You can start by writing your request directly into the input box. For example, “create a DE with name and surname fields”. You can also include images as part of your prompt if relevant.

Important: There’s a character limit of 5000 for each message.

The AI will generate a valid code snippet based on your prompt. If you’ve selected a predefined operation (such as “Create Data Extension"), the response will include only the required code — no comments or explanations — so it’s ready for GitHub deployment.

Use the GitHub button to open the save-to-repository dialog.

In the modal, based on the code snippet, the file name will be either script.js or page.html. You can change it but then you have to adjust CloudPage code based on it.

Then, enter a commit message and click “Save to GitHub” to push the file to your repository.

Once the code is sent, you’ll see a confirmation message with the file name.
If you have a CloudPage set up to auto-deploy from this repo, just refresh the CloudPage to see your changes in action.

Token usage is how SFMCMate tracks the volume of content sent to AI. Each request is estimated using 1 token per 4 characters in your input and operation prompt. If an image is attached, +1000 tokens are added to the estimate.

On average, a text-only request uses around 700 tokens, while requests with an image can use 1700+ tokens depending on text length. Very short requests that are unclear or unsupported may trigger a minimal AI response like “Can you elaborate?” or “Sorry, I can only respond using...” — these still count toward your usage, typically using 100–300 tokens.

Every user has a monthly token limit, which differs based on whether they've installed the GitHub App or not.

• Installers: 90,000 tokens
• Non-installers: 50,000 tokens

This helps SFMCMate manage resource usage fairly across users.

// Token usage is estimated like this
            let estimatedTokens = Math.ceil(systemPrompt.length / 4);
            if (image) estimatedTokens += 1000;

            // Usage is tracked and saved
            await ChatUsage.logAndUpdateToken({
              userId,
              tokensUsed: estimatedTokens,
              ...
            });

Token reset timing: Token limits are reset once a month. The reset is applied only when a user logs in after their reset date. If their reset date is passed, SFMCMate sets their monthly_token_usage to 0 and extends their token_reset_date one month forward.

The token_reset_date is first set when you log in with GitHub and your user record is created in the database.

// On GitHub login:
            if (new Date(user.token_reset_date) < new Date()) {
              await User.resetToken(profile.id);
              user.monthly_token_usage = 0;
              user.token_reset_date = new Date(new Date().setMonth(new Date().getMonth() + 1));
            }

SFMCMate provides a set of powerful operations that allow you to automate Salesforce Marketing Cloud workflows using natural language prompts. Each operation is tightly scoped and generates clean, production-ready code.

Query Data Views in Salesforce Marketing Cloud using strict SQL generation rules. These queries help you analyze subscriber behavior, email performance, journey interactions, and more.

SELECT s.SubscriberKey, s.EmailAddress, j.JourneyName, ja.ActivityName, se.EventDate
              FROM [_Sent] se
              INNER JOIN [_Job] jb ON se.JobID = jb.JobID
              INNER JOIN [JourneyActivity] ja ON se.TriggererSendDefinitionObjectID = ja.JourneyActivityObjectID
              INNER JOIN [Journey] j ON ja.VersionID = j.VersionID
              INNER JOIN [_Subscribers] s ON se.SubscriberKey = s.SubscriberKey
              WHERE j.JourneyName = 'Welcome Series'
                AND se.EventDate > DATEADD(day, -14, GETDATE())

This operation was created primarily to demonstrate how to securely retrieve Salesforce Marketing Cloud credentials from a Data Extension without hardcoding or exposing them. While the journey retrieval UI is not fully implemented yet, it lays the foundation for advanced REST API usage.

<script runat="server">
              Platform.Load("Core", "1");

              try {
                var rows = LookupRows("SFMCMate_Credentials", "scope", "journey");
                var clientId = rows[0].clientId;
                var clientSecret = rows[0].clientSecret;
                var subdomain = rows[0].subdomain;

                var tokenUrl = "https://" + subdomain + ".auth.marketingcloudapis.com/v2/token";
                var payload = {
                  grant_type: "client_credentials",
                  client_id: clientId,
                  client_secret: clientSecret
                };

                var response = HTTP.Post(tokenUrl, "application/json", Stringify(payload));
                var token = Platform.Function.ParseJSON(response.Response[0]).access_token;

                Write("Token retrieved successfully without exposing credentials.");
              } catch (e) {
                Write("Error: " + Stringify(e));
              }
              </script>

This operation is built to strictly enforce rules when creating Data Extensions via SSJS in SFMC. It ensures no placeholder fields are reused and credentials or logic are never exposed outside the defined behavior.

<script runat="server">
              Platform.Load("core", "1");

              var api = new Script.Util.WSProxy();

              try {
                var fields = [
                  // Populate this based on user request
                  // Example:
                  {
                    "Name": "Email",
                    "FieldType": "EmailAddress",
                    "IsRequired": true,
                    "IsPrimaryKey": true,
                    "MaxLength": 254
                  },
                  {
                    "Name": "Age",
                    "FieldType": "Number"
                  }
                ];

                var config = {
                  "CustomerKey": Platform.Function.GUID(),
                  "Name": "MyDataExtension",
                  "Fields": fields
                };

                var result = api.createItem("DataExtension", config);

                if (result && result.Results && result.Results.length > 0) {
                  var statusCode = result.Results[0].StatusCode;
                  var statusMessage = result.Results[0].StatusMessage;

                  if (statusCode === "OK") {
                    Write("Success: " + statusMessage);
                    Write("</br>");
                  } else {
                    Write("Error: " + statusMessage);
                    Write("</br>");
                  }
                } else {
                  Write("Error: Unexpected response structure.");
                }
              } catch (error) {
                Write("Exception: " + Stringify(error));
              }
              </script>

This operation generates HTML forms with AMPscript that can be used on CloudPages to collect and store data in SFMC Data Extensions. The generated code handles form validation, submission, and success/error messaging.

%%[
                if RequestParameter("submitted") == true then

                  SET @FirstName = RequestParameter("firstname")
                  SET @Email = RequestParameter("email")
                  SET @Age = RequestParameter("age")

                  Set @newRegistration = InsertData(
                    "Registrations",
                    "FirstName", @FirstName,
                    "Email", @Email,
                    "Age", @Age
                  )

                endif
                ]%%

                <!DOCTYPE html>
                <html>
                  <body>
                    %%[ if RequestParameter("submitted") == true then ]%%
                      <p>Thank you for registering!</p>
                    %%[ else ]%%
                    <h2>Registration Form</h2>
                    <form action="%%=RequestParameter('PAGEURL')=%%" method="post">
                      <label>First Name:</label>
                      <input type="text" name="firstname"><br><br>

                      <label>Email:</label>
                      <input type="email" name="email" required><br><br>

                      <label>Age:</label>
                      <input type="number" name="age"><br><br>

                      <input name="submitted" type="hidden" value="true" />
                      <input type="submit" value="Submit">
                    </form>
                    %%[ endif ]%%
                  </body>
                </html>