beihzb/dsh-notebook

A stateful, agent-controllable Jupyter workspace: a persistent ipykernel the DSH agent can read, edit, execute and inspect, giving a real Agent ↔ Kernel ↔ Artifact loop with per-cell version snapshots and revert.

dsh-notebook is a stateful, agent-controllable Jupyter workspace for DeepSeek Harness: a persistent kernel that the DSH agent can read, edit, execute and inspect, turning a notebook from a static artifact into an operationable target. Agent tools such as nb_get, nb_edit_cell, nb_run_cell, nb_context, nb_list_vars, nb_inspect_object, nb_cell_history, nb_revert_cell, nb_error_context and nb_edit_and_run_cell give a real Agent ↔ Kernel ↔ Artifact loop with structured access to outputs and tracebacks, so the agent edits a cell, reruns it, reads stdout or the traceback and iterates while operating on live state (e.g. an Anndata or a loaded GPU model) rather than assembling strings. The UI adds live tqdm progress bars, clickable traceback frames, VS Code-aligned cell states and execution semantics, per-cell version snapshots, and a React error boundary that keeps the notebook view alive when an output fails to render.

Tools & Capabilities ★ 3
View on GitHub ↗

Install

dsh plugin add @beihaizb/dsh-notebook

Copied verbatim from the README (the unscoped patch name previously failed to resolve, so the scoped package @beihaizb/dsh-notebook loads out of the box). The README also shows a Web-profile form and a version-admission note.

Compatibility

A DSH plugin pairing a persistent ipykernel + jupyter_client sidecar with a VS Code-aligned notebook UI. Variables persist across cells and notebooks save and reload; the agent drives a live computational session rather than running one-shot scripts.

Details

Alternatives

havingautism/dsh-notebooks · fff122/dsh-research-notes · zhaoolee/notes

More plugins in Tools & Capabilities

Browse more in Tools & Capabilities

Guides for Tools & Capabilities plugins