Quick Start

This guide takes you from zero to a running workflow in under 10 minutes. You'll create an account, set up a project, add an LLM API key, create an agent, build a workflow, and run it.

---

What You'll Build

By the end of this guide, you'll have a working workflow that:

1. Accepts a topic as input (manual trigger)
2. Runs an LLM agent that writes a short summary about the topic
3. Outputs the result

---

Step 1 — Create Your Account

1. Go to the TARX app and click Sign up.
2. Choose Continue with Google (fastest) or enter your email and a password.
3. If you used email, check your inbox for a verification email and click the link.
4. After signing in, you'll be prompted to choose a username (your public handle, like a GitHub username). Pick something you like — this appears in all your resource URLs.

Username rules

Your username must be lowercase, 3–20 characters, letters/numbers/hyphens only. It becomes your URL prefix: /{username}/.... Choose carefully — it cannot be changed later.

---

Step 2 — Create Your First Project

After setting your username, you land on the Projects screen.

1. Click New Project.
2. Give it a name — something like "My First Project".
3. The URL slug is auto-generated from the name (e.g., my-first-project). You can edit it if you want.
4. Click Create Project.

You're now inside your project. The URL is /{username}/my-first-project/workflows.

---

Step 3 — Add an LLM API Key

TARX is BYOK — you need to add your own API key to use LLM agents.

1. Click Settings in the left sidebar (gear icon at the bottom).
2. You're taken to /{username}/settings. Click the Keys & Secrets tab.
3. Click Add API Key.
4. Choose a Provider — for this guide, pick OpenAI or Anthropic (whichever key you have).
5. Give it a name like "My OpenAI Key" — this is what appears in the agent editor dropdown.
6. Paste your API key value into the Key field.
7. Click Save.

Your key is now encrypted at rest using industry-standard encryption. The raw value is never stored in plaintext.

Where to get API keys

• OpenAI: platform.openai.com/api-keys
• Anthropic: console.anthropic.com/keys
• Google: aistudio.google.com/app/apikey

---

Step 4 — Create Your First Agent

Now let's create a reusable agent.

1. Navigate back to your project. Click Agents in the left sidebar.

2. Click New Agent.

3. You land on the Agent Editor (full-page, no modals). Fill in:

   Field	Value
   Name	summarizer (no spaces — this becomes the expression variable {{summarizer.output}})
   Icon	Pick any emoji
   API Key	Select the key you just added
   Model	Pick a fast model from the live list — e.g. gpt-4o-mini (OpenAI) or claude-haiku-4-5 (Anthropic). The dropdown is populated from your key, so you don't need to memorize model strings.
   System Prompt	See below

4. For the System Prompt, enter:

   You are a concise summarizer. When given a topic, write a clear 3-sentence summary 
   covering: what it is, why it matters, and one interesting fact. 
   Keep your response under 100 words.

5. Leave Temperature at 0.7 and Max Tokens at 512.

6. Skip the Capabilities, Skills, RAG, and MCP sections for now.

7. Click Save Agent.

Agent naming

Use short, lowercase names without spaces. The agent name becomes the expression variable used in workflows — {{summarizer.output}}, {{writer.output}}, etc.

---

Step 5 — Build Your First Workflow

1. Click Workflows in the left sidebar.
2. Click New Workflow.
3. Give it a name: "Topic Summarizer". Click Create.
4. You land on the visual canvas (React Flow editor).

Add Nodes

The left panel shows the Node Palette. Let's build our flow:

Add a Trigger node:

• The canvas starts with a Trigger node by default (Manual type).
• Click the Trigger node to open the inspector panel on the right.
• Set Trigger Type to Manual.
• Set Default Input to something like "quantum computing" — this is the test value used when you run manually.

Add an Agent node:

• Drag an Agent node from the palette onto the canvas.
• Drop it to the right of the Trigger node.
• Click the Agent node to open the inspector.
• Select your summarizer agent from the dropdown.
• In the Input field, enter: {{trigger.output}}
  • This passes whatever the trigger received into the agent.

Connect the nodes:

• Hover over the Trigger node — you'll see a small circle (handle) on the right edge.
• Click and drag from that handle to the Agent node's left handle.
• A green line connects them.

Add an Output node:

• Drag an Output node from the palette.
• Drop it to the right of the Agent node.
• Connect the Agent node's right handle to the Output node's left handle.
• In the Output inspector, set the Value to: {{summarizer.output}}

Your canvas now shows:

[Manual Trigger] ──→ [Agent: summarizer] ──→ [Output]

Save the Workflow

Click Save in the top-right toolbar (or Cmd+S / Ctrl+S).

---

Step 6 — Run the Workflow

1. Click the Run button (play icon) in the top toolbar.
2. A run modal appears. The Input field shows your default value ("quantum computing"). You can change it to any topic.
3. Click Execute.
4. Watch the canvas — nodes animate as they execute:
   • The Trigger node flashes blue (starting)
   • The Agent node pulses orange (LLM call in progress)
   • The Agent node turns green (complete)
   • The Output node turns green (workflow done)
5. A result banner appears at the bottom with the output text.

Live streaming

You can watch execution in real time because TARX uses Server-Sent Events (SSE) to stream node status updates. This works for long-running workflows too.

---

Step 7 — View the Execution Details

1. Click Executions in the left sidebar.
2. You'll see your just-completed execution at the top.
3. Click on it to expand the details.
4. Each node shows:
   • Status (completed / failed / skipped)
   • Duration (how long the node took)
   • Input (what was passed in)
   • Output (what was returned)
5. Expand the Agent: summarizer node to see the full LLM response.

---

What's Next?

You've built and run your first TARX workflow. Here's where to go from here:

I want to...	Go to
Add more power to my agent (web search, web scraping)	Capabilities
Build a more complex workflow with branching logic	Conditions
Loop over a list of 100 items	Loops
Trigger workflows automatically	Triggers
Add knowledge from my own vector database	RAG Sources
Let Bridge create workflows for me	Bridge Overview
Visualize workflow data with charts	Canvas
Invite team members	Projects

---

You didn't use Bridge yet

Bridge is your AI command center. Once you're comfortable with the manual flow, try asking Bridge: "Create an agent called researcher that uses web search and summarizes findings". It'll do everything in seconds.