The Claude Code Leak Rewrite

The Claude Code Leak Rewrite | Generated by AI

Home 2026.04

Question: What is the ultraworkers/claw-code GitHub repository, and is it worth using?

Answer:

Overview of `ultraworkers/claw-code`

This repository has a dramatic backstory that explains everything about what it is and why it went viral.

The Origin: The Claude Code Leak

On March 31, 2026, Anthropic accidentally shipped a .map sourcemap file inside a Claude Code npm update. Within minutes, the 600k lines of code were mirrored, analyzed, ported to Python and other languages, and uploaded across the internet.

The leak exposed approximately 512,000+ lines of TypeScript, 44 hidden feature flags, an always-on background agent called KAIROS, and a stealth mode designed to hide Anthropic employee contributions to open-source projects.

What `claw-code` Actually Is

claw-code is not a dump of leaked proprietary code. The maintainers carefully avoided storing the original leaked snapshot, focusing instead on a functional Python-first rewrite and a high-performance Rust port.

The repository was born as a mirror that later became a complete rewrite of the Claude Code app, first in Python and later in Rust, and crossed 100,000 stars on GitHub — making it the fastest-growing repo in GitHub history, surpassing even the previous record holder.

The repo is now a Rust-dominant project (92.9% Rust, 7.1% Python) with these main components:

crates/api-client — API client with provider abstraction and streaming
crates/runtime — session state, MCP orchestration, prompt construction
crates/tools — tool manifest definitions and execution framework
crates/claw-cli — interactive REPL and project bootstrap flows

Is It “Good”? An Honest Assessment

Pros:

Massive community traction — 153k stars and 101k forks is unprecedented
Clean-room approach — designed to avoid direct copyright infringement by reimplementing architecture from scratch
Rust port — aims to deliver a faster, memory-safe harness runtime compared to the original TypeScript
Educational value — useful for studying AI agent harness engineering, tool wiring, and orchestration patterns

Cons / Cautions:

The Python workspace is not yet a complete one-to-one replacement for the original system; it still contains fewer executable runtime slices than the archived source.
Legal uncertainty — the legal status of claw-code remains unresolved. No litigation has been filed against the clean-room rewrite as of early April 2026, but Anthropic has not published a formal post-mortem either.
Ownership in flux — the repo itself notes it is temporarily locked during an ownership transfer, with active maintenance happening at ultraworkers/claw-code-parity.
Malware risk in the ecosystem — a malicious GitHub repository used the Claude Code exposure as a lure, distributing an infostealer (Vidar) and a proxy malware (GhostSocks) disguised as a leaked source archive. Always verify what you’re downloading from related repositories.

Bottom Line

claw-code is a fascinating engineering and community phenomenon, useful for learning about AI agent harness design. However, it is not production-ready as a full Claude Code replacement, its legal status is unresolved, and the ecosystem around it has been polluted with malware lookalikes. If you’re a developer curious about agent internals, the Rust port is interesting to explore — but do so cautiously and only from the verified repo.

References:

Back Donate