Deepnote

Delete a notebook from a Deepnote project. A project version is saved first, so the notebook can be restored from version history.

delete_notebook

Delete a Deepnote project with all its notebooks. The project moves to the workspace trash and can be restored from there for 30 days. Its schedules are deleted and are not restored, and its machine is stopped.

delete_project

Detach a Deepnote workspace integration from a project. Use list_integrations to discover integration IDs.

detach_integration

Duplicate an existing Deepnote notebook inside its current project.

duplicate_notebook

Generate an absolute Deepnote URL for a project or notebook.

generate_project_url

Fetch the full Markdown body of a Deepnote documentation article. Use `list_docs` first to find the slug.

get_doc

Check whether a Streamlit app is running, starting, or unavailable.

get_streamlit_app_status

Get Deepnote block details, including content and metadata. Optionally include the outputs the block shows in the live editor, capped unless fullOutputs is true.

get_block

Return info about the API key calling the MCP server: key name and type, the user that created it, the workspace, and the access level (viewer, editor, admin, etc.).

get_me

Get Deepnote integration details and cached table structure. For large integrations, search tables by relevance with search. Use list_integrations to discover integration IDs.

get_integration

Check whether a Deepnote project's machine is stopped, starting (and what it is waiting for), running, stopping, or in an error state, its machine type (vCPUs, memory and GPUs), and the sandbox environment it runs in. This is the machine the editor, live-mode runs and hosted Streamlit apps use; detached runs and sessions run on machines of their own, which this tool does not report. Cheap enough to poll after starting or stopping the machine.

get_machine

Get Deepnote notebook details, including blocks and input variables.

get_notebook

Get a Deepnote notebook run. By default returns a presigned snapshot download URL when available; optionally request inline snapshot content or the executed blocks with their outputs instead. Block outputs are capped unless fullOutputs is true.

get_run

Get Deepnote project metadata, notebooks, attached integrations, the recursive file inventory, static website settings, and the environment (Docker image) its machine runs.

get_project

Return the Deepnote docs navigation tree (sections containing articles, in display order). Use this to discover which article to fetch with `get_doc` when answering product questions.

list_docs

List the Streamlit apps a Deepnote project serves, with the file each one runs and the URL it is served on.

list_streamlit_apps

List the environments (Docker images) projects in the Deepnote workspace can run on: Deepnote's images and the workspace's custom images, with their image reference, language, and whether they are the default, GPU, or deprecated. Custom images from a private registry also need the registry integration attached to the project.

list_environments

List Deepnote workspace folders with their parent relationships in one response, optionally filtered by name. Omit the filter to receive the complete tree. Use this to resolve a folder name or nested folder path to the `folderId` accepted by `create_project`.

list_folders

List SQL blocks that use a Deepnote integration, optionally narrowed to a single project. Use list_integrations to discover integration IDs.

list_integration_block_usages

List notebooks that contain SQL blocks using a Deepnote integration, optionally narrowed to a single project. Use list_integrations to discover integration IDs.

list_integration_notebook_usages

List projects connected to a Deepnote integration, optionally narrowed to a single project. Use list_integrations to discover integration IDs.

list_integration_project_usages

List Deepnote workspace integrations, optionally filtered by name or type.

list_integrations

List the machine types (sizes) projects in the Deepnote workspace can run on: Deepnote's machine types and the workspace's custom machines, with their vCPUs, memory and GPUs, and whether the workspace's plan and this month's free quota allow each one. Use it to pick a machine big enough for the workload. Also lists the sandbox environments and which one new projects use. A custom machine only works for projects in its sandbox environment.

list_machine_types

List historical runs for a Deepnote notebook.

list_notebook_runs

List Deepnote workspace projects, optionally filtered by name.

list_projects

Publish a small HTML, CSS, and JavaScript site to the project's top-level `_deepnote_static/` directory and share it as an app. File paths are relative to that directory. Never publish API keys, personal access tokens, `.env` files, or other secrets. Use the Deepnote CLI for larger sites or when terminal access is available. Use `update_project` to change access without deleting files; disabling sharing also disables viewer API access.

publish_static_site

Read a small text file, or a range of its lines, from a Deepnote project, such as a script, config, or CSV to inspect or edit. Not meant for large or binary files.

read_file

Reorder blocks in a Deepnote notebook by moving one or more blocks to a new position.

reorder_notebook_blocks

Stop a Deepnote project's machine and start a new one, which recovers a machine in an error state. Kernel state, such as variables, is lost for everyone working in the project. Returns without waiting for the new machine to boot: poll `get_machine` until it is running. This is the machine the editor, live-mode runs and hosted Streamlit apps use; machines of detached runs and sessions are not affected.

restart_machine

Search Deepnote workspace resources across projects, notebooks, blocks, and integrations.

search

Start a Deepnote project's machine, which uses machine time until it stops. Returns without waiting for the machine to boot: poll `get_machine` until it is running. Starting a machine that is already starting or running does nothing. This is the machine the editor, live-mode runs and hosted Streamlit apps use; detached runs (the `create_run` default) start machines of their own, so there is no need to start this one before them.

start_machine

Stop a Deepnote project's machine so it no longer uses machine time. Kernel state, such as variables, is lost for everyone working in the project; project files are kept. Stopping a machine that is not running does nothing. This is the machine the editor, live-mode runs and hosted Streamlit apps use; machines of detached runs and sessions keep running.

stop_machine

Update an existing block in a Deepnote notebook.

update_block

Change a Deepnote project's machine type or sandbox environment (ids from list_machine_types), or how long its machine keeps running without activity. This is the machine the editor, live-mode runs and hosted Streamlit apps use; to run one detached run on another machine type, pass machineType to create_run instead. Changing the type or sandbox environment of a running machine requires restart: true, which restarts it and loses kernel state, such as variables, for everyone working in the project.

update_machine

Rename an existing Deepnote notebook. Naming it "Init" designates the project init notebook, which can run as a prelude to other notebooks in the project.

update_notebook

Update static website sharing and viewer API access settings for a Deepnote project, whether its hosted Streamlit apps may call the public API as their viewers, the environment (Docker image) its machine runs, and the folder it is in. Static website API access requires sharing, and disabling sharing also disables API access. Find environment IDs with `list_environments`; changing the environment restarts a running machine, which clears kernel state. Find folder IDs with `list_folders`, or create a folder with `create_folder`.

update_project

Write UTF-8 text or base64 content to a file in a Deepnote project. The path is relative to the project root. Set `overwrite` to replace an existing file; otherwise a unique path is generated and returned.

write_file