You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
[FEATURE]:Allow reasoning summaries to be manually expandable per message (not just global toggle) #31825
I have verified this feature I'm about to request hasn't been suggested before.
Describe the enhancement you want to request
name: Feature Request
description: Allow reasoning summaries to be manually expandable per message
title: "[FEATURE]: Allow reasoning summaries to be manually expandable per message (not just global toggle)"
labels: [enhancement]
body:
type: textarea
id: description
attributes:
label: Description
description: A clear and concise description of what you want to happen.
value: | Problem
Currently, the OpenCode desktop UI has a single global setting: "Show reasoning summary" (on/off) that controls whether all reasoning summaries are displayed or hidden. There is no way to see reasoning for individual messages without toggling the global setting, which then shows all reasoning summaries at once.
**Desired Behavior**
Allow reasoning summaries to be set to "collapsed" by default (via the existing toggle), but make each summary **manually expandable by clicking** on it — similar to how the shell/CLI version works, where you can choose to expand a specific reasoning block per message.
**Use Case**
- I want to keep the conversation clean (reasoning hidden by default)
- But occasionally I want to peek at the reasoning for a *specific* response without having to toggle the global setting on/off
**Suggestion**
Change the behavior so that when "Show reasoning summary" is OFF, each reasoning block is rendered as a collapsible element (e.g., a small "Show reasoning" toggle per message), and clicking it expands only that one block. This gives per-message granularity while preserving a clean default view.
validations:
required: true
type: textarea
id: alternatives
attributes:
label: Alternatives
description: A clear and concise description of any alternative solutions or features you've considered.
value: |
1. Keep the global toggle but add a third state: "Hidden / Collapsible / Always visible"
2. Add a keyboard shortcut to expand/collapse reasoning for the last message
3. No change — use shell version instead (which already supports this behavior)
type: textarea
id: context
attributes:
label: Additional context
value: |
- The shell/CLI version of OpenCode already supports per-message expandable reasoning — this request is about bringing parity to the desktop UI
- OS: Windows
- OpenCode version: latest
Reacted by Cpt, Mike Zheng, ezeanthony444, Cae2, Anton and Ahmad Ahmadi
I mean this is the "industry standard", all harnesses do it this way, this is my main pain point in using opencode now. Apart from the reasoning not being "collapsible" its very difficult to differentiate the reasoning from the actual resulting output, because its just a slightly smaller font size.
I liked opencode, but the UI is the main reason I am now trying out alterantives (Reasonix - first impression is a clean UI and thinking collapsible like I expect), even in two hierarchies:
First grouping together one whole loop of activities in one:
And then neatly expandable / collapsable activities so I can choose to what I want to dig into:
In the OpenCode v2 TUI, reasoning can be expanded per message. With thinking hidden (toggle it with the "Expand/Collapse thinking" command), each reasoning block shows as a "Thought · N steps" row; click it to expand only that block.
Added in #36901; the expanded state is kept since #48489.
Closing as completed. If something doesn't work on v2, please open a new issue.
Activity
yale-Liyao commentedon Jul 1, 2026
Not just reasoning summaries, but also shell, should be foldable too
baschny commentedon Aug 5, 2026
I mean this is the "industry standard", all harnesses do it this way, this is my main pain point in using opencode now. Apart from the reasoning not being "collapsible" its very difficult to differentiate the reasoning from the actual resulting output, because its just a slightly smaller font size.
I liked opencode, but the UI is the main reason I am now trying out alterantives (Reasonix - first impression is a clean UI and thinking collapsible like I expect), even in two hierarchies:
First grouping together one whole loop of activities in one:
And then neatly expandable / collapsable activities so I can choose to what I want to dig into:
Very intuitive and clean.
neriousy commentedon Oct 1, 2026
In the OpenCode v2 TUI, reasoning can be expanded per message. With thinking hidden (toggle it with the "Expand/Collapse thinking" command), each reasoning block shows as a "Thought · N steps" row; click it to expand only that block.
Added in #36901; the expanded state is kept since #48489.
Closing as completed. If something doesn't work on v2, please open a new issue.