repere

Data explorer for datasets too large for spreadsheets.

Run SQL queries on CSV, Parquet, and Excel files locally. No uploads, no servers-your data never leaves your machine.

Why repere?

• 100% Local - Data stays on your machine. Nothing is uploaded to any server.
• Handle Large Files - Process CSV, Parquet, XLSX, and JSON files with millions of rows.
• SQL Power - Full SQL support via DuckDB with autocomplete.
• Visual Pipeline - See your data transformations as a DAG on an interactive canvas.
• Shareable - Export your work as a self-contained HTML file anyone can open.
• Web or Desktop - Use in any modern browser, or download the portable desktop app.

Features

Category	Operations
Query	Filter, Sort, Limit
Columns	Select, Add, Remove, Rename, Reorder, Cast type
Edit	Edit cells, Fill nulls, Replace values
Aggregate	Group By, Pivot (with subtotals/grand totals), Unpivot
Window	ROW_NUMBER, RANK, LAG/LEAD, Running aggregates
Combine	Join (inner/left/right/full/cross), Union, Distinct
Visualize	15+ chart types (bar, line, pie, scatter, heatmap, treemap, box plot), Dashboards
Analyze	Column profiling, Statistics, Histograms, Outlier detection, Correlation matrix
AI	Natural language data exploration with intelligent agents
Python	pandas, numpy, matplotlib via Pyodide (web) or native (desktop)
SQL	Raw SQL queries with syntax highlighting and autocomplete
Export	CSV, Parquet, XLSX, JSON, JSONL, or shareable HTML

Supported Formats

• CSV - Comma-separated values with automatic type inference
• Parquet - Columnar format, great for large datasets
• XLSX - Microsoft Excel workbooks
• JSON / JSONL - JSON arrays and newline-delimited JSON

Quick Start

Use the App

Web: Visit repere.ai

Desktop: Download the portable binary for your platform from Releases:

• macOS (Apple Silicon / Intel)
• Windows
• Linux

Develop

# Clone the repository
git clone https://github.com/mattismegevand/repere.git
cd repere

# Install dependencies
bun install

# Start development server
bun run dev

For desktop app development (Tauri):

# Install Rust and Tauri CLI first
bun run tauri dev

How It Works

1. Load datasets - Drop CSV, Parquet, XLSX, or JSON files to create root nodes
2. Apply operations - Filter, sort, group, join, pivot to create derived views
3. Lazy evaluation - Each view is a DuckDB VIEW, computed on-demand
4. Visual canvas - See and navigate your full transformation pipeline
5. Share - Export as self-contained HTML or save as .repere session file

Keyboard Shortcuts

Key	Action
⌘K	Command palette
⌘O	Open file
⌘S	Save session
⌘Z / ⌘⇧Z	Undo / Redo
⌘/	Toggle canvas/table view
⌘`	SQL editor
⌘?	All shortcuts

Tech Stack

• React 19 + TypeScript + Vite + Bun
• DuckDB-WASM (web) / Native DuckDB (desktop via Tauri)
• TanStack Virtual - Virtualized data grid
• React Flow - Interactive pipeline canvas
• ECharts - Charts and sparklines
• Zustand - State management
• Tailwind CSS v4

Contributing

Contributions are welcome! See CLAUDE.md for development guidelines, code style, and architecture notes.

# Before submitting a PR
bun run test
bun run lint
bun run format
bun run build

Author

Mattis Megevand

License

MIT