新文章一种更接近工程师思考方式的 AI 记忆架构阅读全文 →

The AI Brain for
Engineering Systems.

Operator X02 · Coding is Art. Feel it. Enjoy it.

Not another AI coding assistant. Operator X02 understands your code, hardware, build system and architecture — so you can build, debug and maintain whole engineering systems, not just files.

main.rs — operator-x02 AI ACTIVE
Project
main.rs
nvidia_commands.rs
arduino_commands.rs
ssh_manager.rs
Hardware
jetson_orin.rs
pi_remote.rs
1use tauri::Builder;
2use nvidia_commands::*;
3
4// Jetson GPU stats — real-time tegrastats stream
5#[tauri::command]
6async fn get_gpu_stats(host: String)
7  -> Result<JetsonStats, String> {
8  let stats = tegrastats_stream(&host).await?;
OPERATOR X02 AI
Context: main.rs · Jetson Orin Nano
You
Add real-time Jetson GPU stats streaming to main.rs
Operator X02 AI
I'll create an async command that streams tegrastats over SSH, add the nvidia_commands module, and wire it into main.rs.
Created nvidia_commands.rs
Wrote get_gpu_stats()
Updated main.rs
Rust 1.76· Jetson: COM3· AI: Claude· Build OK
~95 MB
Memory idle
vs ~280MB VS Code
1.8s
Cold start
vs 3.2s VS Code
7
AI providers
including offline
28
ADB commands
Android native
MIT
License
free forever
The one thing to remember

Operator X02 builds an AI understanding of your entire engineering projectnot just your code.

The whole lifecycle

Engineering doesn't stop at code.

Real projects move through a full lifecycle — and most AI tools only touch the middle of it. Operator X02 understands every stage.

📋
Requirement
01
🏛️
Architecture
02
⌨️
Implementation
03
🛠️
Build
04
🚀
Deploy
05
🧪
Testing
06
🔧
Maintenance
07
Operator X02 understands all of them — not just implementation.
Why Operator X02?

Engineering,
not just coding.

Most AI tools help write code. Operator X02 helps engineers understand and develop complete systems — software, hardware, build system and architecture together, not one file at a time.

Generate codeUnderstand entire projects
Single file contextWhole project context
Chat assistantEngineering partner
Guesses your hardwareReads your schematic ↗
v1.5.8 · live Watch the AI read a real Arduino shield — components, nets, pinouts
Engineering taskOperator X02Typical AI IDE
Generate code
Understand architectureLimited
Build debuggingLimited
Hardware deployment NativePlugin
Read a schematic or datasheet Native
Project knowledgeLimited
Engineering workflowPartial
Comparing value across the workflow — not a feature checklist.
FeatureOperator X02VS CodeCursor
Memory idle~95 MB~280 MB~400 MB
Multi-provider AI NativeExtensionLimited
Arduino support NativeExtension
Raspberry Pi SSH NativeExtension
Android (28 ADB) Native
NVIDIA Jetson Native
Git + SVN Both nativeGit onlyGit only
Offline AI (Ollama)Extension
MISRA C / ISO 26262 AI-aware
Open source MIT
Engineering Brain

One understanding layer for your whole system.

Operator X02 builds a project understanding layer — so instead of searching files manually, you just ask natural questions and get answers grounded in the real system.

🧠
Engineering Brain
Project understanding layer
📄
Source Code
Every module, indexed
📚
Documentation
READMEs & specs
🏛️
Architecture
How it fits together
🛠️
Build System
Toolchains & targets
🔗
Dependencies
Cross-module links
🔌
Hardware Targets
Jetson, Pi, MCUs
🧭
Design Decisions
The why behind it
💬
Ask anything
Natural questions
Connected, not scattered — the AI reasons over the whole project, not a single open file.
What that understanding is actually for
🗺️
Architecture Intelligence
Understand systems visually. Generate diagrams straight from the code you already have.
  • Architecture overview
  • Module dependency
  • Data flow
  • Call graph
🧯
Build Intelligence
AI helps explain build failures in plain language — and points at the real cause.
  • Missing dependency
  • Compiler error
  • Toolchain issue
  • Environment mismatch
🔎
Large Project Intelligence
Built for large C/C++ and mixed codebases with thousands of files and modules.
  • Semantic search
  • Cross-module understanding
  • Repository navigation
  • Project-wide analysis
Engineering Graph

See how your project actually connects.

Source, hardware targets, build system and documentation aren't seven folders — they're one graph. Tap any node to trace what it touches.

