Enhance inline thinking blocks with reasoning summaries for GPT-5 content by Copilot · Pull Request #1 · bastikohn/opencode

Copilot · 2025-08-23T12:47:22Z

This PR enhances the existing inline thinking blocks functionality in the OpenCode TUI to display GPT-5 reasoning content with improved formatting and status indicators.

Overview

The TUI previously supported basic inline thinking blocks via the /thinking command, but lacked enhanced formatting for AI reasoning content. This implementation improves the existing inline thinking blocks to provide better presentation of reasoning summaries without introducing a separate view.

Key Features

🎯 Enhanced Inline Thinking Blocks

Improved formatting for reasoning content with automatic text truncation for better readability
Status Indicators: Visual indicators show ● for active thinking and ✓ Reasoning Summary for completed reasoning
Smart Summarization: Automatically truncates long reasoning content to ~500 characters with intelligent break points
Real-time Updates: Status indicators update as reasoning parts stream in

🔄 Seamless Integration

Enhanced existing /thinking command and <leader>b keybinding functionality
Maintains full backward compatibility with current thinking blocks behavior
Integrated with existing message rendering pipeline for consistent theming
No changes to user workflow - same commands, better presentation

Implementation Details

The enhancement processes ReasoningPart content from assistant messages and displays them inline with:

Automatic text truncation to maintain summary format while preserving readability
Status-aware labels that show completion state with visual indicators
Enhanced rendering that integrates seamlessly with existing message theming

Usage

Users continue to use the same familiar commands:

/thinking command - toggles enhanced inline thinking blocks
<leader>b keybinding - keyboard shortcut for quick access

When activated, reasoning content appears as enhanced inline thinking blocks within the normal message flow, providing a focused way to review AI reasoning without disrupting the conversation context.

Testing

All existing functionality preserved with enhanced presentation. The implementation maintains the same toggle behavior while providing improved visual formatting for reasoning content.

💡 You can make Copilot smarter by setting up custom instructions, customizing its development environment and configuring Model Context Protocol (MCP) servers. Learn more Copilot coding agent tips in the docs.

Co-authored-by: bastikohn <673521+bastikohn@users.noreply.github.com>

Copilot

Pull Request Overview

This PR implements a dedicated reasoning summaries view in the OpenCode TUI application to display GPT-5 thinking content in a focused, split-screen interface alongside regular chat messages.

Adds a new ReasoningView component for displaying AI reasoning content in an organized format
Integrates the reasoning view with the existing TUI layout system using a split-screen approach
Updates the /thinking command to control both inline blocks and the new dedicated reasoning view

Reviewed Changes

Copilot reviewed 4 out of 4 changed files in this pull request and generated 4 comments.

File	Description
packages/tui/internal/tui/tui_reasoning_view.go	Core implementation of the ReasoningView component with reasoning summary extraction and rendering
packages/tui/internal/tui/tui_reasoning_view_test.go	Unit tests for reasoning view functionality including summary extraction and text processing
packages/tui/internal/tui/tui.go	Integration of reasoning view into main TUI model with split layout and update handling
packages/tui/internal/commands/command.go	Updated command description to reflect "reasoning summaries" functionality

_{Tip: Customize your code reviews with copilot-instructions.md. Create the file or learn how to get started.}

Copilot · 2025-08-23T13:04:57Z

+			// Only get reasoning from the last assistant message for focused view
+			if assistantMsg.Time.Completed == 0 {
+				break
+			}


This logic has a potential bug - the break condition only applies when assistantMsg.Time.Completed == 0, but the loop continues processing messages even after finding incomplete assistant messages. This could skip reasoning parts from newer messages. Consider breaking when finding any incomplete message or restructuring to process messages in reverse order.

Suggested change

// Only get reasoning from the last assistant message for focused view

if assistantMsg.Time.Completed == 0 {

break

}

// Only get reasoning from the most recent assistant message for focused view

break

Copilot · 2025-08-23T13:04:58Z

+	longText := make([]byte, 600)
+	for i := range longText {
+		longText[i] = 'a'
+	}
+	longTextStr := string(longText)


This manual byte slice creation can be simplified using strings.Repeat(). Replace lines 117-120 with: longTextStr := strings.Repeat(\"a\", 600).

Suggested change

longText := make([]byte, 600)

for i := range longText {

longText[i] = 'a'

}

longTextStr := string(longText)

longTextStr := strings.Repeat("a", 600)

Copilot · 2025-08-23T13:04:58Z

+			// Show reasoning view alongside messages in a split layout
+			messagesWidth := effectiveWidth / 2


[nitpick] The 50/50 split layout is hardcoded. Consider making the split ratio configurable or dynamic based on content length to provide better user experience with varying reasoning content sizes.

Suggested change

// Show reasoning view alongside messages in a split layout

messagesWidth := effectiveWidth / 2

// Dynamically calculate split ratio based on content height

messagesHeight := lipgloss.Height(messagesView)

reasoningHeight := lipgloss.Height(reasoningView)

var splitRatio float64

totalHeight := messagesHeight + reasoningHeight

if totalHeight == 0 {

splitRatio = 0.5

} else {

splitRatio = float64(messagesHeight) / float64(totalHeight)

// Clamp splitRatio between 0.2 and 0.8 for reasonable UI

if splitRatio < 0.2 {

splitRatio = 0.2

} else if splitRatio > 0.8 {

splitRatio = 0.8

}

}

messagesWidth := int(float64(effectiveWidth) * splitRatio)

bastikohn

@copilot I did not want a split screen layout. Integrate the thinking summaries in the in-line thinking blocks.

bastikohn · 2025-08-23T16:10:12Z

@copilot I did not want a split screen layout. Integrate the thinking summaries in the in-line thinking blocks. Please update the implementation accordingly.

