MachLab — AI for Robotics
AI-powered coding assistant for ROS 2, embedded C++, and drone development.
Phase 1 MVP — VS Code Extension
What it does
| Feature |
How to trigger |
| Ask AI about selected code |
Highlight → Ctrl+M (or Cmd+M on Mac) |
| Auto-detect build errors |
Just run colcon build — MachLab watches the terminal |
| Fix build error with AI |
Click the status bar item or run MachLab: Fix Last Build Error |
| Open AI chat panel |
Command Palette → MachLab: Open AI Chat |
| Show workspace context |
Command Palette → MachLab: Show Workspace Context |
Setup
1. Get a free NVIDIA NIM API key
- Go to build.nvidia.com
- Sign up / log in → click Get API Key
- Copy the key (starts with
nvapi-…)
2. Set your key in VS Code
Open Settings → search machlab.nimApiKey → paste your key.
Or via the command palette:
Preferences: Open User Settings (JSON)
Add:
"machlab.nimApiKey": "nvapi-your-key-here"
3. Open your ROS 2 workspace
Open the folder that contains your src/ and (optionally) colcon.meta.
MachLab will auto-detect your packages on startup and inject them as context into every AI query.
How it knows your project
On activation, MachLab scans for:
package.xml — package name, build type (ament_cmake / ament_python), dependencies
CMakeLists.txt — add_executable() targets (node names)
setup.py — Python node entry points
- Topic name string literals in
src/ files
This context is injected into the AI system prompt so answers are specific to your workspace.
Configuration
| Setting |
Default |
Description |
machlab.nimApiKey |
"" |
NVIDIA NIM API key |
machlab.nimModel |
meta/llama-3.3-70b-instruct |
Model to use |
machlab.errorSnifferEnabled |
true |
Watch terminals for build errors |
machlab.autoAnalyzeWorkspace |
true |
Analyse on startup |
Build from source
cd machlab
npm install
npm run compile # or: npm run watch
# Press F5 in VS Code to launch Extension Development Host
To package a .vsix:
npm run package
# → machlab-0.1.0.vsix
Roadmap
- Phase 2 — One-click Wi-Fi deploy to RPi/Jetson, Docker cross-compilation,
.bag/ulog telemetry reader
- Phase 3 — VS Code Marketplace launch, user metrics, community feedback
- Phase 4 — Native standalone IDE (VSCodium fork), live sensor graphs, 3D physics simulator