Project the whole repository Architecture how it fits together Source every module, indexed Hardware jetson · pi · mcu Documentation readmes & specs Build System toolchains & targets 🧠 Engineering Brain one understanding layer
  1. Projectthe whole repository
  2. Architecturehow it fits together
  3. Sourceevery module, indexed
  4. HardwareJetson · Pi · MCU
  5. DocumentationREADMEs & specs
  6. Build Systemtoolchains & targets
  7. 🧠 Engineering Brainone understanding layer
Tap a node to trace its connections
Most AI tools see the file you have open. Operator X02 walks the graph — so an answer about a build failure can reach the hardware target and the design decision behind it.
Engineering Memory

Your project shouldn't restart from zero every day.

Operator X02 remembers previous decisions, module relationships, coding conventions, project structure and engineering knowledge — so the longer you work, the sharper it gets. 100% local, no token limits, no cloud.

OPERATOR X02 · MEMORY RECALL ACTIVE
What did we fix yesterday in auth.ts?
🔍 Trigger detected: "yesterday" + "fix"
🧠 History search: 3 matches · 15 ms
🔄 SVN status: 2 modified files
⚡ Context injected: 2,335 chars (invisible to you)
🧠 MEMORY🔄 SVN
Yesterday we fixed the JWT token expiry bug in auth.ts — the token wasn't auto-refreshing before expiry, causing logout loops. I can see auth.ts and main.ts modified in SVN; the fix is in place and ready to commit.
fix(auth): resolve JWT refresh loop on token expiry
172+
Trigger patterns
Phrases like "yesterday", "what changed", or "why" auto-surface the right past context.
History depth
No context window. No forgotten conversations. Searchable back to day one.
~15 ms
Local search
Everything indexed on your machine — instant recall, zero cloud round-trips.
How a single question becomes a context-aware answer
01
💬
User Message
"What did we fix yesterday?"
02
🔍
172+ Triggers
Auto-detect past references
03
🧠
History Search
~15 ms local lookup
04
🔄
VCS Context
SVN / Git status + diffs
05
Context Inject
Invisible to the user
06
AI Response
Context-aware answer
You see a clean conversation · the AI receives full history + VCS context · 100% local, no cloud
Build Mode

One prompt. One installer.

Describe the tool you need. Operator X02 plans it, writes it, checks it, compiles it with cargo and Tauri — and hands back something you can double-click. Locally. Without leaving the editor.

OPERATOR X02 · BUILD MODE LOCAL
>Build me a serial monitor for ESP32 with live plotting and CSV export
01
Prompt
Plain language, in the chat you already use.
02
Plan
Purpose, MVP scope, storage, risks — shown to you first.
03
Generate
Project scaffold and source files written to disk.
04
Validate
Static checks run before cargo is ever called.
05
Compile
cargo and Tauri, on your machine, offline.
06
Installer
A double-clickable Windows installer.
serial-monitor_setup.exe
Built on your machine · no cloud build service · no upload of your source
🪟 Windows ready 🐧 Linux in progress 🔒 Offline capable
Example output. Build Mode ships in v1.5.9.1a and later — read the release notes →
What you get

Outcomes, not just features.

AI-first from day one — not bolted on. Each capability is here for what it lets you do, from never losing working code to deploying straight onto real hardware.

🧠
Never lose working code
8-stage Surgical Edit Engine. A Rust pipeline that backs up before it touches a file, blocks destructive deletions, and patches only the lines that change — accountable, reversible, in ~36 ms.
CORE
🔄
Your AI, zero lock-in
Multi-provider routing. Claude, GPT-4o, Gemini, Groq, DeepSeek, Ollama — pick which model owns which task, or auto-route per request. Switch anytime.
AI
🚀
It reads before it writes
Auto Mode. Surgical line-level patches instead of full-file rewrites. Atomic multi-file batches, backup per command, and one-click rollback. Zero destroyed code.
AUTO
Watch your device, live
Jetson live dashboard. Stream tegrastats, watch GPU and thermal graphs, detect CUDA, and one-click SSH-deploy to Jetson — all without leaving the editor.
V1.5
💾
Never explain your project twice
Engineering Memory. 172+ triggers surface the right past conversation in ~15 ms. The AI recalls your architecture, decisions, and trade-offs — and gets sharper every session.
MEMORY
📹
Turn what you see into code
Camera AI vision. Point at a PCB, schematic, or whiteboard. The AI reads what you see and turns it into working code — no manual transcription.
VISION
🔌
Extend it to your stack
Plugin architecture. Add Python, C#, Kotlin, Arduino, or Flet plugins — new languages and toolchains without rebuilding the app.
EXTEND
🛡️
Ship safety-critical with confidence
Functional-safety aware. Tuned for MISRA C:2012, ISO 26262, IEC 61508, and IEC 62304 — purpose-built for ADAS and safety-critical embedded work.
SAFETY
🌿
Just ask what changed
AI-aware SVN & Git. Native TortoiseSVN and Git — real modified files, AI-written commit messages, diff analysis, and merge help on request.
VCS
Hardware Native

Designed for real devices.

