Skip to main content

Cursor Notebook Bridge

Edit Jupyter notebooks via live kernel tools.

Skill metadata​

SourceBundled (installed by default)
Pathskills/software-development/cursor-notebook-bridge
Version1.0.0
AuthorVibeOS
LicenseMIT
Platformslinux, macos, windows
Tagscursor, notebook, jupyter, ipynb, data-science
Related skillsjupyter-live-kernel, cursor-capabilities-bridge

Reference: full SKILL.md​

info

The following is the complete skill definition that VibeOS loads when this skill is triggered. This is what the agent sees as instructions when the skill is active.

Cursor Notebook Bridge

Cursor’s EditNotebook tool edits .ipynb cells in the IDE. VibeOS has no separate notebook core tool — use skill jupyter-live-kernel (execute/edit via JupyterLab API) or edit the JSON with read_file / patch when offline.

When to Use​

  • User asks to edit/run notebook cells “как EditNotebook в Cursor”
  • Data-science loops that need a live kernel
  • Small .ipynb text fixes without a running server

Prerequisites​

  • For live kernel: JupyterLab installed (uv tool install jupyterlab)
  • Skill jupyter-live-kernel scripts available (bundled under data-science)
  • Offline fallback: file tools only

How to Run​

Live kernel (preferred)​

Follow jupyter-live-kernel: start JupyterLab, then:

# Discover / execute / edit (paths from that skill’s scripts/)
uv run <jupyter_live_kernel.py> notebooks --compact
uv run <jupyter_live_kernel.py> execute --path notebook.ipynb --code 'print(1)' --compact
uv run <jupyter_live_kernel.py> edit --path notebook.ipynb replace-source \
  --cell-id <id> --source 'x = 1' --compact

Offline JSON edit​

  1. read_file the .ipynb.
  2. Change cells[i].source carefully (keep valid JSON).
  3. Prefer patch over rewriting the whole file.
  4. Do not invent outputs — clear or leave existing outputs unless asked.

Quick Reference​

CursorVibeOS
EditNotebook insert/editjupyter-live-kernel edit
Run celljupyter-live-kernel execute
No serverread_file + patch on .ipynb
New core notebook toolDo not add — skill path only

Pitfalls​

  • Breaking notebook JSON (trailing commas, wrong cell type)
  • Running untrusted notebook code without user intent
  • Adding a core notebook_* tool (footprint) when the skill already covers it

Verification​

  • Cell source updated; notebook still opens in Jupyter / VS Code
  • Execute path returns stdout for a trivial print