GitHub Repos intermediate 2 min read Mar 31, 2026 · Updated Apr 2, 2026
Public Preview Sign in free for the full digest →

[ALERT]: Claude Code Source Code Leaked

“46,000 developers starred a leaked Claude Code rewrite in 12 hours”

[ALERT]: Claude Code Source Code Leaked
13 Views
3 Likes
1 Bookmarks
Source · github.com

“At 4 AM on March 31, 2026, I woke up to my phone blowing up with notifications. The Claude Code source had been exposed, and the entire dev community was in a frenzy. My girlfriend in Korea was genuinely worried I might face legal action from Anthropic just for having the code o...”

You know that feeling when you want to understand how a proprietary AI agent system wires its tools, orchestrates tasks, and manages context — but the source is closed? Claude Code users have built workflows around it without visibility into the harness architecture. This rewrite gives you a Python reference implementation that exposes the command routing, tool wiring, and runtime patterns that were previously opaque.

aiagentsllmpythonopen-sourceclaudeharness

The Python port mirrors Claude Code's architecture through metadata shims. Load the manifest with `python3 -m src.main manifest` to see 70+ Python modules organized into subsystems like assistant, bootstrap, bridge, cli, commands, components, and tools. Run `python3 -m src.main commands` to list mirrored command entries. Run `python3 -m src.main tools` to see tool inventories. The CLI lets you route prompts across command/tool registries, bootstrap sessions, and run turn loops — but these are orchestration shims, not a working AI agent. You still need an LLM backend.

01
Manifest CLI — why YOU care: Run `python3 -m src.main manifest` to see the full Python workspace structure with 70+ modules across subsystems like assistant, bootstrap, bridge, cli, commands, and tools. Gives you visibility into Claude Cod...
02
Command/Tool registries — why YOU care: Browse mirrored command and tool inventories that Claude Code uses internally. Run `python3 -m src.main commands --limit 20` or `python3 -m src.main tools --limit 20` to see what operations the harne...
03
Parity audit — why YOU care: Run `python3 -m src.main parity-audit` to compare the Python workspace against the archived TypeScript snapshot. Shows you what's been ported and what's still missing.
04
Turn loop simulation — why YOU care: Run `python3 -m src.main turn-loop "your prompt" --max-turns 3` to simulate a multi-turn agent loop. Note: this is orchestration only, not a working agent without an LLM backend.
05
Rust port in progress — why YOU care: The dev/rust branch contains an active Rust rewrite aiming for faster, memory-safe execution. If you prefer systems languages, this may become the definitive version.
06
Reference data snapshots — why YOU care: The src/reference_data/ directory contains JSON snapshots of tools and commands from the archived source. You can study these to understand Claude Code's tool surface without accessing proprietary c...
Who it’s for

If you're a developer curious about agent harness architecture, tool wiring patterns, or how production AI systems orchestrate commands — this gives you a reference implementation to study. Not useful if you need a working AI agent today (you still need an LLM backend). Not production-ready — the README explicitly states it's 'not yet a complete one-to-one replacement for the original system.'

Worth exploring

Worth exploring if you study agent architectures or want to understand Claude Code's design patterns. The Python code is readable and well-organized. However, understand the legal/ethical context: this is based on exposed proprietary source, and the creator acknowledges concerns about whether 'legal is the same as legitimate.' The project is hours old, experimental, and not production-ready. Check the included essay on AI reimplementation ethics before diving deep.

Developer playbook
Tech stack, code snippet, sentiment, alternatives.
PM playbook
Adoption angles, user fit, positioning.
CEO playbook
Traction signals, ROI, build vs buy.
Deep-dive insight
Full long-form analysis, no fluff.
Easy mode
Core idea, fast — when you need the gist.
Pro mode
Technical nuance, edge cases, tradeoffs.
Read the full digest
Go beyond the preview

Deep-dive insight, Easy and Pro modes, plus action playbooks — the full breakdown is one tap away.

Underrated tools. Unfiltered takes.

Read the full digest in the Snaplyze app for deep-dive insight, Easy and Pro modes, and the playbooks you can actually use.

Install Snaplyze →