Saddle Quickstart

Saddle is LittleHorse's managed workflow and streaming platform. In this quickstart, you will build a workflow that sends a Slack message, expose it through a webhook, and trigger it from a webhook event.

The complete runnable example, configuration template, and screenshots are in the Saddle quickstart in the LittleHorse Developer Hub.

What You Will Build

• A send-slack TaskDef deployed as a Saddle Job or run by a local Java worker.
• A WfSpec named quickstart-wf built with the Saddle Workflow Builder.
• A JSON webhook that publishes requests to a Streamlet.
• A Workflow Trigger that starts quickstart-wf for each webhook request.

Prerequisites

• Access to a Saddle environment.
• A Saddle user who can configure clients, secrets, jobs, workflows, webhooks, and triggers.
• Java 21 or newer if you run the local mock worker.
• Docker only if you use the optional local development tooling.

1. Start The Task Worker

The workflow uses a task named send-slack. Choose one of these options:

Use A Saddle Job

Use the prebuilt saddle-bag-slack worker if you have a Slack bot token:

1. Create a Slack app with the chat:write scope and install it to your workspace.
2. Invite the app to the channel where it will post messages.
3. In Saddle, open Configure > Secrets and create slack-token with the bot token.
4. Open Deploy > Saddle Jobs, click Add Saddle Job, select saddle-bag-slack, and configure its Slack token to use the slack-token secret.
5. Click Create Job. The job registers the send-slack TaskDef.

Run The Local Mock Worker

The Developer Hub includes a worker that prints the message instead of calling Slack. From the Developer Hub repository root:

cp examples/saddle/00-quickstart/.env.example examples/saddle/00-quickstart/.env

In Saddle, open Configure > Clients, create a Kernel Client, and put its generated credentials in .env. Then start the worker:

./gradlew :examples:saddle:java:00-quickstart:run

The worker registers send-slack and remains running while it polls for tasks.

2. Build The Workflow

In Saddle, open Orchestrate > Workflows, click Add Workflow, and create a workflow named quickstart-wf.

Add two STR variables:

• channel
• message

Drag a Task node onto the canvas and select send-slack. Map its channel and message parameters to the workflow variables. Add an Exit node and connect Start > send-slack > Exit.

Click Deploy WfSpec, then Register Workflow.

3. Create The Webhook

Open Connect > Webhooks, click Add Webhook, and configure:

• Name: quickstart-webhook
• Path: /quickstart
• Body: JSON

Use this body schema:

{
  "$id": "https://example.com/schemas/quickstart-webhook.json",
  "$schema": "http://json-schema.org/draft-07/schema#",
  "type": "object",
  "properties": {
    "channel": { "type": "string" },
    "message": { "type": "string" }
  },
  "required": ["channel", "message"],
  "additionalProperties": false
}

Validate the schema and create the webhook. Saddle creates the wh.quickstart-webhook Streamlet for its requests.

4. Create The Workflow Trigger

Open Stream > Workflow Triggers, click Add Trigger, and configure:

• Name: quickstart-trigger
• Type: Workflow Spec
• Workflow: quickstart-wf
• Major Version: 0.0
• Streamlet: wh.quickstart-webhook

Leave the wfRunId mapping empty so LittleHorse generates an ID for each run. Map the webhook body to the workflow variables:

• message: $.value.body.message
• channel: $.value.body.channel

Create the trigger.

5. Test The Workflow

Open the quickstart-webhook details page and copy its generated listener URL. Send a request:

curl --request POST '<WEBHOOK_LISTENER_URL>' \
  --header 'Content-Type: application/json' \
  --data '{
    "channel": "<CHANNEL>",
    "message": "Hello from Saddle!"
  }'

Replace the placeholders with the listener URL and a channel value. Confirm that a new quickstart-wf run completes in Saddle. The Slack worker posts the message to Slack; the local mock worker prints it in its terminal.

For the full walkthrough and screenshots, continue to the Developer Hub Saddle quickstart.