Free Pascal · Lazarus · LCL

Your desktop coding workspace

Lazy Coding Agent brings project navigation, an editor, chat sessions and utility panels together in a native desktop application.

Ask, Plan and Agent modes expose different tool permissions. Review the application's recorded run activity and tool inputs/outputs in its execution details.

Packaged builds are on GitHub Releases. The README describes AI/model execution as in active integration; treat early releases accordingly.

Futuristic neon owl emblem for Lazy Coding Agent

Lazy Coding Agent

Desktop workspace

Projects, editor, agent and terminal in one window

The main application window brings workspace navigation, an open plan document, an agent response, and the integrated terminal together.

Lazy Coding Agent desktop window with workspace navigation, a chat session, an agent response, a plan document, and an integrated terminal.
The main application window shows workspace navigation, a chat session with an agent response, an open plan document, and the terminal.
Detached workspace panels

Detached panels in floating windows

The screenshot shows the Chat / AI Agent and Utility Panel windows separately beside the workspace window.

Lazy Coding Agent workspace with separate Chat / AI Agent and Utility Panel windows and an open plan document.
Separate Chat / AI Agent and Utility Panel windows are visible alongside the workspace window.
Reviewed in source and help

A workspace with visible boundaries

These are implementation descriptions from reviewed application files and help, not a claim that every feature is finished or user-tested on your machine.

Workspace

Navigate, edit and chat

Reviewed documentation describes a project tree, single-file editor, Markdown preview, chat, and dockable utility panels.

Workspace details →

Mode-gated tools

Choose how the assistant acts

Ask is read-oriented, Plan can save plan files, and Agent can use write, patch and shell tools. Workspace boundaries apply by default.

Modes and boundaries →

Execution detail

Inspect recorded activity

Session logs and execution detail expose application-recorded context and tool I/O; they are not a display of hidden model reasoning.

Agent loop details →

Maturity and availability: Installers for Windows, Linux, macOS, and arm64 (including Raspberry Pi) are published on GitHub Releases. The README still describes AI/model execution as in active integration—downloads are available, but the product is evolving.