// Full CUDA ML JupyterLab image with real-time collaboration and CRDT MCP server. Base: nvidia. Port 8888. GPU-accelerated ML training + collaborative notebooks. MUST be invoked before building, deploying, or troubleshooting the jupyter-ml image.
[HINT] Download the complete skill directory including SKILL.md and all related files
| name | jupyter-ml |
| description | Full CUDA ML JupyterLab image with real-time collaboration and CRDT MCP server. Base: nvidia. Port 8888. GPU-accelerated ML training + collaborative notebooks. MUST be invoked before building, deploying, or troubleshooting the jupyter-ml image. |
jupyter-ml:
base: nvidia
layers:
- agent-forwarding
- jupyter-ml
- notebook-templates
- dbus
- ov
ports:
- "8888:8888"
platforms:
- linux/amd64
The jupyter-ml layer is a Tier 2 environment-owning meta-layer that composes:
llama-cpp ā llama.cpp prebuilt binaries + GGUF toolsunsloth ā vLLM 0.19 cu130 inference engine + fine-tuning (unsloth + unsloth-zoo) + vLLM torch.compile patchjupyter-mcp ā CRDT MCP extension (fastmcp + jupyter_mcp package)Additional layers from the image:
agent-forwarding ā SSH/GPG agent forwardingnotebook-templates ā Starter notebooks (data layer, seeds ~/workspace)dbus ā D-Bus session busov ā Overthink CLIquay.io/fedora/fedora:43
āāā fedora
āāā fedora-nonfree (RPM Fusion repos)
āāā nvidia (NVIDIA drivers + container toolkit)
āāā jupyter-ml
| Port | Service |
|---|---|
| 8888 | JupyterLab + MCP endpoint at /mcp |
| Name | Path | Purpose |
|---|---|---|
| workspace | ~/workspace | Persistent notebook storage |
| models | ~/.cache/huggingface | HuggingFace model cache (from unsloth sub-layer) |
This image receives env_provides variables from infrastructure layers when they are deployed:
| Variable | Injected by | Value |
|---|---|---|
OLLAMA_HOST | /ov-ollama:ollama | http://ov-ollama:11434 |
PGHOST | /ov-foundation:postgresql | ov-postgresql |
PGPORT | /ov-foundation:postgresql | 5432 |
REDIS_URL | /ov-foundation:redis or /ov-foundation:valkey | redis://ov-<image>:6379 |
These variables are injected automatically into the container environment at ov config time when the corresponding service is deployed. No manual -e flags needed.
/mcp ā 11 tools for programmatic notebook access (server manages CRDT rooms invisibly; see /ov-jupyter:jupyter-mcp for the auto-attach + canonicalization design)ov image build jupyter-ml
ov config jupyter-ml
ov start jupyter-ml
ov status jupyter-ml
ov logs jupyter-ml -f
# JupyterLab: http://localhost:8888
# MCP endpoint: http://localhost:8888/mcp
ov shell jupyter-ml -c "pixi run verify-pytorch"
ov shell jupyter-ml -c "pixi run verify-vllm"
ov shell jupyter-ml -c "pixi run verify-unsloth"
ov shell jupyter-ml -c "pixi run verify-mcp"
ov shell jupyter-ml -c "pixi run verify-collaboration"
| jupyter | jupyter-ml | |
|---|---|---|
| Base | fedora | nvidia |
| CUDA | No | Yes |
| Arch | amd64 + arm64 | amd64 |
| MCP | CRDT (11 tools) | CRDT (11 tools) |
| Collaboration | Yes | Yes |
| ML Stack | No | Full |
| Volume | workspace | workspace + models |
| Notebook dir | ~/workspace | ~/workspace |
/ov-jupyter:jupyter-ml-notebook ā Same stack with fine-tuning notebooks/ov-jupyter:jupyter ā Lightweight variant (no CUDA, multi-arch)/ov-foundation:python-ml ā ML base without JupyterMCP testing: inherits 3 deploy-scope mcp: checks from the jupyter-ml layer (ping, list-tools, call list_notebooks). Run ov eval live jupyter-ml --filter mcp or probe ad-hoc with ov eval mcp list-tools jupyter-ml. See /ov-build:mcp.
MUST be invoked before building, deploying, configuring, or troubleshooting the jupyter-ml image.
/ov-build:image ā image family umbrella (image: entries in overthink.yml, build/validate/inspect/list)/ov-build:build ā build.yml vocabulary (distros, builders, init-systems)