duckdb

10 repositories

Canner/WrenAI

A Git-versioned semantic layer that lets AI agents write trusted SQL — WrenAI gives agents like Claude Code, Cursor and LangChain a governed semantic layer plus an AI context layer, encoding metric definitions, joins and business meaning as reviewable YAML/Markdown in your own repo. It connects to 22+ sources and turns natural-language questions into governed SQL and shareable dashboards, aimed at teams wanting AI-generated BI they can audit.

PythonAttention 47
★18,346
perspective-dev/perspective

A streaming query engine for data grids, pivots, and charts in any runtime — Perspective pairs a drag-and-drop grid, pivot table, and WebGL charting UI with a C++ query engine compiled to WebAssembly, Python, and Rust. Notable extras include a 64-bit memory64 build for in-browser datasets over 4GB, OPFS/mmap paging to disk, DuckDB support, and Jupyter and React integrations for embedded analytics.

RustAttention 70
★11,609
opengeos/GeoLibre

A privacy-first GIS app with 1,000+ browser-based WASM tools — GeoLibre is a free, cloud-native GIS platform built on Tauri, React, MapLibre GL JS and DuckDB-WASM. It runs in browsers, desktop, mobile and Jupyter, offers 1,000+ WebAssembly geoprocessing tools with no server needed, and lets users switch between MapLibre, Mapbox, CesiumJS or ArcGIS rendering engines in one project.

TypeScript🇺🇸 North AmericaAttention 46
★8,747
shy3130/tick-stock-panel

A self-hosted quant workbench for A-share strategy building — TSP is a self-hosted, zero-ops A-share quant workbench combining custom strategies, screening, monitoring and backtesting in one container with no external database. It exposes 65 open API endpoints and 12 MCP tools, and uses LLMs for strategy customization and stock review, aimed at individual traders wanting to self-host their own quant setup.

Python🇨🇳 Greater ChinaAttention 53
★8,092
lance-format/lance

A lakehouse file format built for multimodal AI workloads — Lance is an open lakehouse format combining vector search, BM25 full-text search, and SQL analytics on the same dataset, with fast random access and data versioning. It converts from Parquet in two lines of code and works with Pandas, DuckDB, Polars, PyArrow, Ray, and Spark, targeting ML training and multimodal data storage (images, video, audio, embeddings).

Rust🇺🇸 North AmericaAttention 54
★7,373
feigeCode/navop

One native Rust app for databases, SSH, SFTP, and AI agents — Navop bundles database clients (MySQL, PostgreSQL, ClickHouse, Oracle, DuckDB and more), SSH/SFTP, terminals, remote desktop, monitoring and AI agents into one native app. Built with GPUI/Rust and GPU-accelerated rendering, it skips WebView for a lighter, faster desktop tool aimed at engineers managing varied infra and data sources.

RustAttention 33
★2,195
ClickHouse/ClickBench

A Reproducible, Multi-Engine Benchmark for Analytical Databases — ClickBench tests dozens of OLAP, OLTP, serverless, and specialized databases (including DuckDB, BigQuery, Snowflake, Doris, DataFusion, chDB) against a real anonymized web-analytics dataset. Results are reproducible in ~20 minutes via documented shell scripts, making it useful for anyone comparing analytical database performance before adopting one.

Shell🇺🇸 North AmericaAttention 16
★1,170
dbt-labs/dbt-charts

Dashboards as YAML, written for agents and audited by humans — dbt Charts (CLI `dct`) compiles declarative YAML board definitions, wrapped around SQL, into interactive dashboards. It targets the mess of vibe-coded dashboard stacks by giving AI agents a simple syntax to write and humans a clear artifact to audit, with a playground and MCP support for agent workflows.

Python
★598
hugues31/duckpad

A visual SQL workspace built on DuckDB with cell-level lineage — Duckpad is an open-source SQL workspace built on DuckDB that lets you compose data pipelines visually, explore results, and trace any result cell back to its source values. It adds Git-based versioning, repeatable Jobs, and a context-aware AI assistant, aimed at data engineers who want transparency and control over transformations.

Python🇪🇺 EuropeAttention 0
★304
Dayan-18/earthquake-olap-duckdb-wasm

A No-Server OLAP Stack: USGS Quakes, Parquet, and DuckDB-WASM in the Browser — This project turns two years of USGS earthquake data into a Parquet star schema that DuckDB-WASM queries entirely client-side. GitHub Actions handles daily ETL, 10 data-quality checks, and redeployment to GitHub Pages, with no servers, databases, or AWS involved. Useful as a template for anyone building analytics dashboards without backend infrastructure.

JavaScript
★1