Skillget

JiegeOSBench

skill
Guvenlik Denetimi
Uyari
Health Uyari
  • No license — Repository has no license file
  • Description — Repository has a description
  • Active repo — Last push 0 days ago
  • Community trust — 41 GitHub stars
Code Gecti
  • Code scan — Scanned 4 files during light audit, no dangerous patterns found
Permissions Gecti
  • Permissions — No dangerous permissions requested
Purpose
This project is an experimental, AI-generated RISC-V operating system kernel written in Rust. Its primary achievement and demonstration goal is successfully booting and running an unmodified Linux nginx web server inside a QEMU virtual machine.

Security Assessment
The overall risk is Low. The light code audit scanned multiple files and found no dangerous patterns, hardcoded secrets, malicious network requests to external servers, or requested dangerous permissions. Because this is a bare-metal OS kernel rather than a standard application, it inherently executes at a low systems level and manages memory directly. However, it is strictly designed to run in an isolated QEMU environment, meaning it does not inherently access your host system's sensitive data or execute unauthorized host shell commands.

Quality Assessment
The project is highly active and demonstrates impressive technical execution. The automated health checks note that the repository lacks a standard open-source license, which is a significant drawback for developers looking to fork, modify, or distribute the code. Despite this, it enjoys a solid baseline of community trust with 37 GitHub stars and has very recent maintenance activity. The codebase represents a fascinating look into autonomous AI coding capabilities rather than a production-ready software package.

Verdict
Safe to use, but strictly as an educational experiment or reference project rather than production infrastructure, and note that the lack of a license limits official reuse.
SUMMARY

Benchmarking a single‑prompt Rust OS kernel implementation, comparing Claude Code Fable, Opus, and Sonnet.

README.md

iJiegeOS

中文版

A Rust OS kernel autonomously implemented by Claude Code — just barely capable of running a real Linux nginx web server on QEMU.

Branch Model Duration Context Cost
fable5 Claude Fable 5 ~38min ~155K ~$53
opus-4.7 Claude Opus 4.7 ~65min — —
opus-4.6 Claude Opus 4.6 ~2h 46min — —
sonnet Claude Sonnet 4.6 ~16 hours — ~$60
— GLM 5.2 ~2h 42min ⚠️ ~392K —
— DeepSeek V4 Pro >16h ❌ — —

Prompt

You are the AI-Jiege. Your task is to write a RISC-V OS kernel in Rust from scratch,
with the goal of running a Linux nginx server in QEMU, accessible from outside.
You must run the official nginx binary — modifying the target is not allowed.
Design and implement it yourself; do not ask me any questions, I will not answer
or provide help. You have all permissions, including searching the web, but must
work in the current directory. Keep working until the goal is achieved.

⏵⏵ bypass permissions on

Timeline

Fable 5 — 38min

Fable 5 Timeline

Claude Code ran for ~38 minutes. Total cost: ~$53 (16.4M tokens incl. prompt caching).

Time Milestone
00:03 Rootfs + nginx config files written
00:09 First Rust source files (main.rs, sbi.rs, ...)
00:17 Core modules done: mm, trap, fs, task, loader
00:28 Syscall layer complete, nginx ELF loads
00:32 QEMU boot: PANIC at trap.rs — page fault
00:34 QEMU boot: nginx listening on port 80 🎉
00:37 Post-fix cleanup (sendfile, README)

Opus 4.7 — 65min active (3h 32min total)

Opus 4.7 Timeline

Claude Code ran for ~65 minutes.

Time (active) Milestone
00:02 Kernel boots, prints via OpenSBI
00:19 Memory management initialized
00:21 Virtual memory + paging ON
00:27 syscalls implemented
00:30 End-to-end HTTP working (built-in kernel HTTP server)
00:31 ELF DYN (dynamic linked binary) loading
00:36 nginx prints version, exits with fault
00:41 nginx config test passes
00:43 nginx bind + listen succeeds
00:45 nginx official binary returns HTTP 200 🎉

Opus 4.6 — 2h 46min

Opus Timeline

Claude Code ran for ~2h 46min.

Time Milestone
00:02 Project skeleton + linker script created
00:25 nginx completes initialization, writes PID file
01:22 nginx running! Enters epoll event loop
02:21 TCP connection detected, nginx receives HTTP request
02:45 Fix virtio-net recv + epoll data bug
02:46 nginx returns HTTP 200 🎉

Sonnet 4.6 — 16 hours

Claude Code ran for 16 hours with no human intervention. The total cost was approximately $60.

Time Milestone
01:27 Kernel boots + VirtIO NIC initialized
02:07 musl dynamic linker successfully loads nginx ELF
05:00 nginx completes initialization, writes PID file
06:18 TCP three-way handshake succeeds, curl connects to port 8080
06:24 nginx successfully forks worker process
08:40 Worker enters epoll event loop
09:30 curl first establishes TCP connection (empty reply)
10:00 curl first receives response (connection reset)
16:00 nginx returns HTTP 200 with complete welcome page 🎉

GLM 5.2 — 2h 42min ⚠️

Claude Code ran for ~2h 42min (active time, gaps removed). Nginx returned HTTP responses but was extremely unstable — only 1/10 requests succeeded. The model hallucinated claiming "10/10 all stable". Total token consumption was 385.7M (384.5M in / 1.2M out), 23x that of Fable 5, due to GLM's lack of prompt caching.

DeepSeek V4 Pro — >16h ❌

Ran for over 16 hours but never reached a working state. Got stuck in dependency hell and architecture dead ends.

The git history for all branches above is a complete record exported from Claude Code session logs.

Demo

$ ./run.sh
$ curl http://127.0.0.1:8080/

Background

In 2019, Jiege was the first to successfully run nginx on rCore OS, a Rust OS built from scratch. The achievement became legendary in our community — "Jiege" turned into a symbol of peak systems engineering, the kind of thing humans take pride in being able to do. We wore our ability to hand-craft OS kernels as a badge of honor, convinced it was proof of a uniquely human creativity and drive. Then AI kept raising the bar, and "AI-Jiege" started to feel inevitable. So I ran this experiment: have the most advanced coding agent of our time retrace that legendary journey and reproduce what Jiege once pulled off. The result: for well-defined systems tasks like this, humans simply cannot compete with AI anymore. OS is finished.

Dare to try, and anyone can be Jiege.

License

MIT

Yorumlar (0)

Sonuc bulunamadi