First-class support for the boards and targets engineers actually ship on — SSH, flash, serial monitor, remote development and deployment, built into the core, not bolted on as extensions.

🔌
Arduino
One-click compile and flash, built-in serial monitor, pin visualizer, and an AI sketch generator. Boards auto-detected.
✓ Fully supported
🍓
Raspberry Pi
Deploy over SSH, run remote commands, and browse the Pi's filesystem as if it were local — straight from the editor.
✓ Fully supported
📱
Android
28 ADB commands, Gradle builds, AI Logcat crash analysis, an Arduino IoT bridge, and an ADB auto-installer — in 95 MB.
✓ Fully supported
NVIDIA Jetson
Live GPU dashboard, tegrastats streaming, CUDA detection, and remote deploy via the built-in SSH manager.
★ Installer edition
🐧
Linux Embedded
Develop, deploy and debug on embedded Linux targets over SSH — remote filesystem, remote commands, live deployment.
✓ Fully supported
Schematic + datasheet context

Code that matches the board — not code that guesses it.

Drop a board schematic, an MCU datasheet or a pin-map spreadsheet straight into the chat. Operator X02 reads the actual design — components, nets, pinouts — and writes firmware against the real pins instead of generic placeholders.

Live since v1.5.8 — see it read a real board
📐 schematic.pdf 📄 datasheet.pdf 📊 pin-map.xlsx
Components, nets and pinouts read
#defines and init written against real pins
! Mismatches flagged when code and board drift apart
🔐 SSH ⚡ Flash 📟 Serial Monitor 🌐 Remote Development 🚀 Deployment
Embedded AI · Safety-Critical Development
Write AI code. Flash it.
Test it on real hardware.
Not simulated. Not emulated. Deploy models to a Jetson Orin, flash MISRA-C-checked firmware to an Arduino, SSH into a Pi — then watch real sensor data stream back, live, in the same window you wrote the code.
🟢 Jetson Edge AI · NEW V1.5 ▶ Live Hardware Demos 🛡️ MISRA C · ISO 26262 📡 Real Sensor Data
AI providers

Your AI, your choice.

X02 ships with a free built-in AI — start coding with no API key, no account. Add your own providers when you want more power, or run Ollama for 100% offline, air-gapped work.

Claude (Anthropic)
OpenAI GPT-4o
Groq Llama
Deepseek
Gemini (Google)
Ollama (local) FREE · OFFLINE
Your IDE. Your rule.
Every developer has their own style, their own taste, their own way of creating. X02 doesn't impose a workflow — it adapts to yours. Bring your own provider, your own keys, your own models. Or use nothing at all. This is your canvas. Code it your way.
FREE
One IDE. Every Platform.

6 platforms.
3 live demos.
Zero switching.

From edge AI on Jetson to Android apps, Arduino firmware, safety-critical MISRA C, web fullstack, and ML pipelines — six target platforms, one 95 MB IDE, zero context-switching. See the full breakdown with side-by-side comparisons, live demo videos, and a feature deep-dive.

Explore All Platforms
Jetson · Android · Arduino demos  ·  Full comparison table  ·  FAQ
✍️
From the Blog

Stories, Insights & Dev Logs.

Behind the scenes of building an AI IDE — our philosophy, updates, and lessons learned.

🔄 Auto Mode ⚔️ The Craftsman's Path 🛡️ Surgical Edit Engine ⚡ Embedded AI
Read the Dev Logs
Built in the open

Judge it before you install it.

No waitlist, no private beta, no “contact sales”. The source, every release and every open issue are public — you can read the code and the history before you install anything.

Get started

Two ways to get X02.

Grab the installer for the full experience in one click, or clone the source to study, fork, and contribute.

For developers
Build from Source
Clone, study, and contribute. Requires Node.js 18+ and Rust 1.70+.
View Source on GitHub
# prerequisites: Node 18+, Rust 1.70+
git clone github.com/csheng21/operator-x02-ide
npm install
npm run tauri dev
Note: 4 core modules are community stubs when built from source. Download the installer for the full experience.
All IDE features (Git, SVN, Arduino…)FULL
Surgical Edit EngineSTUB
NVIDIA JetsonSTUB
SSH Remote ManagerSTUB
— feel it. enjoy it. —

Coding is Art.
Start creating yours
today.

Free forever. MIT license. No account, no telemetry, no cloud lock-in — just your code, your machine, and an AI that actually speaks embedded.

MIT License  ·  Windows Beta  ·  No account  ·  Offline capable  ·  Free built-in AI