Skip to content

Create Your First Agent

This guide walks you through creating a custom agent from scratch. By the end, you’ll have a working agent with a personality and tools tailored to your needs.

Make sure you have a running Frona AI instance and are logged in. If not, follow the Quickstart first.

From the chat interface, look for the agent selector in the sidebar. You’ll see the built-in agents listed. To create a new one, click the option to add a new agent.

Fill in the core fields:

  • Name: give your agent a descriptive name (e.g., “Content Writer”, “Data Analyst”)
  • Description: a short summary of what the agent does. This helps other agents understand what to delegate to it.

The model group determines which LLM your agent uses. The available groups depend on your configuration:

  • primary: general-purpose, good for most tasks
  • coding: optimized for code generation and technical tasks
  • reasoning: better for complex analysis and research

Pick the one that matches your agent’s purpose.

Choose which tools your agent can access. Some common combinations:

Research agent: web_search, web_fetch, browser_navigate, browser_extract, remember

Coding agent: cli, read_file, produce_file, remember

General assistant: web_search, web_fetch, remember, delegate

You can always change the tool selection later.

Start a chat with your new agent. Send it a message related to its purpose and see how it responds. Check that:

  • It uses the tone and style you expect
  • It can access the tools you enabled
  • It handles your requests appropriately
  • Start with fewer tools and add more as needed. An agent with too many tools can get confused about which to use.
  • The description matters for delegation. If other agents need to delegate to yours, make the description clear about what it handles.
  • You can update the agent’s identity fields to customize its personality (name, traits, response style).
  • Use the delegate tool on your System Assistant to route tasks to your custom agent automatically.