Advanced·10 min read

Building Your First Swarm

Connect multiple agents together to solve complex problems through Swarm Architecture.

Building Your First Swarm

A single agent is great for simple Q&A. A Swarm is necessary for multi-step, complex workflows where different skills are required.

What is a Swarm?

A swarm typically consists of:

  • Router Agent: The boss. Decides what needs to be done.
  • Worker Agents: The specialists. (e.g., "Python Coder", "Copywriter").
  • Reviewer Agent: The QA. Checks the work before showing the user.

Example: The Content Creator Swarm

Let's build a swarm that researches a topic and writes a blog post.

Step 1: Create the Researcher

Create a new bot. Give it the "Web Search" tool. Prompt it:

"You are a researcher. Given a topic, find the 3 most recent articles and summarize the key facts."

Step 2: Create the Writer

Create a second bot. Prompt it:

"You are a copywriter. Take the facts provided by the researcher and write a 500-word blog post in a casual tone."

Step 3: Link them in HiveLang

In your orchestration script, link them:

ruby
agent ContentSwarm { trigger: "new_topic" actions: [ call Researcher with { topic: input.topic } => facts call Writer with { facts: facts } => draft return draft ] }

Now your swarm will independently research and write!

Quick tip

Use the test pane to iterate quickly. Every change you make is live — no need to save first.

Important

API keys are shown only once. Store them securely and never commit them to version control.

Best practice

Test your bot with edge cases before deployment. Try empty inputs, long messages, and special characters.

Pro tip

Chain multiple specialized bots in a workflow for better results than one general-purpose bot.

Building Your First Swarm