1. apps/shared/OpenClawKit/Sources/OpenClawChatUI/ChatViewModel.swift, line 874-889 (link)

   Errored run skips history refresh when partial streaming content exists

   For the "error" state, appendFinalAssistantMessage falls through to streamedAssistantMessage() when chat.message is nil (typical for errored runs). If any partial streaming text was buffered, streamedAssistantMessage() creates a synthesized message and returns true. This causes appendFinalAssistantMessage to return true, which suppresses refreshHistoryAfterRun().

   The result: partial, incomplete content from a failed run is permanently appended to self.messages as if it were a finalized assistant message, and the client never reconciles with the server. On next reload the partial message would disappear (since the server history for an errored run won't include it), creating a visible inconsistency.

   The "final" and "aborted" cases are intentional here — those runs legitimately complete. "error" is different: the server hasn't produced a final message, so skipping the history reconciliation is incorrect.

   Consider separating the error branch so it always triggers a history refresh, even when partial content exists:

   case "final", "aborted":
       let appendedFinalMessage = self.appendFinalAssistantMessage(from: chat)
       self.pendingToolCallsById = [:]
       self.streamingAssistantText = nil
       if !appendedFinalMessage {
           Task { await self.refreshHistoryAfterRun() }
       }
   case "error":
       self.errorText = chat.errorMessage ?? "Chat failed"
       if let runId = chat.runId {
           self.clearPendingRun(runId)
       } else if self.pendingRuns.count <= 1 {
           self.clearPendingRuns(reason: nil)
       }
       self.pendingToolCallsById = [:]
       self.streamingAssistantText = nil
       Task { await self.refreshHistoryAfterRun() }

   Prompt To Fix With AI

   This is a comment left during a code review.
   Path: apps/shared/OpenClawKit/Sources/OpenClawChatUI/ChatViewModel.swift
   Line: 874-889
   
   Comment:
   **Errored run skips history refresh when partial streaming content exists**
   
   For the `"error"` state, `appendFinalAssistantMessage` falls through to `streamedAssistantMessage()` when `chat.message` is nil (typical for errored runs). If any partial streaming text was buffered, `streamedAssistantMessage()` creates a synthesized message and returns `true`. This causes `appendFinalAssistantMessage` to return `true`, which suppresses `refreshHistoryAfterRun()`.
   
   The result: partial, incomplete content from a failed run is permanently appended to `self.messages` as if it were a finalized assistant message, and the client never reconciles with the server. On next reload the partial message would disappear (since the server history for an errored run won't include it), creating a visible inconsistency.
   
   The `"final"` and `"aborted"` cases are intentional here — those runs legitimately complete. `"error"` is different: the server hasn't produced a final message, so skipping the history reconciliation is incorrect.
   
   Consider separating the error branch so it always triggers a history refresh, even when partial content exists:
   ```swift
   case "final", "aborted":
       let appendedFinalMessage = self.appendFinalAssistantMessage(from: chat)
       self.pendingToolCallsById = [:]
       self.streamingAssistantText = nil
       if !appendedFinalMessage {
           Task { await self.refreshHistoryAfterRun() }
       }
   case "error":
       self.errorText = chat.errorMessage ?? "Chat failed"
       if let runId = chat.runId {
           self.clearPendingRun(runId)
       } else if self.pendingRuns.count <= 1 {
           self.clearPendingRuns(reason: nil)
       }
       self.pendingToolCallsById = [:]
       self.streamingAssistantText = nil
       Task { await self.refreshHistoryAfterRun() }
   ```
   
   How can I resolve this? If you propose a fix, please make it concise.

2. apps/shared/OpenClawKit/Sources/OpenClawChatUI/ChatViewModel.swift, line 895-908 (link)

   Hardcoded alias pair may not generalize to other session configurations

   matchesCurrentSessionKey only handles one specific canonical/alias pair: "agent:main:main" ↔ "main". Any other gateway session key following a similar convention (e.g., "agent:work:work" ↔ "work", or "agent:staging:staging" ↔ "staging") would fail the match and silently drop events.

   If the alias pattern is always agent:<name>:<name> ↔ <name>, a general rule would be more resilient:

   // Generalize: "agent:<name>:<name>" is the canonical form of alias "<name>"
   let parts = incomingNormalized.split(separator: ":", maxSplits: 2)
   if parts.count == 3 && parts[0] == "agent" && parts[1] == parts[2] && parts[2] == currentNormalized {
       return true
   }
   let currentParts = currentNormalized.split(separator: ":", maxSplits: 2)
   if currentParts.count == 3 && currentParts[0] == "agent" && currentParts[1] == currentParts[2] && currentParts[2] == incomingNormalized {
       return true
   }

   If "main" truly is the only valid alias in production today, a comment to that effect would help future maintainers understand the constraint.

   Prompt To Fix With AI

   This is a comment left during a code review.
   Path: apps/shared/OpenClawKit/Sources/OpenClawChatUI/ChatViewModel.swift
   Line: 895-908
   
   Comment:
   **Hardcoded alias pair may not generalize to other session configurations**
   
   `matchesCurrentSessionKey` only handles one specific canonical/alias pair: `"agent:main:main"` ↔ `"main"`. Any other gateway session key following a similar convention (e.g., `"agent:work:work"` ