Sundeep/eng 2362 handle user messages emitted by sub agents #823
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Updates the pending approval status bar to display the count of pending approvals when there are multiple (e.g. "3 Pending Approvals" vs "Pending Approval"). This helps users understand how many approvals are waiting. 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude <noreply@anthropic.com>
Previously only tool_use events captured parent_tool_use_id from the stream event, causing sub-agent messages to appear as regular user messages in the UI. Now all event types (text, tool_result, thinking, system) properly capture this field to enable distinguishing sub-agent activity from actual user input. 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude <noreply@anthropic.com>
Contributor
There was a problem hiding this comment.
Important
Looks good to me! 👍
Reviewed everything up to d6ba1eb in 1 minute and 1 seconds. Click for details.
- Reviewed
152lines of code in2files - Skipped
0files when reviewing. - Skipped posting
2draft comments. View those below. - Modify your settings and rules to customize what types of comments Ellipsis leaves. And don't forget to react with 👍 or 👎 to teach Ellipsis.
1. hld/session/manager.go:1030
- Draft comment:
Added propagation of ParentToolUseID to the conversation event (and bus event) for system, message, tool_call, tool_result, and thinking events. Ensure that this field is consistently set and handled across the event processing pipelines. - Reason this comment was not posted:
Comment did not seem useful. Confidence is useful =0%<= threshold50%This comment is asking the PR author to ensure that a field is consistently set and handled across the event processing pipelines. This falls under the category of asking the author to ensure something is done, which is against the rules.
2. humanlayer-wui/src/components/internal/SessionDetail/components/ActiveSession.tsx:308
- Draft comment:
IntroducedpendingApprovalsCountstate and updated the pending approvals visibility check. Verify that filtering by ApprovalStatus.Pending and subsequent UI message (singular/plural) works as expected. - Reason this comment was not posted:
Comment did not seem useful. Confidence is useful =0%<= threshold50%The comment is asking the PR author to verify that the filtering and UI message work as expected. This falls under the category of asking the author to ensure the behavior is intended or tested, which is against the rules.
Workflow ID: wflow_TXzSba3vbzurYeDp
You can customize by changing your verbosity settings, reacting with 👍 or 👎, replying to comments, or adding code review rules.
Contributor
|
nice LGTM |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
What problem(s) was I solving?
Sub-agent messages were appearing as regular user messages in the UI because only
tool_useevents captured theparent_tool_use_idfield from Claude stream events. This made it impossible to distinguish between actual user input and activity from sub-agents (agents launched via the Task tool).Additionally, when multiple approvals were pending, the status bar only showed "Pending Approval" without indicating how many approvals were waiting.
What user-facing changes did I ship?
parent_tool_use_idfor all conversation event types, enabling future UI work to distinguish sub-agent messages from user messagesHow I implemented it
Backend changes (hld/session/manager.go:1028-1269):
parent_tool_use_idcapture to all conversation event types inprocessStreamEvent():tool_useevents, causing sub-agent messages to lose their parent contextConversationEvent) and event bus publicationsFrontend changes (humanlayer-wui/src/components/internal/SessionDetail/components/ActiveSession.tsx:309-925):
pendingApprovalsCountstate to track total pending approvalscheckPendingApprovalVisibility()to count all pending approvals, not just check for existenceHow to verify it
make check test- All tests passed (412 unit tests, 46 integration tests)Manual Testing
Sub-agent message tracking:
parent_tool_use_idpopulated:Multiple approval count:
Description for the changelog
Fixed sub-agent message tracking by capturing
parent_tool_use_idfor all conversation event types. Added count display to pending approval status bar when multiple approvals are waiting.