Enterprise Flutter: The Tom Framework

A Dart/Flutter enterprise & AI ecosystem — explore every module and component, what it does, and how far along it is.

Go and see the Tom Overview
0
released
21
published
16
in the works
5
not started

Tom — Enterprise Flutter/Dart AI Platform

Tom is an Enterprise Flutter and Dart AI platform: an end-to-end ecosystem for building AI-powered applications in a single language. It pairs a Flutter front end and a Dart server with agentic AI, a Dart scripting engine (D4rt / Dartscript), runtime reflection, and a spec-driven software factory that turns specifications into tested code. These are the areas Tom covers today.

Tom Reflection: Runtime Reflection & Source-Code Analysis for Dart

Dart reflection without dart:mirrors: runtime reflection lets a Flutter or Dart app inspect types, read and write fields, construct objects, and invoke methods by name, while build-time source analysis emits a serialized model of any codebase's API. It is the metaprogramming foundation that lets Tom apps assemble themselves at runtime and powers the API reference on this site.

Tom D4rt: Dart Scripting for Agentic UIs and On-the-Fly Scripting

D4rt (Dartscript) is a sandboxed Dart interpreter that runs and generates Dart code at runtime — no ahead-of-time compilation. It is the engine behind AI scripting, agentic UI that renders live inside a Flutter client, and on-the-fly scripting of business logic, all inside a permission-guarded sandbox that blocks untrusted file, process, and network access.

Tom Brain: Autonomous AI Agent with Graph Memory

Tom Brain is an autonomous AI agent that speaks the OpenAI-compatible LLM API but thinks in a structured loop — analysis, recall, self-criticism, and revision — over a self-organising graph memory. It runs agent procedures as sandboxed D4rt scripts, turning a raw large language model (LLM) into a reasoning engine that produces answers, not just tokens.

Tom Software Factory: Spec-Driven AI Code Generation (DocSpecs / CodeSpecs)

The software factory is a staged, AI-assisted code-generation pipeline that goes from idea to specification (DocSpecs / CodeSpecs) to working, tested Dart code — with full traceability from a business requirement to its implementation and its test. It is spec-driven development for AI: explicit specs, examples as specs, and each generated unit verified before the next is built on it.

Tom Build Tools: Dart CLI Build Orchestration & DevOps

buildkit is Tom's build orchestrator for the whole Dart/Flutter workspace — the main build tool, intended for all kinds of projects, not just the framework. Built on the tom_build_base CLI framework, it runs build pipelines and commands (versioner, compiler, runner, publisher) and drives git across every repository in dependency order — the DevOps line of the platform.

Tom Core: Enterprise Flutter App Framework (Client + Server)

Tom Core is the Enterprise Flutter application framework every Tom app is built on: a Flutter client and a Dart server joined by a typed interface on a shared kernel. It guarantees, by construction, a resource-driven configurable UI, full internationalization (i18n), uniform authentication and authorization, and uniform deployment, operations, and cloud provisioning for every app.

Tom for VS Code: Dart Scripting Inside the Editor

The Tom VS Code extension lets you automate and extend VS Code in Dart — the same language as the framework. A thin TypeScript host drives a Dart bridge that runs VS Code scripting through the D4rt interpreter, giving Dart scripts full access to the VS Code API for AI-assisted, editor-native development.

A note on this project and this site

Tom is built as a spec-driven AI-development project — a deliberate man-machine collaboration in which a human author and AI agents work together, from specification through implementation, test, and documentation.

The core librariestom_basics, tom_core_kernel, tom_core_flutter and tom_core_server — are hand-authored, then AI-documented and AI-tested. The rest of the framework's business logic is produced through the spec-driven AI flow.

This website is generated entirely by AI, with three deliberate exceptions written by hand: the About page, the Process page, and this introductory note.

We will add the remaining components to the site over the next few weeks. Some of them are not yet ready to use — when a component becomes ready, we will say so with a release note here on the website.

