FIX: Thought Chain UI Improvements

[tahierhussain]

What

• Updated initial thought chain message from "Preparing database information..." to "Gathering context..."
• Consolidated EnhancedPromptInfo and PromptInfo into a single PromptInfo component
• Made thought chain collapsible (collapsed by default, showing latest message with step count)
• Added React performance optimizations (memoization, removed unused props)
• Fixed duplicate loader issue when collapse is expanded during streaming

Why

• The previous message "Preparing database information..." was too specific and didn't accurately describe the initial processing step
• Having two separate components (EnhancedPromptInfo and PromptInfo) added unnecessary complexity
• Collapsible thought chain improves UX by reducing visual clutter while still allowing users to see full history
• React optimizations prevent unnecessary re-renders and improve performance

How

• Backend: Changed the content string in llm_context.py
• Frontend:
  • Renamed EnhancedPromptInfo.jsx to PromptInfo.jsx and removed legacy component
  • Added Ant Design Collapse component for collapsible UI
  • Removed unused props (isThoughtChainReceived, llmModel, coderLlmModel, chatIntent)
  • Added useMemo for processedChain and errorDetails objects
  • Added condition to hide header loader when expanded (isInProgress && !isExpanded)

Can this PR break any existing features. If yes, please list possible items. If no, please explain why. (PS: Admins do not merge the PR without this section filled)

• No, this PR should not break existing features
• The thought chain functionality remains the same, only the UI presentation has changed
• All props that were removed were already unused
• The component API is backward compatible

Database Migrations

• None

Env Config

• None

Relevant Docs

• None

Related Issues or PRs

• None

Dependencies Versions

• No new dependencies added

Notes on Testing

• Test thought chain display during prompt execution (streaming)
• Verify collapsible behavior: collapsed by default, shows "(N previous steps)" count
• Verify expanded view shows all messages in chronological order
• Verify loader only appears in header when collapsed, and in body when expanded
• Test error scenarios to ensure error popovers still work correctly

Screenshots

Collapsed

Expanded

Checklist

I have read and understood the Contribution Guidelines.

[greptile-apps]

Greptile Summary

This PR consolidates EnhancedPromptInfo and PromptInfo into a single PromptInfo component, introduces a collapsible thought-chain UI using Ant Design's Collapse, and updates the initial backend message to "Gathering context...". All five issues flagged in previous review rounds have been resolved:

• The "Processing..." shimmer now correctly returns null for completed conversations (shouldStream guard on line 566).
• The "(0 previous steps)" label is hidden when previousCount === 0.
• The errorDetails variable shadow was renamed to msgErrorDetails.
• React keys now use the message's display text instead of [object Object].
• The overly-broad disapproval-reason fallback regex was removed — only the explicit [DISAPPROVE REASON] prefix is matched.

Changes by file:

• llm_context.py: Correct, minimal string change.
• Conversation.jsx: llmModels, EnhancedPromptInfo, and several unused useMemo values removed; errorDetailsMemo added for a stable object reference passed to PromptInfo.
• PromptInfo.jsx: Thought chain is now collapsible (collapsed by default), the header spinner is correctly suppressed when expanded, and the component returns null for historical conversations with no active stream.
• ThoughtChainEnhancements.css: Clean new utility classes for the collapsible layout.
• EnhancedPromptInfo.jsx: Deleted — no longer needed.
• Minor nit: The err parameter in the copyErrorToClipboard catch block is declared but never used (see inline comment).

Confidence Score: 5/5

Safe to merge — all previous P1/P2 issues are resolved and no new blocking issues were found.

All five issues identified in prior review rounds have been addressed. The only remaining finding is a cosmetic P2 lint nit (unused catch variable). The component consolidation is clean, removed props were confirmed unused, and the collapsible UX change is backward-compatible.

No files require special attention; PromptInfo.jsx is the most heavily changed file but its logic is sound.

Important Files Changed

Filename	Overview
backend/backend/application/context/llm_context.py	Straightforward string change from 'Preparing database information...' to 'Gathering context...'
frontend/src/ide/chat-ai/Conversation.jsx	Removed EnhancedPromptInfo import, unused llmModels prop, and computed values; added errorDetailsMemo for a stable prop reference
frontend/src/ide/chat-ai/EnhancedPromptInfo.jsx	File deleted — functionality consolidated into PromptInfo.jsx
frontend/src/ide/chat-ai/ExistingChat.jsx	No longer passes llmModels to Conversation; all other Conversation props remain unchanged
frontend/src/ide/chat-ai/PromptInfo.jsx	Merged EnhancedPromptInfo into PromptInfo; added collapsible thought chain via Ant Design Collapse; all five previous review issues resolved; minor unused catch variable remains
frontend/src/ide/chat-ai/ThoughtChainEnhancements.css	Added CSS for collapsible thought chain: .thought-chain-collapse, .thought-chain-collapse-header, and related layout classes

Flowchart

%%{init: {'theme': 'neutral'}}%%
flowchart TD
    A["Conversation.jsx\nreceives message prop"] --> B["Derives errorDetailsMemo\n(useMemo)"]
    B --> C["PromptInfo\nthoughtChain / shouldStream / errorDetails"]
    C --> D{thoughtChain\nnot empty?}
    D -- Yes --> E["processedChain useMemo\npairs attempt + DISAPPROVE REASON"]
    E --> F["renderThoughtChain()"]
    F --> G["Ant Design Collapse\ncollapsed by default"]
    G -- Collapsed --> H["Header: latest message\n+ N previous steps count"]
    G -- Expanded --> I["Body: full timeline\nrenderMessage() for each step"]
    D -- No --> J{shouldStream?}
    J -- Yes --> K["Fallback shimmer\n'Processing...'"]
    J -- No --> L["return null\n(completed conversation)"]

Loading

Prompt To Fix All With AI

This is a comment left during a code review.
Path: frontend/src/ide/chat-ai/PromptInfo.jsx
Line: 140-148

Comment:
**Unused `err` parameter in catch block**

The caught error is declared but never used — neither logged nor surfaced in the notification description. This will trigger a `no-unused-vars` ESLint warning and silently discards the failure reason, making clipboard issues harder to debug.

```suggestion
      .catch(() => {
        notification.error({
          message: "Copy failed",
          description: "Could not copy to clipboard",
          placement: "topRight",
          duration: 2,
        });
      });
```

How can I resolve this? If you propose a fix, please make it concise.

_{Reviews (6): Last reviewed commit: "fix: remove stale llmModels prop passed ..." | Re-trigger Greptile}

[abhizipstack]

LGTM

[wicky-zipstack]

LGTM — clean consolidation, all prior review issues resolved. Two minor P2s for follow-up: unused llmModels required prop in Conversation.jsx, and latest message duplication when Collapse is expanded.

[tahierhussain]

LGTM — clean consolidation, all prior review issues resolved. Two minor P2s for follow-up: unused llmModels required prop in Conversation.jsx, and latest message duplication when Collapse is expanded.

@wicky-zipstack Thanks for pointing it out. Both P2 items have been addressed as well.