fix: agent streaming - #2880
Open
tevanburen wants to merge 8 commits into
Open
Conversation
Contributor
✅ Snyk checks have passed. No issues have been found so far.
💻 Catch issues earlier using the plugins for VS Code, JetBrains IDEs, Visual Studio, and Eclipse. |
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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Description
Small set of fixes to the agent-run harness and subagent infrastructure, surfaced while wiring up Playground's subagent UI.
Changes Made
AgentRuntimeManager.getRun: always populatependingActions(empty by default) instead of only when the run isINPUT_REQUIRED, so consumers get a consistent field on every response.SemossAgentHarness/HarnessToolExecutor: reattach_meta(including execution mode) to subagent tool calls (spawn/named/check/wait) — previously they had none, since they're not backed by room/workspace MCP metadata — and stop suppressing their tool-item stream events, so they render like ordinary tool calls instead of being hidden from the client.SubAgentToolSynthesizer: explicitly declareSMSS_MCP_EXECUTION=autoon all subagent tool definitions. Without it, reattaching_metacausedMCPUtility.getValidMcpExecutionto default them toask, which incorrectly paused every spawn/check/wait call for user approval.ModelInferenceLogsUtils.getUserConversations: exclude rooms with a non-nullPARENT_ROOM_IDfromGetPlaygroundRooms— every subagent spawn creates a real Room row, and these were leaking into the playground room list as if they were user-initiated conversations.How to Test
GetPlaygroundMessages— theSpawnSubAgent/WaitForSubAgenttool call should carry_meta.SMSS_TOOL_KINDand_meta.SMSS_MCP_EXECUTION="auto".GetPlaygroundRooms— the subagent's own room should not appear in the list.GetAgentRunon a run that isn't paused —pendingActionsshould be an empty array, not missing.Notes
agent-streamingPR — Playground's subagent tool-card/stream rendering depends on the_meta/execution-mode fix here.