Features

Previous

AGX is a hybrid desktop and web-based analytics application built with Tauri , SvelteKit, and Plot, designed to query and explore data in ClickHouse and Apache Iceberg. It combines native performance with web flexibility, offering a modern interface for data analysts and engineers. The public repository is available at https://github.com/agnosticeng/agx . Below are the key features of AGX, enabling efficient data exploration and querying.

AGX leverages Tauri to deliver native desktop application performance on macOS and Linux, using system webviews (WebKitGTK on Linux, WKWebView on macOS) for rendering and a Rust backend for efficiency and security. With a minimal footprint (as low as 600 KB), it also runs as a web app at https://agx.app , combining the speed of native execution with the accessibility of web technologies like SvelteKit.

AGX provides an interactive SQL query editor with syntax highlighting, streamlining query composition and debugging. The editor supports complex SQL queries for ClickHouse and Apache Iceberg, offering real-time feedback and a user-friendly experience for crafting analytical queries.

AGX integrates with large language models (LLMs) to enhance query creation and data exploration:

  • Agnostic AI: Built-in AI assistance suggests SQL queries, optimizes syntax, and provides insights based on data patterns.
  • Ollama: Supports local Ollama models for query assistance in the web version. Configure with OLLAMA_ORIGINS="https://agx.app" ollama serve to enable AI-driven query recommendations, leveraging models like those from the Ollama project .

The schema browser allows users to navigate and inspect the structure of datasets in ClickHouse or Iceberg tables. It displays tables, columns, data types, and relationships, enabling users to understand data organization before querying, which is particularly useful for complex datasets like those from https://github.com/agnosticeng/datasets .

Query results are presented in a clean, tabular format, optimized for readability and analysis. AGX supports sorting, filtering, and exporting results, making it easy to interpret large datasets and share findings.

AGX supports drag-and-drop functionality for loading dataset files (e.g., CSV, Parquet, or SQL scripts) into the interface. This simplifies importing data from repositories like https://github.com/agnosticeng/datasets for immediate querying, enhancing workflow efficiency.

AGX runs natively on macOS and Linux, leveraging Tauri’s cross-platform capabilities for efficient desktop performance. Additionally, it operates as a web application accessible via modern browsers at https://agx.app , ensuring flexibility for users across different systems without requiring installation.

AGX integrates with the Agnostic Asynchronous Query Processor (AGP) for asynchronous query execution against ClickHouse clusters. Users can specify a custom AGP server via a query parameter (e.g., https://agx.app?proxy=https://your-agp-server). The default AGP proxy uses datasets from https://github.com/agnosticeng/datasets , enabling seamless querying of pre-loaded data. This integration supports AGP’s features like query deduplication and tier-based resource allocation, ensuring efficient query processing.

Previous Tutorial