Quickstart Guide

Build and run your first AI agent in 5 minutes with this step-by-step tutorial

Quickstart Guide

This guide walks you through creating your first AI agent on Flezi AgentBox — from sign-up to execution. You'll have a working agent in under 5 minutes.

Local Setup (CLI)

Want to run Flezi AgentBox locally and connect it to your IDE? The agentbox CLI sets everything up in one command:

curl -fsSL https://raw.githubusercontent.com/agentbox/cli/main/install.sh | bash
agentbox init

This handles Docker, config, LLM keys, and MCP IDE auto-connect. See the full CLI Local Setup Guide.

Web Platform Setup

To use the hosted web platform instead, follow the steps below.

Prerequisites

• A modern web browser (Chrome, Firefox, Edge, or Safari)
• An email address for registration
• An API key from a supported LLM provider (OpenAI, Anthropic, or Google)

Step 1: Create Your Account

Step 2: Set Up Your API Key

To create and run agents, you need to configure at least one LLM provider API key.

Step 3: Open the Builder Lab

Step 4: Build Your Agent

The Builder Lab guides you through three phases. Here's an example building a Code Review Agent based on the BMAD methodology:

Phase 1: Discovery

The Discovery phase defines what your agent does using categorized sticky notes:

| Note Category | Example Content | |---------------|----------------| | Goal (Blue) | Find security vulnerabilities and code quality issues | | Risk (Red) | SQL injection, XSS, missing error handling | | Constraint (Amber) | Output must include file paths and severity levels | | Persona (Pink) | Senior SE, analytical, ultra-succinct |

Review and customize your notes, then click Continue to Logic.

Phase 2: Logic

The Logic phase defines how your agent processes information using a visual DAG editor:

• Service nodes (hexagon) — Load context: "Load Code Context"
• Function nodes (rounded rectangle) — Process: "Analyze Quality Facets"
• Gate nodes (diamond) — Filter: "Severity Filter"
• Data nodes (cylinder) — Output: "Review Report"

Connect nodes with edges to define execution flow. Validation ensures no cycles or disconnected nodes.

# Example: Code Review Agent DAG
dag:
  nodes:
    - { id: "n1", type: "service", label: "Load Code Context" }
    - { id: "n2", type: "function", label: "Analyze Quality Facets" }
    - { id: "n3", type: "gate", label: "Severity Filter" }
    - { id: "n4", type: "data", label: "Review Report" }
  edges:
    - { source: "n1", target: "n2" }
    - { source: "n2", target: "n3" }
    - { source: "n3", target: "n4" }

Click Continue to Output when ready.

Phase 3: Synthesis

The Output phase generates your agent's final blueprint:

• Review the synthesized agent configuration
• Verify the execution pipeline
• Preview sample input/output

Click Publish to make your agent available.

Step 5: Run Your Agent

Input:  "Review this function for security issues:
        function getUserData(userId) {
          const query = `SELECT * FROM users WHERE id = ${userId}`;
          return db.execute(query);
        }"

Output: ## Findings
        ### [CRITICAL] Security: SQL Injection Vulnerability
        **File:** src/index.ts:2
        **Fix:** Use parameterized queries: db.execute(query, [userId])

        ### [LOW] Correctness: Missing Error Handling
        **File:** src/index.ts:3
        **Fix:** Add try-catch for database errors

Step 6: Share and Iterate

Congratulations! You've built and run your first agent. Here's what you can do next:

• Publish to Marketplace: Make your agent discoverable by other users
• Set Pricing: Configure execution pricing in the agent settings
• Add Version Notes: Document changes as you improve your agent
• Create Collections: Group related agents into themed bundles
• Explore BMAD Agents: Check out the BMAD Code Reviewer, Architect Winston, and QA Engineer Quinn on the marketplace for inspiration

Next Steps

Now that you have a working agent, explore these guides to learn more:

• Core Concepts — Understand agents, pipelines, and templates
• Builder Lab Guide — Deep dive into the three-phase workflow
• Marketplace Guide — Learn about discovery, reviews, and storefronts