DOCSEARCHbyAlgolia
Submit Docs

QRY Documentation

website·Indexed 31 days ago

https://docs.qry.dev

Documentation for QRY, a platform for chatting with databases using AI. Features include AI-powered notebooks with SQL and Python, scheduled tasks with multiple output channels, RAG-based workspace memory, domain context for business logic, and a Skills IDE for teaching the AI organizational knowledge and reusable scripts.

Tokens
95.7K
Snippets
217
Records
725
Agent score
58%

What's inside QRY

  1. Overview of Attribute-Based Access Control (ABAC)

    Attribute-Based Access Control (ABAC) provides fine-grained, row-level security that complements Role-Based Access Control (RBAC). While RBAC determines which tables a user can access, ABAC determines which specific rows within those tables are visible based on attributes (tags) and policies. Enforcement happens in two layers: AI Guidance (generating queries with correct filters) and Server-Side Validation (blocking queries that lack required filters).

  2. Overview of QRY Git Folders

    Git Folders provide professional version control for QRY notebooks, Forge pipelines, and jobs. It uses a hybrid architecture where the database serves as the primary source of truth for speed, while a Git Sync Service manages versioning, collaboration, and synchronization with remote repositories. This ensures work is never lost even if Git synchronization fails.

  3. Overview of Model Context Protocol (MCP) in QRY

    The Model Context Protocol (MCP) is an open standard that allows QRY to securely connect with external data sources and tools. It acts as a universal adapter, enabling QRY to discover tools from an MCP server and integrate them into the AI's toolset. This allows for extensibility (adding capabilities without code changes) and interoperability across different AI applications.

  4. Overview of QRY Speech-to-Text System

    The QRY speech-to-text system provides universal browser-compatible voice input. It features dual provider support and integrated Voice Activity Detection (VAD). The system is currently production-ready and fully implemented.

  5. Overview of qry-forge migration capabilities

    qry-forge is a migration platform designed to automate the transition of data warehouses from source dialects (Teradata, Oracle, SQL Server, and Cloudera) to target platforms (BigQuery and Cloudera/Iceberg). It automates DDL translation, schema mapping, row-count validation, and gate approvals. The process is organized into Projects, which consist of ordered Waves of Artifacts (tables, views, stored procedures, UDFs, or macros).

  6. Overview of Workspaces for collaborative analysis

    Workspaces act as a shared environment for teams to perform data analysis. They provide several collaborative features:

    • Shared Conversations: Team members can see, continue, and contribute to the same data discussions.
    • Collective Files: Documents uploaded to a workspace are available to all members and serve as a searchable knowledge base for the AI.
    • Shared AI Memory: Context provided to the AI (e.g., project names or specific definitions) is remembered at the workspace level for all members.
    • Activity Tracking: Members can monitor ongoing work within the workspace.

  7. Overview of QRY Unified Memory Tool

    The Unified Memory Tool provides provider-agnostic persistent memory that works across all supported LLM services, including Claude, Gemini, and OpenAI. This feature is production-ready and fully implemented.

  8. Overview of QRY Notebooks

    QRY Notebooks provide a Jupyter-like environment for creating reusable, parameterized AI prompts and structured analysis workflows. They allow users to organize prompts, document analytical thinking, share notebooks via workspaces, and schedule recurring analysis tasks. Notebooks consist of five cell types: Markdown (documentation), Prompt (parameterized AI queries), Python (code execution), SQL (database queries), and Result (captured outputs).

  9. Overview of Forge migration waves

    In Forge, a wave is the primary unit of work for migrating workloads from Teradata, Oracle, or SQL Server to BigQuery or Cloudera. A wave consists of a named set of artifacts (tables, views, procedures, UDFs, and macros) that are migrated together. Waves allow developers to sequence large migrations into smaller, reviewable, and deployable chunks.

  10. Overview of QRY DataFlow

    DataFlow is a conversational data pipeline tool that enables users to move data between connected databases using plain English descriptions. It replaces traditional ETL processes by automatically determining the optimal transfer strategy, handling schema mappings, and providing AI-driven auto-healing for pipeline failures.

  11. Overview of the QRY Scheduled Tasks System

    The Scheduled Tasks System in QRY allows for automated recurring or one-time data analysis tasks. It is currently production-ready and fully operational. Key capabilities include:

    • Natural language scheduling
    • Output delivery via conversation, email, or Telegram
    • Conditional alerts with built-in cooldown periods
    • Quota limits to manage resource usage
    • Automatic disabling of tasks upon failure

  12. Overview of QRY and DataHub Integration

    QRY integrates with DataHub, an enterprise metadata platform, to provide AI assistants with business-level context for data queries. Instead of relying solely on database schemas, QRY fetches rich metadata from DataHub to understand the meaning, ownership, and logic behind tables and fields. This ensures that AI-generated queries align with organization-specific business logic (e.g., specific calculation methods for 'churn score' or 'revenue').