v2.51.0

New Features

  • AI Agent Infrastructure Enhancements

    • Implemented execution tracking with detailed history for agent runs.
    • Added markdown support for action progress messages and improved breadcrumb hierarchy.
    • Introduced GraphQL streaming support for real-time agent execution.
    • Introduced new Web Search action linked to demo agents.
    • Added support for filtering SQL logging by user identity.
  • AI Test Harness Redesign

    • Redesigned UI with custom buttons, JSON syntax highlighting, and modal previews.
    • Added support for multiple execution modes and dynamic UI switching.
    • Implemented focus management and improved execution feedback display.
  • AI Prompt Execution Enhancements

    • Added customizable form interface for prompt runs with model/response format configuration.
    • Clarified model priority handling and improved drag-and-drop support.
  • Metadata & Migration Support

    • Introduced comprehensive validation system to MetadataSync with smart field handling.
    • Added new Qwen 3 32B model (Alibaba Cloud vendor) and updated schema references.
    • Upgraded migration metadata versioning from v2.50.x to v2.51.x.
  • Advanced Configuration Features

    • Enabled SQL Server connection pooling configuration via MJServer.
    • Enhanced Advanced Settings with code editor integration and tabbed interface.

Improvements

  • Core Actions Framework

    • Made parameter handling case-insensitive and improved execution reliability.
    • Added detailed logging options with configurable verbosity levels.
  • Explorer & Settings UIs

    • Modernized UI components with new designs and layout improvements.
    • Refined dashboard performance and loading behavior within MJ Explorer shell.
  • Action Execution Monitor

    • Enhanced UI with node-depth-aware indentation, dynamic component creation, and streaming updates.
    • Improved clarity and responsiveness of progress indicators.
  • Development & Documentation

    • Added connection pooling and circular dependency resolution guides.
    • Cleaned up internal structure of AI packages to break circular dependencies via shared type centralization.

Bug Fixes

  • Fixed circular dependency between @memberjunction/ai-prompts and @memberjunction/aiengine that blocked builds.
  • Resolved transaction concurrency errors (EREQINPROG) in SQLServerDataProvider.
  • Fixed layout issues in AI Dashboards and tab switching inconsistencies.
  • Eliminated duplicate test harness windows and improved SafeValue display handling.
  • Suppressed spurious errors during metadata sync record creation.
  • Corrected breadcrumb logic and dashboard x-axis labeling.