黑洞智能体BLACK HOLE AGENTS
Docs

Multi-Window Research Workbench

Researchers need to keep reference materials and model comparisons visible while an Agent organizes files and generates reports. Splitting research tasks into independent windows allows the main window to continue searching for information without interrupting the Agent's state.

  • Main window: Keep reference conversations and global search;
  • Independent window: Open the research Agent task and pin it to the top;
  • Second tab: Open the knowledge base or notes to verify original sources at any time.
The assistant model selector offers Multi-select for sending the same question to multiple models. (Chinese interface.)The assistant model selector offers Multi-select for sending the same question to multiple models. (Chinese interface.)

The assistant model selector offers Multi-select for sending the same question to multiple models. (Chinese interface.)

Workflow

1. Compare research questions in [Chat]

Use multi-model comparison to identify consensus, conflicts, and items requiring verification. Do not rush to generate the final report.

2. Prepare directories for the research Agent

Separate raw materials from output directories, bind relevant knowledge bases, and require external information to retain links.

3. Split the task into an independent window

Right-click the Agent tab and select [Open in New Window]. Use window pinning if needed. The main window continues searching for information without interrupting the Agent's state.

4. Summarize and review

Check the report in the [Files] panel on the right side of the Agent, and open the original conversation in the main window to verify conflicts. Only retain conclusions that can be traced back to sources or references.

Right-click a tab to pin it, open it in a new window or close other tabs. (Chinese interface.)Right-click a tab to pin it, open it in a new window or close other tabs. (Chinese interface.)

Right-click a tab to pin it, open it in a new window or close other tabs. (Chinese interface.)

WindowSuggested Content
Main windowReference conversations, global search, and notes
Independent windowOne running Agent task
Pinned tabMost frequently used knowledge base or task for the project
Completion criteriaSources, task status, and final notes correspond to each other; windows are retracted or closed after the task ends

This setup is suitable for research requiring continuous observation of long-running tasks. It is not recommended to open many windows simultaneously just to "look more professional."

On this page