Marimo brings reactive Python notebooks natively into PyCharm

Marimo has released an early plugin that embeds its reactive notebooks directly inside PyCharm and other JetBrains IDEs, addressing one of its most-requested features. Notebooks live alongside project code and can import from the library or application under development; marimo’s module reloader propagates edits from PyCharm’s editor into notebook cells, creating a tight edit-and-observe feedback loop without leaving the IDE. Opening a notebook auto-starts a local marimo server, with the plugin handling port and process lifecycle so there’s no separate terminal or browser tab to juggle.

Because marimo notebooks are stored as plain .py files—identified by a marimo import and matching App declaration—they diff and version cleanly through PyCharm’s VCS tooling, and ordinary Python files still open normally. By default notebooks run on the project’s configured interpreter and dependencies, but a uv-backed sandbox mode isolates them entirely, storing dependencies inline via PEP 723 script metadata so experimental packages stay out of the project environment. Existing PyCharm tools like the database inspector work side by side with the notebook, though the two aren’t directly integrated and connections must be configured separately.

The plugin also supports agent pairing: a “Pair with AI” option launches a marimo-pair session with Claude, Codex, or opencode in a terminal, or copies a prompt for use with other agents including JetBrains’ Junie. Both marimo and the plugin are open source, and the team is soliciting feedback on this preview via the plugin’s GitHub repository.