信されました。通常、テストエージェントは数分以内に承認されます。次に、デバイスへのテストメッセージングを有効にします。

テストデバイスを追加すると、テスターの招待はすぐには送信されません。システムはアクティベーションを少なくとも 120 秒間遅延させ、招待が到着するまでに最大 20 分かかることがあります。コンソールには、おおよそのアクティベーション時間が表示されます。デバイスを追加する前に待つ必要はありません。システムは自動的に遅延を処理します。

iOS デバイス (iOS 18 以降の iPhone) では、テスターの招待がメインの受信トレイではなく、メッセージアプリの不明な送信者フォルダに表示される場合があります。招待が表示されない場合は、不明な送信者フォルダを確認してください。

次のプロンプトをコピーし、AI エージェントまたはコーディングアシスタントに貼り付けます。

## RCS Setup Assistant Prompt

Help me set up RCS messaging in AWS End User Messaging using the AWS CLI.
The service is `pinpoint-sms-voice-v2`. Walk me through each step with exact
CLI commands. Ask me for all required details before generating any commands.

**Important rules for generating commands:**
- All commands use the `pinpoint-sms-voice-v2` service.
- Use `create-rcs-agent` exactly as spelled — NOT `create-r-c-s-agent`.
- Use the term "testing" — NOT "sandbox".
- There is NO `describe-messages` API. Do not generate it.
- `create-rcs-agent` does NOT accept brand asset parameters (no display name,
  no logo, no banner, no color). Brand assets are registration fields only.
- `create-verified-destination-number` uses `--rcs-agent-id`, NOT
  `--origination-identity`.

### Step 1: Create an RCS Agent

Use `create-rcs-agent`. This creates the agent resource only.
Optional parameters: `--deletion-protection-enabled`, `--opt-out-list-name`,
`--tags`.
The response returns `RcsAgentId` and `RcsAgentArn` — save both.

### Step 2: Create and submit a testing registration

This configures brand assets and submits for approval. It requires multiple
API calls in sequence:

a. `create-registration --registration-type TEST_RCS_LAUNCH_REGISTRATION`
   → returns `RegistrationId`. Save it.

b. `create-registration-association --registration-id <id> --resource-id <agent-id>`
   → links the registration to the agent.

c. Upload images as attachments (two calls):
   `create-registration-attachment --attachment-body fileb://<logo-path>`
   `create-registration-attachment --attachment-body fileb://<banner-path>`
   → each returns `RegistrationAttachmentId`. Save both.

d. Set ALL required registration fields using `put-registration-field-value`
   with `--registration-id`, `--field-path`, and the appropriate value flag
   (`--text-value`, `--select-choices`, or `--registration-attachment-id`).

   Required fields (ALL must be set or registration will be DENIED):
   - `agentDetails.brandName` (text, 2-65 chars)
   - `agentDetails.serviceName` (text, 1-100 chars)
   - `agentDetails.senderDisplayName` (text, 1-40 chars)
   - `agentDetails.useCase` (select: OTP, TRANSACTIONAL, PROMOTIONAL, MULTI_USE)
   - `agentDetails.agentDescription` (text, 1-100 chars)
   - `agentDetails.logoImage` (attachment ID from step c, 224x224 PNG)
   - `agentDetails.bannerImage` (attachment ID from step c, 1440x448 PNG/JPEG)
   - `agentDetails.accentColor` (text, hex code e.g. #0066CC)
   - `agentDetails.privacyPolicyUrl` (text, valid URL)
   - `agentDetails.termsAndConditionsUrl` (text, valid URL)
   - `agentDetails.averageMonthlyRcsFrequency` (select: 10, 100, 1000+)
   - `agentDetails.monthlyRcsVolume` (text, 1-100000)
   - At least ONE contact method WITH its label:
     agentDetails.contactWebsite + agentDetails.contactWebsiteLabel, OR
     agentDetails.contactPhoneNumber + agentDetails.contactPhoneLabel, OR
     agentDetails.contactEmailAddress + agentDetails.contactEmailLabel

e. Verify all fields: `describe-registration-field-values --registration-id <id>`
   Any field showing `DeniedReason: MISSING_REQUIRED_FIELD` must be set.

f. Submit: `submit-registration-version --registration-id <id>`

g. Poll status: `describe-registrations --registration-ids <id>`
   Wait for `RegistrationStatus: COMPLETE`.

**Error recovery:** If registration is DENIED, you must:
1. `create-registration-version --registration-id <id>` (creates new draft)
2. Re-populate ALL fields from scratch (new versions do NOT inherit values)
3. Fix the issue noted in `DeniedReasons`
4. Re-submit

### Step 3: Add a test device

**Prerequisite:** Step 2 must be COMPLETE and the agent's `TestingAgent.Status`
must be `ACTIVE` (check with `describe-rcs-agents`). Then wait at least 120
seconds after the agent becomes ACTIVE.

Use `create-verified-destination-number --destination-phone-number <E.164>
--rcs-agent-id <agent-id>`.

The device status will be `PENDING`. The user must accept the RCS tester
invitation on their physical device. Check status with
`describe-verified-destination-numbers` — wait for `VERIFIED`.

### Step 4: Send a test RCS message

**Prerequisite:** Step 3 device must be `VERIFIED`.

Use `send-text-message --destination-phone-number <E.164>
--origination-identity <agent-arn> --message-body "<text>"
--message-type TRANSACTIONAL`.

Returns `MessageId`.

### Step 5: Verify delivery

For testing: check the test device — the message appears from the branded
RCS agent.

For production monitoring: set up event destinations BEFORE sending messages
using `create-event-destination` (SNS, CloudWatch Logs, or Firehose). Event
destinations do not retroactively capture events for already-sent messages.
CloudWatch metrics in the `AWS/SMSVoice` namespace provide aggregate stats.

---

**Before generating commands, ask me for:**
- Brand name, service name, and sender display name
- Agent description (what the agent does, what messages users receive)
- Use case type: OTP, TRANSACTIONAL, PROMOTIONAL, or MULTI_USE
- Logo file path (224x224 PNG) and banner file path (1440x448 PNG/JPEG)
- Brand accent color hex code (e.g. #0066CC)
- Privacy policy URL and terms & conditions URL
- One contact method with label: website URL, phone number, or email
- Estimated monthly RCS volume and per-user message frequency
- Test device phone number in E.164 format (e.g. +12065550100)