Kiwano

Cross-platform desktop app

Your folder, your tables, and an assistant—in one focused workbench.

Kiwano is an Electron + React desktop app that pairs a project file tree with fast tabular and document previews. A Python engine handles CSV, TSV, Excel, and more over a simple NDJSON protocol, while chat stays grounded in the files you open—with Gemini, Kiwano Cloud, or a local model.

Build & run locallySee the layout

The workbench

Three panes, one flow—mirror of the desktop UI you already use.

Open a workspace folder, pin files into assistant context, and preview data on the right. The layout deliberately echoes the shipping app: files on the left, chat in the center, structured viewers on the right.

Files
Project tree · search · context checkboxes
  • reports/
  • sales_q1.csv
  • notes.docx
  • figures/
Assistant
Summarize dtypes and outliers in the open sheet.
On it—using the CSV preview and workspace paths you shared.
Data
sales_q1.csvnotes.docx
region
sku
units
revenue
EMEA
A-12
42
12840

Features

Built for analysts, builders, and anyone living next to messy folders.

  • Rich previews

    CSV and TSV via pandas, Excel workbooks, Word documents, images, and text—streamed through a dedicated Python child process.

  • Context-aware chat

    Pin files from the tree, thread active tabs, and let the model reason over paths and previews—tools can mutate the workspace when you approve plans.

  • Flexible AI backends

    Use Google Gemini with a developer key, sign into Kiwano Cloud, or run a local GGUF model through llama.cpp when you want everything on-machine.

  • Engine you can repair

    The Kiwano engine installs into an isolated venv when needed; Settings exposes logs, package installs, and health so previews stay reliable.

  • Sessions that stick

    Open tabs and the active file persist under .kiwano in your project, so returning to a folder feels like resuming—not starting over.

  • Web companion (this site)

    This Next.js service complements the desktop app with account and LLM API surfaces—the product story starts here; the heavy lifting stays in the app.

Get started

Run the desktop app from the Kiwano repository.

Clone kiwano_app, install Node.js dependencies, set up the Python engine from python/requirements.txt, then npm run dev. Watch for the Engine status pill—when it reads ok, previews and tools are ready.

Quick commands

cd kiwano_app
npm install
pip install -r python/requirements.txt
npm run dev