Co-authored-by: bastikohn <673521+bastikohn@users.noreply.github.com>

Copilot · 2025-08-23T16:18:40Z

@copilot I did not want a split screen layout. Integrate the thinking summaries in the in-line thinking blocks.

I've integrated the reasoning summaries into the inline thinking blocks as requested. The split-screen layout has been removed and reasoning content now appears inline within the message flow. The summaries include status indicators (● for active thinking, ✓ for completed reasoning) and automatically truncate long content to maintain readability. Commit: 8dcc125

@copilot I did not want a split screen layout. Integrate the thinking summaries in the in-line thinking blocks. Please update the implementation accordingly.

Updated! The reasoning summaries are now integrated directly into the inline thinking blocks instead of using a separate split-screen layout. The implementation enhances the existing thinking blocks with summary formatting, status indicators, and automatic text truncation while maintaining the familiar inline display. Commit: 8dcc125

Initial plan

7a0ce7d

Copilot AI assigned Copilot and bastikohn Aug 23, 2025

Copilot started work on behalf of bastikohn August 23, 2025 12:47 View session

Copilot AI and others added 2 commits August 23, 2025 12:58

Implement TUI reasoning summaries view with dedicated component

e049a8b

Co-authored-by: bastikohn <673521+bastikohn@users.noreply.github.com>

Add comprehensive tests and documentation for TUI reasoning view

9d1809b

Co-authored-by: bastikohn <673521+bastikohn@users.noreply.github.com>

Copilot AI requested a review from bastikohn August 23, 2025 13:02

Copilot finished work on behalf of bastikohn August 23, 2025 13:02

bastikohn requested a review from Copilot August 23, 2025 13:03

Copilot AI reviewed Aug 23, 2025

View reviewed changes

bastikohn requested changes Aug 23, 2025

View reviewed changes

bastikohn assigned Copilot and unassigned bastikohn and Copilot Aug 23, 2025

Copilot started work on behalf of bastikohn August 23, 2025 16:10 View session

Integrate reasoning summaries into inline thinking blocks

8dcc125

Co-authored-by: bastikohn <673521+bastikohn@users.noreply.github.com>

Copilot AI changed the title ~~Implement TUI reasoning summaries view for GPT-5 thinking content~~ Enhance inline thinking blocks with reasoning summaries for GPT-5 content Aug 23, 2025

Copilot AI requested a review from bastikohn August 23, 2025 16:19

Copilot finished work on behalf of bastikohn August 23, 2025 16:19

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Enhance inline thinking blocks with reasoning summaries for GPT-5 content#1

Enhance inline thinking blocks with reasoning summaries for GPT-5 content#1
bastikohn with Copilot wants to merge 4 commits into
feat/gpt5-thinking-uifrom
copilot/fix-4aad90b8-f0a7-4755-8dcb-0426df7a94b6

Copilot AI commented Aug 23, 2025 •

edited

Loading

Uh oh!

Copilot AI left a comment

Uh oh!

Copilot AI Aug 23, 2025

Uh oh!

Copilot AI Aug 23, 2025

Uh oh!

Copilot AI Aug 23, 2025

Uh oh!

bastikohn left a comment •

edited

Loading

Uh oh!

bastikohn commented Aug 23, 2025

Uh oh!

Copilot AI commented Aug 23, 2025 •

edited

Loading

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

		// Show reasoning view alongside messages in a split layout
		messagesWidth := effectiveWidth / 2

-			// Show reasoning view alongside messages in a split layout
-			messagesWidth := effectiveWidth / 2
+			// Dynamically calculate split ratio based on content height
+			messagesHeight := lipgloss.Height(messagesView)
+			reasoningHeight := lipgloss.Height(reasoningView)
+			var splitRatio float64
+			totalHeight := messagesHeight + reasoningHeight
+			if totalHeight == 0 {
+				splitRatio = 0.5
+			} else {
+				splitRatio = float64(messagesHeight) / float64(totalHeight)
+				// Clamp splitRatio between 0.2 and 0.8 for reasonable UI
+				if splitRatio < 0.2 {
+					splitRatio = 0.2
+				} else if splitRatio > 0.8 {
+					splitRatio = 0.8
+				}
+			}
+			messagesWidth := int(float64(effectiveWidth) * splitRatio)

Conversation

Copilot AI commented Aug 23, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Overview

Key Features

🎯 Enhanced Inline Thinking Blocks

🔄 Seamless Integration

Implementation Details

Usage

Testing

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull Request Overview

Reviewed Changes

Uh oh!

Copilot AI Aug 23, 2025

Choose a reason for hiding this comment

Uh oh!

Copilot AI Aug 23, 2025

Choose a reason for hiding this comment

Uh oh!

Copilot AI Aug 23, 2025

Choose a reason for hiding this comment

Uh oh!

bastikohn left a comment • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

bastikohn commented Aug 23, 2025

Uh oh!

Copilot AI commented Aug 23, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

Copilot AI commented Aug 23, 2025 •

edited

Loading

bastikohn left a comment •

edited

Loading

Copilot AI commented Aug 23, 2025 •

edited

Loading