July 1, 2026 — Data Engineering Agent Release Notes

This release (Release 2026.07.01, v2.3.0_Data Agent) introduces new features, enhancements, and fixes. Note that these updates roll out gradually across the following regions and will be fully deployed within one to two weeks of the release date, depending on your region.

• Domestic regions

  • Alibaba Cloud (Shanghai)
  • Tencent Cloud (Shanghai / Beijing / Guangzhou)
  • AWS (Beijing)
  • Alibaba Cloud (Hangzhou)

• International regions

  • Alibaba Cloud (Singapore)
  • AWS (Singapore)

🖥️ Frontend Experience

Bug Fixes

• Streaming output continuity fix: Fixed an issue where streaming output was interrupted or inconsistent when re-entering a conversation after a page refresh. Streaming output now resumes correctly after refresh.

Experience Improvements

• Output display optimization: Improved the current-task output area and added display of the Agent's execution plan.

• Conversation flow simplification: After a response completes, intermediate execution steps are automatically collapsed, showing only the key results for a cleaner interface.

• Homepage and global interaction polish: Overall style and interaction refinements across the homepage and other pages.

• Folder context support: Entire folders can now be added as conversation context, so you no longer need to add files one at a time.

⚙️ Backend Performance and Stability

• Model info caching: Added a caching layer for model display information to reduce redundant requests and improve load speed.
• SQL output display optimization: SQL-type outputs are no longer displayed as standalone items, reducing interface clutter.
• Long-term memory and context compression optimization: Improved the long-term memory mechanism and context compression strategy to increase accuracy and efficiency across multi-turn conversations.
• High-risk operation interception (new): Added an interrupt confirmation mechanism for high-risk tool calls such as delete task, preventing accidental operations.

✨ New Features

1. Onboarding Guidance

A new onboarding / scenario guidance feature helps users understand and get started with the product's core capabilities faster.

2. Data Source and Task Management Enhancements

• Data source sampling and connectivity test tools (new): Supports sampling preview and connectivity testing for data sources.

• Task save configuration enhancements:

  • Added support for parameterized configuration, skip execution, run timeout, output table, and other configuration options
  • Note: The owner field is not configurable by the Agent due to permission restrictions
  • All other parameters can be modified

• Task dependency policy upgrade: Task dependencies now support advanced dependency policy configuration.

• Flow and subnode parameterization: Both Flows and subnodes support parameter-level modifications; subnodes support custom timeout configuration.

• Task group management: Task groups now support two basic operations: Add Member and Remove Member.

• Data source save capability expansion: Python, Shell, and JDBC task types now support saving data source configuration.

3. Report Generation

• Added report output, with one-click download support. Current supported formats: Markdown and HTML.
• Added a Data Lineage Drawing Skill that automatically generates data lineage diagrams. Select a target task or folder and ask: "Help me analyze the task lineage for this folder." The Agent returns a business overview, task dependency DAG, data warehouse layer breakdown, and field-level lineage. ![]

4. Data Integration Enhancements

• Added an offline integration tool
• Added a multi-table offline integration tool, supporting batch table-level offline data integration scenarios