Agent streaming - #3534
Open
tevanburen wants to merge 36 commits into
Open
Conversation
✅ 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. |
tevanburen
force-pushed
the
agent-streaming
branch
from
August 11, 2026 14:42
f21bbac to
3684dd1
Compare
neelneelneel
requested changes
Aug 13, 2026
| */ | ||
| export const submitAgentRun = async ( | ||
| params: { | ||
| roomId: string; |
Contributor
There was a problem hiding this comment.
Can we comment each of the parameters / the return?
tevanburen
marked this pull request as ready for review
August 14, 2026 20:35
Contributor
|
Yeah skill file for this @tevanburen |
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
4 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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Description
Adds a full agent mode (RunAgent) to Playground — a server-driven agent harness loop that streams messages, tool calls, and subagent activity back to the client, as an alternative to the existing client-orchestrated chat flow. Requires the companion Semoss backend
agent-streamingbranch to function.Changes Made
SDK (
libs/sdk)agent.ts/agent.types.ts/agent-subscription.ts: pure API wrappers forRunAgent,GetAgentRun,GetSubagentRuns, and tool-decision pixels, plus asubscribeAgentRunpolling/reconciliation subscription and shared item-event merging — decoupling the wire protocol from Playground-specific state.Agent mode entry & UI (
room-input.tsx,new-room-page.tsx,constants.ts, slash-command context)/agent-harness(and hidden/harness) slash commands to switch a room into agent mode.Streaming & tool execution (
agent-harness.ts,response-message.store.ts,response-message-tool*.tsx,tools-view.tsx)agent-harness.tsstore: submits agent runs, applies streamed item events (message/reasoning/tool/subagent) onto the room's message tree, reconnects an in-progress run after a page reload, and resumes a paused run on a tool decision.ToolsViewfor tools with no resolvable MCP app behind them.Subagents
AGENT_RUNtable on reload — not just the tail turn's — positioned under the message that spawned them.Cancellation
How to Test
/agent-harnessto switch into agent mode.Notes
agent-streamingbranch — subagent_meta/execution-mode fixes, the playground room-list filter excluding subagent rooms, and theGetSubagentRuns/StopAgentRunreactors all live there.SocketClient.executeCommand, so a working Stop button is follow-up work, tracked separately.