fix(platform): prevent chat loading state flicker during tool call transitions

[larryro]

Summary

• Add hasIncompleteAssistantMessage check to useMessageProcessing that detects when the last assistant message is still in a non-terminal state (streaming or pending)
• Include this flag in the isLoading computation in ChatInterface to keep the loading indicator active during status transitions between streaming and pending (e.g., during tool calls)
• Add comprehensive tests for useMessageProcessing and useChatPendingState hooks covering the new behavior and existing functionality

Test plan

• Unit tests added for hasIncompleteAssistantMessage covering: no messages, no assistant messages, success/failed/streaming/pending statuses, and multi-message scenarios
• Unit tests added for useChatPendingState covering pending state clearing on streaming, fallback, and human input responses
• Manual verification: trigger a tool call in chat and confirm the loading indicator stays active throughout the streaming → pending → streaming cycle

🤖 Generated with Claude Code

Summary by CodeRabbit

Release Notes

• Bug Fixes

  • Improved chat interface loading state to reduce UI flicker during message transitions, providing a smoother user experience when the chat transitions between different message states.

• Tests

  • Added comprehensive test coverage for chat pending state management and message processing workflows.

[greptile-apps]

Greptile Summary

Fixes a UI flicker where the chat loading indicator would briefly disappear during tool call transitions. When an assistant message transitions from streaming → pending (while a tool executes) → streaming (when the agent resumes), the prior isLoading logic (isPending || !!streamingMessage) would evaluate to false during the pending gap.

• Adds hasIncompleteAssistantMessage to useMessageProcessing — a memoized check that returns true when the last assistant message has a non-terminal status (streaming or pending)
• Includes hasIncompleteAssistantMessage in the isLoading computation in ChatInterface, bridging the gap during tool call transitions
• Adds comprehensive unit tests for both useMessageProcessing and useChatPendingState hooks

Confidence Score: 5/5

• This PR is safe to merge — it's a focused UI fix with a small, well-understood surface area and comprehensive test coverage.
• The change is minimal (one new memoized boolean + one OR condition), logically correct for all four known status values, and well-tested. The hasIncompleteAssistantMessage flag is partially redundant with !!streamingMessage during the streaming phase, but this is harmless in a boolean OR. No risk of stuck states beyond what already exists with the streamingMessage check, since the Convex SDK guarantees transition to terminal states.
• No files require special attention

Important Files Changed

Filename	Overview
services/platform/app/features/chat/hooks/use-message-processing.ts	Adds hasIncompleteAssistantMessage computed via useMemo — checks if last assistant message status is non-terminal (streaming/pending). Clean logic, correct use of findLast, proper memoization dependency.
services/platform/app/features/chat/components/chat-interface.tsx	Integrates hasIncompleteAssistantMessage into isLoading boolean. This keeps loading indicator active during streaming→pending→streaming transitions. The flag is partially redundant with !!streamingMessage during the streaming phase, but the redundancy is harmless and the pending coverage is the key improvement.
services/platform/app/features/chat/hooks/tests/use-message-processing.test.ts	Comprehensive test coverage for hasIncompleteAssistantMessage, streamingMessage, and pendingToolResponse. Covers edge cases: no messages, user-only messages, all four status values, multi-message ordering.
services/platform/app/features/chat/hooks/tests/use-chat-pending-state.test.ts	New tests for useChatPendingState covering streaming clearing, fallback clearing, human input response behavior, and combined scenarios. Well-structured with proper use of renderHook, rerender, and act.

Sequence Diagram

sequenceDiagram
    participant User
    participant ChatInterface
    participant useMessageProcessing
    participant UIMessages

    User->>ChatInterface: Send message
    ChatInterface->>ChatInterface: isPending = true

    UIMessages-->>useMessageProcessing: assistant status = "streaming"
    useMessageProcessing-->>ChatInterface: streamingMessage ✓, hasIncomplete ✓
    ChatInterface->>ChatInterface: isLoading = true (streaming)

    Note over UIMessages: Tool call begins
    UIMessages-->>useMessageProcessing: assistant status = "pending"
    useMessageProcessing-->>ChatInterface: streamingMessage ✗, hasIncomplete ✓
    ChatInterface->>ChatInterface: isLoading = true (hasIncomplete prevents flicker)

    Note over UIMessages: Tool call completes
    UIMessages-->>useMessageProcessing: assistant status = "streaming"
    useMessageProcessing-->>ChatInterface: streamingMessage ✓, hasIncomplete ✓
    ChatInterface->>ChatInterface: isLoading = true (streaming resumes)

    UIMessages-->>useMessageProcessing: assistant status = "success"
    useMessageProcessing-->>ChatInterface: streamingMessage ✗, hasIncomplete ✗
    ChatInterface->>ChatInterface: isLoading = false

Loading

_{Last reviewed commit: ea94fad}

[coderabbitai]

📝 Walkthrough

Walkthrough

This PR introduces a new hasIncompleteAssistantMessage computed field to the useMessageProcessing hook that identifies when the last assistant message in a thread is in a non-terminal state (status not 'success' or 'failed'). The chat interface component then incorporates this flag into its loading state calculation to prevent UI flicker during state transitions, such as when a message status changes from 'streaming' to 'pending' during tool calls. Supporting tests are added for both the hook and related pending state logic.

Estimated code review effort

🎯 2 (Simple) | ⏱️ ~15 minutes

Possibly related PRs

• feat(platform): paginate thread list and reduce initial page sizes #482: Modifies the same useMessageProcessing hook to adjust streaming-related behavior
• feat: chat ux improvements and mobile responsiveness fixes #150: Updates rendering and loading logic in the chat interface component that consumes this hook

🚥 Pre-merge checks | ✅ 2 | ❌ 1

❌ Failed checks (1 warning)

Check name	Status	Explanation	Resolution
Docstring Coverage	⚠️ Warning	Docstring coverage is 25.00% which is insufficient. The required threshold is 80.00%.	Write docstrings for the functions missing them to satisfy the coverage threshold.

✅ Passed checks (2 passed)

Check name	Status	Explanation
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.
Title check	✅ Passed	The title clearly summarizes the main change: preventing chat loading state flicker during tool call transitions, which directly reflects the core fix in the changeset.

_{✏️ Tip: You can configure your own custom pre-merge checks in the settings.}

✨ Finishing Touches

• 📝 Generate docstrings

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Post copyable unit tests in a comment
• Commit unit tests in branch fix/chat-loading-state-flicker

---

_{Comment @coderabbitai help to get the list of available commands and usage tips.}

[coderabbitai]

Actionable comments posted: 1

🤖 Prompt for all review comments with AI agents

Verify each finding against the current code and only fix it if needed.

Inline comments:
In
`@services/platform/app/features/chat/hooks/__tests__/use-message-processing.test.ts`:
- Around line 22-33: The test fixture createUIMessage currently force-casts the
returned object with "as UIMessage"; replace that cast by using TypeScript's
"satisfies UIMessage" on the object literal so the compiler verifies
compatibility without widening types — keep the function signature the same,
merge overrides as before, and remove the trailing "as UIMessage" so the
returned object satisfies UIMessage while preserving precise literal types.