jxtngx/fork-setup
.cursor/skills/fork-setup/SKILL.md
Get DGX Lab running on your own DGX Spark after cloning or forking. Covers prerequisites, install, config, first run, Docker, Tailscale, and common troubleshooting. Use when setting up DGX Lab for the first time or diagnosing setup issues.
devops
Updated Apr 16, 2026
$ install --global
skillsauthnpx skillsauth add jxtngx/dgx-lab fork-setupInstall this skill globally with one command. Works with Claude Code, Cursor, and Windsurf.
Security Scan Results
3 of 9 scanners reported clean
Some scanners were skipped, did not run, or reported a non-clean status. Review each row below.
3
Scanners Passed
9
Scanners in report
Clean
TrivyContainer and dependency vulnerability scanner
95%
Clean
SemgrepStatic code analysis for vulnerabilities
95%
Clean
mcp-scan (Snyk)Model Context Protocol security validation
95%
Skipped
Snyk (dep)Open source security scanning
50%
Skipped
Socket.devSupply chain security analysis
50%
Skipped
VirusTotalMulti-engine malware detection
50%
Skipped
CrowdStrikeAdvanced threat intelligence
50%
Skipped
OSV-ScannerOpen Source Vulnerability database check
50%
Skipped
OWASP Dep-Check
50%
Last scanned: Apr 16, 2026, 5:54 PM100.8s1 file scanned
SKILL.md
- name:
- fork-setup
- description:
- >-
Fork Setup — Getting DGX Lab Running
Prerequisites
| Requirement | Check |
|-------------|-------|
| NVIDIA DGX Spark (or GPU with nvidia-smi) | nvidia-smi returns GPU info |
| Python 3.12+ | python3 --version |
| uv | uv --version |
| Bun 1.3+ | bun --version |
| Docker + Docker Compose (production only) | docker compose version |
Install and first run
git clone <your-fork-url> dgx-lab && cd dgx-lab
cd backend && uv sync && cd ..
cd frontend && bun install && cd ..
make dev
Open http://localhost:3000. Backend is at http://localhost:8000/api/health.
Configuration
All paths are set via env vars in backend/app/config.py. Defaults work out of the box on a Spark:
| Env var | Default | What |
|---------|---------|------|
| DGX_LAB_MODELS_DIR | ~/.cache/huggingface/hub | HuggingFace model cache |
| DGX_LAB_EXPERIMENTS_DIR | ~/.dgx-lab/experiments | Logger experiment data |
| DGX_LAB_TRACES_DIR | ~/.dgx-lab/traces | Agent trace JSONL files |
| DGX_LAB_DATASETS_DIR | ~/.dgx-lab/datasets | Local dataset storage |
| DGX_LAB_MEMORY_TOTAL_GB | 128 | Total memory for fit calculations |
| DGX_LAB_MEMORY_BW_MAX_GBS | 273 | Memory bandwidth ceiling |
If your machine has different specs (e.g. a non-Spark GPU), override the memory vars:
export DGX_LAB_MEMORY_TOTAL_GB=80
export DGX_LAB_MEMORY_BW_MAX_GBS=200
Docker (production)
make build
make up
Serves on port 80 via nginx. Stop with make down. Logs with make logs.
Remote access (Tailscale)
# On the Spark
curl -fsSL https://tailscale.com/install.sh | sh
sudo tailscale up
tailscale serve http://localhost:80
# From your Mac
open https://spark.your-tailnet.ts.net
See docs/remote-access.md for SSH tunnels and systemd auto-start.
Troubleshooting
| Problem | Fix |
|---------|-----|
| uv sync fails | Ensure Python 3.12+. Run uv python install 3.12 if needed. |
| bun install fails | Ensure Bun 1.3+. Run curl -fsSL https://bun.sh/install \| bash to update. |
| Monitor shows "GPU not available" | Verify nvidia-smi works on the host. In Docker, add deploy.resources.reservations.devices for GPU access. |
| Frontend can't reach backend | Dev mode proxies /api/* via Next.js rewrite to localhost:8000. Ensure backend is running. |
| Docker build fails on frontend | Ensure bun.lock exists. Run bun install locally first to generate it. |
| Port 80 already in use | Change the nginx port in docker-compose.yaml: ports: ["8080:80"] |
Project layout (quick reference)
backend/app/main.py ← FastAPI app, router registration
backend/app/config.py ← All env vars and hardware constants
backend/app/routers/*.py ← One router per tool
frontend/apps/web/app/ ← Next.js pages (app router)
frontend/apps/web/components/ ← UI components per tool
frontend/packages/ui/ ← Shared shadcn components
docker-compose.yaml ← frontend + backend + nginx
nginx.conf ← Reverse proxy config
Makefile ← dev, build, up, down, logs, rebuild
Related Skills
jxtngx/langsmith-trace
tools
VerifiedTrustedCommunity
INVOKE THIS SKILL when working with LangSmith tracing OR querying traces. Covers adding tracing to applications and querying/exporting trace data. Uses the langsmith CLI tool.
SKILL.mdUpdated Apr 16, 2026
jxtngx/langsmith-evaluator
tools
VerifiedTrustedCommunity
INVOKE THIS SKILL when building evaluation pipelines for LangSmith. Covers three core components: (1) Creating Evaluators - LLM-as-Judge, custom code; (2) Defining Run Functions - how to capture outputs and trajectories from your agent; (3) Running Evaluations - locally with evaluate() or auto-run via LangSmith. Uses the langsmith CLI tool.
SKILL.mdUpdated Apr 16, 2026
jxtngx/langsmith-dataset
tools
VerifiedTrustedCommunity
INVOKE THIS SKILL when creating evaluation datasets, uploading datasets to LangSmith, or managing existing datasets. Covers dataset types (final_response, single_step, trajectory, RAG), CLI management commands, SDK-based creation, and example management. Uses the langsmith CLI tool.
SKILL.mdUpdated Apr 16, 2026
jxtngx/langgraph-persistence
testing
VerifiedTrustedCommunity
INVOKE THIS SKILL when your LangGraph needs to persist state, remember conversations, travel through history, or configure subgraph checkpointer scoping. Covers checkpointers, thread_id, time travel, Store, and subgraph persistence modes.
SKILL.mdUpdated Apr 16, 2026