Building AI Agents with the Agent Builder & Flow Designer

CodeAgento Team2026-05-1610 min read

AI agents are programs that use LLMs to reason, make decisions, and take actions. CodeAgento provides a complete toolkit for building, testing, and debugging agents, no infrastructure required.

What You'll Build

By the end of this guide, you'll have a working AI agent that can:

Answer questions using custom tools

Follow a multi-step workflow

Be debugged step-by-step

Step 1: Design Your Agent

Open the Agent Builder from the sidebar. Here you'll configure:

**System Prompt**: The instructions that define your agent's behavior

**Model Settings**: Temperature, max tokens, and model selection

**Tools**: The functions your agent can call

Start with a clear system prompt. For example:

You are a helpful code review assistant. When given code, analyze it for bugs, security issues, and style problems. Use the available tools to look up documentation when needed.

Step 2: Create Tools

Switch to the Tool Designer to define the tools your agent can use. Each tool needs:

A name and description (the AI reads these to decide when to call the tool)

A Zod schema defining the parameters

An execution handler

Step 3: Build a Flow

The Flow Designer lets you create multi-step agent workflows visually:

1. Drag agent, tool, condition, and prompt nodes onto the canvas

2. Connect them with edges to define the flow

3. Configure each node's properties in the side panel

Step 4: Test & Debug

Use the Prompt Playground to test individual prompts, then the Agent Debugger to trace full executions:

See every step: input, agent reasoning, tool calls, tool results, and final output

Inspect timing for each step

Identify where the agent went wrong

Step 5: Export

Once your agent works, export the configuration as JSON or scaffold a complete MCP server with the MCP Server Builder.

Start building your first agent.