What is BIGGI?

BIGGI is an open-source AI coding agent built for game development. It runs inside VS Code and combines code editing, tool execution, game-engine workflows, reviewable diffs, and parallel agent sessions.

Use BIGGI for ordinary software work or choose its game-specific agents and Game Loop tools when you are working with Unity, Godot, or a browser game.

One extension, one local runtime

BIGGI ships as a VS Code extension. The extension includes and manages a local runtime that provides agents, tools, sessions, model connections, MCP, indexing, and game integration.

You do not need to install a separate BIGGI CLI, desktop application, or plugin for another IDE. The runtime is an internal part of the extension, even when logs or configuration refer to it as the CLI or server.

What you can do

WorkflowWhat BIGGI provides
Write and change codeChat, file editing, terminal commands, code search, language-server tools, and reviewable diffs
Build gamesSpecialized Gameplay, Game Architect, Tech Art, and Perf agents
Verify gamesRun, inspect, test, screenshot, playtest, and profile supported game projects
Work in parallelAgent Manager sessions with optional git-worktree isolation, terminals, and live diffs
Stay in controlPer-tool permissions, approval prompts, snapshots, cost alerts, and sandbox support
Add project knowledgeAGENTS.md, custom instructions, skills, workflows, semantic indexing, and project memory
Connect other systemsMCP servers, plugins, web search, optional browser automation, and custom providers

How BIGGI works

  1. You open a project in VS Code.
  2. You connect an AI model provider directly from BIGGI settings.
  3. You choose an agent that matches the task.
  4. BIGGI reads the project, proposes or performs approved actions, and reports the result.
  5. You review the conversation, tool output, and diff before keeping the change.

Your source files, sessions, and provider credentials stay on your machine unless you explicitly use a networked provider, web tool, MCP server, browser automation, or sharing feature.

Start here

  1. Check what is available today.
  2. Install the extension.
  3. Connect a model provider.
  4. Run your first task.

If you are already working on a game, continue with the game-development overview. For multiple independent tasks, open Agent Manager.