Getting started

Understanding Fine

New to Fine? Here are some terms and concepts to get familiar with as you learn how to use Fine:


Your workspace is home for your agents, repositories and issues. Invite your team to collaborate on building agents and projects.


The most basic concept in Fine is the Agent. An Agent is an AI teammate that can carry out various coding tasks for you. Agents can automate almost any kind of software development tasks.

Agents can be generic (e.g. Python Agent) or programmed to perform specific kind of task (e.g. Python Unit Tests Agent). You can deploy as many agents as you want in your workspace.


Agents operate on repositories. Repositories are linked to a source codebase (e.g. backend, frontend, mobile).

Local projects

Projects can be linked to a local directory on your computer. This allows agents to make changes to your codebase on your local environment.

Local mode

For privacy and security reasons, Fine does not store your source code. Instead, the client will link your local source code to a proxy. This allows agents to make changes to your codebase on your local environment without Fine storing your code.


Agents need instructions to perform tasks. In Fine, we use Issues to write specifications for agents. They describe the scope of change and provide clear instructions for the agent to follow.

Notebooks are powerful because they leave a trail of documentation for the changes made by the agent. Over time, agents can learn from historical notebooks and improve their performance.

Command palette

The command palette lets you quickly take any action applicable to the current Notebook or context. It is accessible via the ⌘+P key (⇧+P in Windows).

Practice Software 3.0