[go: up one dir, main page]

AI-autonomous software engineering experiment

Can AI build and maintain
a browser?

ZeroWeb tests whether AI can sustain a large systems project. One human designs, decides and verifies; AI develops the Rust browser and its product code, with almost no code written by hand.

Experimental · Not ready for everyday browsing or production use

EXPERIMENT / 01STATUS: RUNNING
Code commits15,100+developed autonomously by AI · one human: design, decisions, verification
full WPT pass rate78.85%1,789,714 / 2,269,629 upstream subtests passing, measured against the entire WPT corpus · see the data →

The experiment

The browser is the test case.

A browser is large enough to test more than code generation. It requires long-term architecture work, standards implementation, debugging, security boundaries, graphics, and cross-platform integration. Verification runs throughout.

What AI does

AI analyzes tasks, changes code, adds tests, runs checks, and investigates failures. The work continues across modules.

How the work is constrained

Specifications, module boundaries, review rules, CI, WPT, reftests, and browser comparisons decide whether a change is acceptable.

What counts as progress

A larger repository is not the goal. Features must compile, run, and pass the checks appropriate to the behavior being changed.

Test subject
A cross-platform browser
Implementation
Browser core and product code in Rust
Validation
Builds, tests, WPT and browser comparisons

The result is judged by working software and reproducible checks, not generated line count.

Runnable code

Two entrypoints into the same project.

01
◌

ZeroBrowser

The desktop entrypoint used to integrate and test the browser subsystems. It is not yet an everyday browser.

View browser application
02
⌘

ZeroWebView

A Rust API for embedding the page pipeline in host applications. The API and compatibility surface are still evolving.

View WebView library

Implementation scope

From request to rendered output.

The repository separates the systems that load, parse, lay out, and render a page so each part can be implemented and verified independently.

  1. 01NetworkResources & navigation
  2. 02DOM + CSSDocument & rules
  3. 03StyleCascade & computed values
  4. 04LayoutGeometry & flow
  5. 05RenderPaint & composite
Read the architecture guide

Engineering method

AI output is checked like engineering work.

RS

Browser code in Rust

The browser core, host layers, and product code are written in Rust. External crates provide bounded low-level capabilities.

◇

Standards-driven

Web behavior is verified against WPT, CSSWG reftests, and browser comparisons.

CI

Checks are required

Formatting, static analysis, unit tests, integration tests, and platform CI guard the repository.

LOG

Progress has sources

Updates link to commits and roadmap items so the experiment can be inspected rather than taken on trust.

Current status

Useful as an experiment, incomplete as a browser.

Core subsystems and runnable entrypoints exist. Real-site compatibility, complete DOM and Web APIs, and everyday browser behavior remain active work.

See the detailed roadmap
Implemented foundationsDOM, CSS, style, layout, rendering, network and security basics
In progressBrowser shell, rendering compatibility, DOM and Web API coverage
Not yet completeWebGL / WebGPU, SVG documents, animation fidelity and real-site behavior

Run the code

Start with a local build.

The repository includes a desktop browser entrypoint and a smaller WebView rendering demo. Platform prerequisites are documented in the README.

ZeroBrowser

Launch the desktop browser entrypoint.

$ make browser

WebView demo

Run the smallest rendering pipeline demo.

$ cargo run --bin webview-demo

These commands build experimental software; see the README for platform prerequisites and V8 setup.

Read the setup guide →

Project journal

Progress, with sources.

A running record of what changed, what is still in motion, and where each update comes from.

Rendering failures fall to 884

33 previously closed goals were archived, and 15 more reached a terminal state. Rendering fixes reduced full-corpus failures from 1,104 to 884; media timing and browser stability also improved.

Read update →

View all updates

Inspect the experiment

Read the code.
Check the evidence.

The repository, development history, tests, roadmap, and weekly updates are public. Contributions are expected to meet the same verification rules.