A frank caveat: we know not every document is perfectly to the point, and a few may even be slightly misleading. We have published them anyway because we believe they are a strong starting point and good enough for an initial launch. They will sharpen as the framework stabilises.

News

15.06.2026 — Released: Tom Basics and Tom D4rt

The first two component families are now public. Tom Basics delivers the foundation every Tom project builds on — the build toolchain (tom_build_base), console and network helpers, crypto, chat tooling and the package/analyzer scanners. Tom D4rt ships the Dart-in-Dart interpreter and its ecosystem — the AST model and generator, the Flutter bridge, and the DCli scripting surface.

Browse them under Components, or jump straight into their guides and API in the Docs and API trees.

Areas

Start with an area — the big picture — then drill into its components and modules.

Reflection

1 member
Reflection is the base technology that lets a Tom application assemble itself at runtime and lets tools read a codebase's API without re-running the analyzer.

D4rt Interpreter

1 member
The D4rt Interpreter is where Tom runs and generates Dart code at runtime, inside a permission-guarded sandbox, without ahead-of-time compilation.

Tom Brain Agent

1 member
Tom Brain is the intelligence layer of the Tom framework: from the outside it looks like an ordinary OpenAI-format LLM API server, but inside it is a single always-running engine that runs procedures — sandboxed D4rt scripts — over a self-organising, "dreaming" graph memory.

Docspecs/Codespecs Software Factory

0 members
The Docspecs/Codespecs Software Factory is the staged, AI-assisted process by which Tom applications are created — from an initial idea, through structured specifications, to working, tested code — with complete traceability from a business requirement to the implementation and the test that covers it.

Build tools

2 members
Build tools are the command-line machinery that develops, builds, tests, documents, and deploys the Tom framework and the applications made with it.

Core Application Framework

1 member
The Core Application Framework is the foundation every Tom application is built on: a Flutter client and a Dart server joined by a typed interface, both resting on a shared kernel.

VS Code Extension and VS Code Scripting

1 member
This area is how the Tom framework is developed inside the editor: a VS Code extension that hosts a Dart bridge, so framework tooling and automation can be written in Dart — the same language as the framework itself — with full access to the VS Code API.

Components

Each component is a family of modules — hover for its status mix, click to open its modules.

Basics

basics · 12 modules
20.6k loc · 1.2k tests · 18.9k test LOC
The foundation layer beneath the Tom stack: a minimal-dependency utility floor and the unified CLI/build framework, surrounded by focused libraries for console, network, crypto, chat, and documentation/workspace tooling.
0 rel 7 pub 2 wip 3 ns

Core

core · 9 modules
499.6k loc · 3.1k tests · 259.3k test LOC
The Flutter-client and Dart-server foundation every Tom application is built on, over a shared kernel and the tom_basics floor. It guarantees a resource-driven UI, i18n, authentication, four-state authorization, and uniform deployment by construction.
0 rel 0 pub 8 wip 1 ns

D4rt

d4rt · 13 modules
1.7M loc · 12.8k tests · 3.2M test LOC
D4rt runs Dart source at runtime in a permission-guarded sandbox. It is the dynamic-execution engine that lets Tom's AI scripting and flow layers generate and run code safely.
0 rel 11 pub 2 wip 0 ns

Reflection

reflection · 5 modules
28.7k loc · 420 tests · 285.5k test LOC
The reflection layer gives Tom programs runtime introspection and dynamic invocation, plus analyzer-based extraction of a codebase's API. It is the base technology behind auto-assembling Tom applications at runtime.
0 rel 2 pub 2 wip 1 ns

Vscode

vscode · 3 modules
97.5k loc · 1.5k tests · 19.5k test LOC
The VS Code extension ecosystem used to develop the Tom framework. A TypeScript extension hosts a Dart bridge that runs Dart scripts inside the editor with full VS Code API access, driven by a typed Dart scripting layer.
0 rel 1 pub 2 wip 0 ns
Three ways in: start with an Area above, browse every Component , or jump straight to a file via the Docs / API trees in the top nav.