Quick Start Guide

Welcome to the HIVE Protocol Quick Start Guide! This guide will walk you through creating your first AI agent swarm from scratch. By the end, you'll have a working multi-agent system capable of handling complex tasks through collaborative AI. The entire process takes about 10-15 minutes, and you'll gain hands-on experience with the core features of the platform.

We'll build a practical example: a research and writing team consisting of two specialized agents. This demonstrates the power of multi-agent collaboration while keeping things simple enough for a first project. Once you understand these basics, you'll be ready to create more sophisticated swarms tailored to your specific needs.

Prerequisites

Before you begin, make sure you have the following ready:

Required:

• A HIVE Protocol account (sign up at the homepage if you haven't already)
• An API key from at least one AI provider

Supported AI Providers:

If you don't have an API key yet, we recommend starting with OpenAI as they offer free trial credits and have straightforward setup. Anthropic is an excellent choice for writing-focused tasks, while Google AI offers generous free tiers and excellent long-context capabilities.

Step 1: Sign In and Navigate the Dashboard

After signing in, you'll land on the Dashboard. Take a moment to familiarize yourself with the navigation sidebar on the left:

• Dashboard: Overview of your swarms, agents, and recent activity
• Swarms: Manage your agent swarms and conversations
• Agents: Create and configure individual AI agents
• Tools: Define custom capabilities for your agents
• Integrations: Connect AI providers and external services
• Settings: Account preferences and security settings

The Dashboard shows your key statistics at a glance, including active swarms, total agents, and recent activity. You'll also see quick-create buttons to get started fast.

Step 2: Connect Your AI Provider

Before creating agents, you need to connect at least one AI provider. Navigate to Settings in the sidebar, then click on the Integrations tab. You'll see cards for each supported provider.

To add your OpenAI API key:

1. Click the OpenAI card
2. Paste your API key in the secure input field
3. Click Save Integration
4. You should see a success message confirming the connection

Tip: Your API key is encrypted before storage using AES-256
encryption. We never log or display your full key after saving.

The integration status will show "Connected" with a green indicator. You can test the connection by clicking "Test Connection" which verifies your key is valid and has available credits.

Troubleshooting Connection Issues:

If you see an error when saving your key, check these common issues:

• Ensure the key is copied completely (no leading/trailing spaces)
• Verify your API account has available credits or an active billing plan
• Check that you're using an API key, not a user token
• For OpenAI, ensure the key hasn't been revoked in your account settings

Step 3: Create Your First Agent

Now let's create our first agent—a Research Specialist. Navigate to Agents in the sidebar and click the Create Agent button in the top right.

Fill in the agent configuration:

Name: Research Specialist

Role: Researcher

Framework: OpenAI (or your connected provider)

Model: gpt-4o (recommended for best results)
       gpt-4o-mini (budget-friendly alternative)

System Prompt:
You are an expert research specialist with a talent for finding,
synthesizing, and presenting information clearly. Your approach:

## Core Responsibilities
- Gather comprehensive information on requested topics
- Evaluate source credibility and note any limitations
- Synthesize findings into clear, structured summaries
- Highlight key insights and unexpected discoveries

## Communication Style
- Be thorough but concise
- Use bullet points and headings for clarity
- Cite your reasoning and acknowledge uncertainty
- Proactively suggest related topics worth exploring

## When Collaborating
- Share relevant findings with other agents promptly
- Ask clarifying questions if the request is ambiguous
- Offer multiple perspectives on complex topics

Click Create Agent to save. You'll see your new agent appear in the agents list with a status indicator.

Step 4: Create a Second Agent

To demonstrate multi-agent collaboration, let's create a complementary agent. This one will take research findings and create polished content from them.

Click Create Agent again and configure:

Name: Content Writer

Role: Writer

Framework: Anthropic (if available) or OpenAI
           Claude excels at writing tasks

Model: claude-sonnet-4-20250514 (excellent for writing)
       gpt-4o (great alternative)

System Prompt:
You are a skilled content writer who transforms research and ideas
into engaging, well-structured content. Your strengths:

## Writing Approach
- Create clear, engaging prose that flows naturally
- Structure content with logical progression
- Adapt tone and style to the intended audience
- Balance comprehensiveness with readability

## Content Standards
- Use active voice and strong verbs
- Vary sentence length for rhythm
- Include relevant examples and analogies
- Create compelling openings and satisfying conclusions

## Collaboration Style
- Build upon research provided by other agents
- Ask for clarification on audience or purpose
- Offer multiple approaches when appropriate
- Accept feedback gracefully and iterate

Save this agent. You now have two specialized agents ready to work together.

Step 5: Create Your First Swarm

With agents ready, it's time to create a swarm—the environment where agents collaborate. Navigate to Dashboard and click Create Swarm, or go to Swarms and click the create button.

Configure your swarm:

Name: Research & Content Team

Task Description:
A collaborative team that researches topics thoroughly and
produces well-written content. The Research Specialist gathers
and synthesizes information, then the Content Writer transforms
findings into polished articles, summaries, or reports.

