antigravity-nb
A Jupyter notebook MCP server for Antigravity — lets AI agents read, edit, and execute .ipynb notebooks directly inside your workspace.
Potencialize seu agente
Instalação do publicador. Não é um endpoint do PlugMCP.
Host a copy with our keys and logs.
Ainda não há uma versão empacotada no registro. Siga o README do repositório.
URL remota ou um comando na sua máquina? Como as duas instalações diferem
Rodar no PlugMCP
A new server you create. It is not this listing's official endpoint. Use the publisher install when you want their URL and their credentials.
AbrirDo publicador
A Jupyter notebook MCP server for Antigravity — lets AI agents read, edit, and execute .ipynb notebooks directly inside your workspace.
A Jupyter notebook MCP server for [Antigravity](https://antigravity.codes) — lets AI agents read, edit, and execute `.ipynb` notebooks directly inside your workspace.
.ipynb files are allowedpip install git+https://github.com/Ian747-tw/Antigravity-Jupiter-Notebook-MCP.git
git clone https://github.com/Ian747-tw/Antigravity-Jupiter-Notebook-MCP.git
cd antigravity-nb
pip install -e .
Open your Antigravity configuration file. It is usually found at:
~/.gemini/antigravity/mcp_config.json
Or go to Settings → MCP Servers inside Antigravity.
Append the following to your mcpServers object:
{
"mcpServers": {
"antigravity-nb": {
"command": "antigravity-nb",
"args": ["serve-agent", "--workspace-root", "${workspaceFolder}"]
}
}
}
Restart Antigravity or trigger Refresh MCP Servers to discover the new tools.
Ask the agent:
"List the tools available from antigravity-nb" "Open a notebook and run the first cell"
Once connected, Antigravity discovers all 10 tools automatically. You can then ask it things like:
"Run the preprocess stage in pipeline.ipynb and show me the outputs" "Edit cell 2 in analysis.ipynb to use pandas instead of a plain list" "Run all cells in notebook.ipynb and tell me if any failed"
If you see ModuleNotFoundError when running a cell, the notebook is executing in a kernel that doesn't have your packages installed. Check available kernels and install one that matches your environment:
# List installed kernels
jupyter kernelspec list
# Install the current environment as a kernel
python -m ipykernel install --user --name myenv
Then pass kernel_name when calling run_cell or run_range, or restart Antigravity and select the correct kernel.
To use the run_pipeline tool, tag your cells in the notebook metadata. In JupyterLab: View → Cell Toolbar → Tags, then add a tag like preprocess, train, or eval to each cell.
The agent can then run specific stages in order while maintaining kernel state across them:
"Run the train and eval stages in model.ipynb"
Run the server manually and pipe stderr to a file to see all tool calls and errors:
antigravity-nb serve-agent --workspace-root . 2>debug.log
tail -f debug.log
edit_cell and delete_cell write a timestamped backup before making changes:
notebook.checkpoint_20250610_143022.ipynb
Pass checkpoint=false to skip if you don't need it.
The CLI is also available for scripting:
# Run a single cell
antigravity-nb run-cell notebook.ipynb 3
# Run a range of cells
antigravity-nb run-range notebook.ipynb 2 7
# Run a tagged pipeline
antigravity-nb run-pipeline notebook.ipynb
antigravity-nb run-pipeline notebook.ipynb --stage preprocess --stage train
# Edit a cell from a file
antigravity-nb edit-cell notebook.ipynb 5 --source-file ./new_cell.py
# List pipeline stages
antigravity-nb list-stages notebook.ipynb
# Start the MCP server manually
antigravity-nb serve-agent --workspace-root .
from antigravity_nb import NotebookAdapter, KernelSession, NotebookRunner
nb = NotebookAdapter("pipeline.ipynb")
ks = KernelSession(nb)
runner = NotebookRunner(nb, ks)
ks.start()
nb.update_cell(0, "print('hello')")
result = runner.run_cell(0)
print(result.outputs)
nb.save()
ks.shutdown()
.ipynb files are allowed--workspace-root; any path outside it is rejectedjupyter_client >= 8.0.0pip install ipykernel for Python)MIT
A Jupyter notebook MCP server for [Antigravity](https://antigravity.codes) — lets AI agents read, edit, and execute `.ipynb` notebooks directly inside your workspace.
Ferramentas
Ainda não há lista de ferramentas. Servidores remotos recebem uma após a primeira verificação de saúde.
Perguntas frequentes
Copie nesta página a configuração stdio para Cursor, Claude Desktop, Claude Code ou VS Code. Ela usa a instalação do publicador, não um proxy do PlugMCP.
Este registro não lista nenhuma variável de ambiente obrigatória.
Antigravity Jupiter Notebook roda localmente via stdio.
Última verificação: 2026-10-27. A Jupyter notebook MCP server for [Antigravity](https://antigravity.codes) — lets AI agents read, edit, and execute `.ipynb` notebooks directly inside your workspace.