Add SMS spam and scam filter template

env-variables.manifest.json

@@ -219,6 +219,7 @@
         "contentKey": null
       }
     ],
+    "send-receive-sms-messages": [],
     "stripe-payment-link-sms": [
       {
         "key": "STRIPE_SECRET_KEY",
@@ -1949,6 +1950,68 @@
         "configurable": true,
         "contentKey": null
       }
+    ],
+    "sms-spam-scam-filter": [
+      {
+        "key": "OPENAI_API_KEY",
+        "required": true,
+        "format": "secret",
+        "description": "API key used to call OpenAI for SMS spam and scam classification",
+        "link": null,
+        "default": null,
+        "configurable": true,
+        "contentKey": null
+      },
+      {
+        "key": "OPENAI_MODEL",
+        "required": false,
+        "format": "text",
+        "description": "OpenAI model used for SMS classification",
+        "link": null,
+        "default": "gpt-5.6-luna",
+        "configurable": true,
+        "contentKey": null
+      },
+      {
+        "key": "FORWARD_WEBHOOK_URL",
+        "required": true,
+        "format": "url",
+        "description": "Webhook URL that should receive inbound SMS requests classified as genuine",
+        "link": null,
+        "default": null,
+        "configurable": true,
+        "contentKey": null
+      },
+      {
+        "key": "FORWARD_WEBHOOK_METHOD",
+        "required": false,
+        "format": "text",
+        "description": "HTTP method used when redirecting genuine SMS requests to the forwarding webhook",
+        "link": null,
+        "default": "POST",
+        "configurable": true,
+        "contentKey": null
+      },
+      {
+        "key": "OPENAI_TIMEOUT_MS",
+        "required": false,
+        "format": "number",
+        "description": "Timeout in milliseconds for OpenAI classification requests",
+        "link": null,
+        "default": "7000",
+        "configurable": true,
+        "contentKey": null
+      },
+      {
+        "key": "TWILIO_SMS_WEBHOOK_URL",
+        "required": true,
+        "format": "text",
+        "description": "The path to the webhook",
+        "link": null,
+        "default": "/sms-spam-scam-filter",
+        "configurable": false,
+        "contentKey": null
+      }
     ]
   }
 }

package.json

@@ -143,6 +143,7 @@
     "ai-assistants-samples",
     "mcp-server",
     "email-events-sms",
-    "send-receive-sms-messages"
+    "send-receive-sms-messages",
+    "sms-spam-scam-filter"
   ]
 }

sms-spam-scam-filter/.owners

@@ -0,0 +1 @@
+danohn

sms-spam-scam-filter/CHANGELOG.md

@@ -0,0 +1,9 @@
+# Changelog
+
+## [Unreleased]
+
+## [1.0.0]
+
+### Added
+
+- Initial release.

sms-spam-scam-filter/README.md

@@ -0,0 +1,93 @@
+# SMS Spam and Scam Filter
+
+This application filters inbound SMS messages before they reach your downstream webhook. Incoming messages are classified with OpenAI as either `spam_scam` or `genuine`. Spam and scam messages are ignored with an empty TwiML response. Genuine messages are redirected to your webhook with TwiML `<Redirect>`.
+
+## Pre-requisites
+
+- A [Twilio account](https://www.twilio.com/try-twilio) with an active phone number that can receive SMS
+- An OpenAI API key
+- A webhook URL that should receive genuine inbound SMS requests
+
+### Environment variables
+
+`.env` is included with placeholder values. Update the values before running. Never commit real secrets in `.env`.
+
+| Variable | Description | Required |
+| :------- | :---------- | :------- |
+| `OPENAI_API_KEY` | API key used to call OpenAI for SMS spam and scam classification | Yes |
+| `OPENAI_MODEL` | OpenAI model used for SMS classification. Defaults to `gpt-5.6-luna` | No |
+| `FORWARD_WEBHOOK_URL` | Webhook URL that should receive inbound SMS requests classified as genuine | Yes |
+| `FORWARD_WEBHOOK_METHOD` | HTTP method used when redirecting genuine SMS requests. Defaults to `POST` | No |
+| `OPENAI_TIMEOUT_MS` | Timeout in milliseconds for OpenAI classification requests. Defaults to `7000` | No |
+| `TWILIO_SMS_WEBHOOK_URL` | The webhook path — pre-set to `/sms-spam-scam-filter`, do not change | Yes |
+
+## Running the project
+
+1. Install the [Twilio CLI](https://www.twilio.com/docs/twilio-cli/quickstart#install-twilio-cli)
+1. Install the [serverless toolkit](https://www.twilio.com/docs/labs/serverless-toolkit/getting-started)
+
+```shell
+twilio plugins:install @twilio-labs/plugin-serverless
+```
+
+1. Initialize the project from the template
+
+```shell
+twilio serverless:init example --template=sms-spam-scam-filter --skip-credentials && cd example
+```
+
+1. Edit `.env` and set `OPENAI_API_KEY` and `FORWARD_WEBHOOK_URL`
+1. Start the server
+
+```shell
+twilio serverless:start
+```
+
+Check the developer console and terminal for any errors, and make sure you've set your environment variables.
+
+### Configuring your Twilio phone number
+
+You can configure the webhook via the Twilio CLI (recommended) or the Console.
+
+**Via CLI:**
+
+Find your phone number SID:
+
+```shell
+twilio api:core:incoming-phone-numbers:list
+```
+
+Then update the SMS webhook:
+
+```shell
+twilio api:core:incoming-phone-numbers:update \
+  --sid <your-phone-number-sid> \
+  --sms-url https://<your-ngrok-id>.ngrok.app/sms-spam-scam-filter \
+  --sms-method POST
+```
+
+**Via Console:**
+
+1. Go to your [Twilio Console Phone Numbers](https://console.twilio.com/us1/develop/phone-numbers/manage/incoming)
+2. Select the phone number you want to use
+3. Under "Messaging Configuration":
+   - Set "A Message Comes In" to **Webhook**
+   - Enter your URL: `https://<your-ngrok-id>.ngrok.app/sms-spam-scam-filter` (or your deployed URL)
+   - Set HTTP method to **POST**
+4. Click **Save**
+
+## Cost estimate
+
+This estimate covers only the OpenAI classification request. It does not include Twilio messaging, carrier, phone number, or webhook hosting costs.
+
+Using `gpt-5.6-luna` pricing of `$0.20` per 1M input tokens and `$1.20` per 1M output tokens, a typical one-segment SMS classification costs about `$0.0001` per checked message.
+
+## Deploying
+
+Deploy your functions and assets with the following command. Note: you must run this command from inside your project folder. [More information about the serverless toolkit is available in the docs.](https://www.twilio.com/docs/labs/serverless-toolkit)
+
+With the [Twilio CLI](https://www.twilio.com/docs/twilio-cli/quickstart):
+
+```shell
+twilio serverless:deploy --service-name sms-spam-scam-filter
+```