Add your agents to the swarm:

1. Click "Add Agent" in the agent selection area
2. Select "Research Specialist" from your agents list
3. Click "Add Agent" again
4. Select "Content Writer"

You should see both agents listed in your swarm configuration. Click Create Swarm to finalize.

Step 6: Understanding the Swarm Interface

Your new swarm opens to a conversation view with several key areas:

Main Chat Area (Center)

This is where the conversation happens. You'll see messages from yourself and each agent, with clear labels indicating who said what. Agent messages include verification badges showing they're cryptographically signed.

Agent Panel (Right Sidebar)

Shows all agents in the swarm with their current status. You can see which agents are active, their roles, and quick stats. Click an agent to see their full configuration.

Context Blocks (Collapsible Section)

Add shared information that all agents can access. This is perfect for project requirements, style guides, or reference materials.

Input Area (Bottom)

Type your messages here. The "@" mention feature lets you direct messages to specific agents, though typically you'll let the swarm coordinator decide which agent should respond.

Step 7: Have Your First Conversation

Let's put your swarm to work! Type a message in the input area and press Enter or click Send.

Try this prompt:

I need to create a comprehensive guide about "best practices for
remote team management." Research Specialist, please gather the
key insights and frameworks used by successful remote-first
companies. Content Writer, once we have the research, please
help structure this into an engaging article outline.

Watch as your agents spring into action! You'll see:

1. Typing indicators showing which agent is processing
2. The Research Specialist responding first with gathered insights
3. The Content Writer building on that research with an outline
4. Verified badges on each message confirming authenticity

The conversation flows naturally, with agents building on each other's contributions. This is the power of multi-agent collaboration—specialized expertise working together.

Step 8: Working with Context Blocks

Context Blocks help maintain shared understanding across your swarm. Let's add one to guide our content style.

Click "Add Context" and configure:

Name: Content Style Guide

Priority: High

Content:
## Target Audience
- Remote team managers and executives
- HR professionals implementing remote policies
- Team leads transitioning to remote-first culture

## Tone & Style
- Professional but approachable
- Evidence-based with practical examples
- Actionable advice over theory
- Global perspective (avoid US-centric assumptions)

## Content Requirements
- Include statistics where available
- Mention both benefits and challenges
- Provide tool recommendations where relevant

Save the context block. Now all agents in this swarm can reference this information when creating content. You can update context blocks at any time as your project evolves.

Step 9: Exploring Collaboration Modes

HIVE Protocol offers three collaboration modes that change how you interact with your agents. Access these from the swarm settings or the mode selector in the interface.

Observe Mode

In Observe mode, agents collaborate autonomously based on the initial task. You watch the conversation unfold without direct intervention. This is great for:

• Understanding how agents approach problems
• Letting complex discussions develop naturally
• Minimal-touch workflows after initial prompting

Collaborate Mode (Default)

Collaborate mode balances autonomy and control. Agents consider your input as suggestions rather than commands. Perfect for:

• Guiding direction while allowing agent creativity
• Providing feedback that shapes the conversation
• Most everyday use cases

Direct Mode

In Direct mode, agents follow your instructions precisely. Use this for:

• Critical tasks requiring exact specifications
• Correcting agents or redirecting conversations
• Tasks with strict requirements

Try switching modes to see how the conversation dynamic changes!

Step 10: Next Steps and Advanced Features

Congratulations! You've created your first multi-agent swarm and experienced the basics of HIVE Protocol. Here's where to go next:

Enhance Your Agents

• [Creating Agents](/docs/agents/creating-agents): Learn advanced agent configuration
• [System Prompts](/docs/agents/system-prompts): Master the art of effective prompting
• [Agent Tools](/docs/agents/agent-tools): Give agents real-world capabilities

Master Swarm Orchestration

• [Agent Coordination](/docs/swarms/agent-coordination): Advanced multi-agent patterns
• [Context Blocks](/docs/swarms/context-blocks): Effective knowledge management
• [Human-in-the-Loop](/docs/swarms/human-in-loop): Control strategies

Connect and Extend

• [Webhooks](/docs/integrations/webhooks): Automate workflows
• [API Reference](/docs/api/authentication): Build custom integrations

Troubleshooting Tips

Agents not responding:

• Check that your AI provider integration is connected
• Verify you have available API credits
• Try refreshing the swarm page

Responses seem slow:

• Response time varies by model (GPT-4 is slower than GPT-3.5)
• Complex prompts take longer to process
• Check your internet connection

Unexpected agent behavior:

• Review the system prompt for clarity
• Ensure context blocks don't have conflicting instructions
• Try being more specific in your messages

Messages show as unverified:

• This is rare but can indicate a system sync issue
• Refresh the page to re-verify signatures
• Contact support if the issue persists

Quick Reference

You're now ready to build powerful AI workflows with HIVE Protocol. Remember, the best way to learn is by experimenting—try different agent configurations, swarm compositions, and collaboration modes to discover what works best for your use cases.

Happy building!