openclaw - 💡(How to fix) Fix Feature request: Split chat from activity/workbench in Control UI [1 comments, 2 participants]

Summary

Add an optional split-view layout to the OpenClaw Control UI: one clean, readable chat column, plus a separate activity/workbench column for tool calls, command output, file edits, progress/status updates, and verification results.

Problem

Right now, conversational replies and execution details can occupy the same visual stream. That is transparent, but it can make the main chat harder to read — especially during longer agent runs with many tool calls, shell outputs, file edits, background tasks, or verification steps.

For day-to-day use, the user often wants two things at once:

1. A clean conversation thread that reads like a normal assistant chat.
2. Full transparency into what the agent is doing when they need to inspect it.

A single mixed stream makes those goals compete.

Related existing issues / discussions

This overlaps with existing requests to reduce tool-call clutter, especially:

• https://github.com/openclaw/openclaw/issues/6634 — option to hide/collapse tool call output in webchat UI.
• https://github.com/openclaw/openclaw/issues/21032 — hide or collapse tool call/result messages in chat view.
• https://github.com/openclaw/openclaw/issues/10985 — bug: “Toggle assistant thinking/working output” does not hide tool call blocks.

This proposal is slightly different: instead of only hiding/collapsing tool output, it suggests preserving that activity in a dedicated workbench/activity column so the chat stays clean while transparency remains visible.

Proposed solution

Introduce a split-view mode with two distinct areas:

Left / main column: Chat

• User messages
• Assistant-facing conversational replies
• Final results
• Clarifying questions
• Important summaries or blockers

This column should stay calm and readable.

Right / activity column: Workbench

• Tool calls and tool results
• Shell command output
• File edits / diffs
• Background task status
• Progress updates
• Verification results: tests, lint, build, health checks, screenshots, etc.
• Expand/collapse controls for noisy output

This column preserves transparency without burying the conversation.

Why this helps

• Keeps the main chat readable during complex runs.
• Makes agent work feel more understandable and inspectable.
• Reduces cognitive load for non-technical users while keeping technical details available.
• Makes long-running tasks easier to follow.
• Creates a natural place for progress, logs, and verification evidence without turning the chat into a terminal transcript.

UX notes

Possible behavior:

• Default to chat-first layout on narrow screens.
• Allow the activity column to be hidden, resized, or collapsed.
• Show compact status indicators in chat when workbench activity occurs, e.g. “Running tests…” or “Edited 2 files,” with details visible in the workbench.
• Group activity by assistant turn or task.
• Keep final evidence visible in chat, but put raw output/details in the workbench.
• Preserve existing transcript/export behavior, ideally with both chat and activity available.

Example

A code change run could appear as:

Chat column

• User: “Fix the bug and verify it.”
• Assistant: “I found the issue, patched it, and tests pass.”

Workbench column

• Read src/foo.ts
• Edited src/foo.ts
• Ran npm test
• Test output
• Verification summary

Acceptance criteria

• Users can read the main conversation without scrolling through raw tool output.
• Tool/action transparency is preserved in a separate visible area.
• No important result is only hidden in the workbench; summaries/blockers still appear in chat.
• Works responsively on smaller screens.

Optional extensions

• Filters for activity types: tools, shell, files, status, verification.
• Pin important verification results.
• Collapse successful routine steps by default.
• Link chat summaries to corresponding workbench activity groups.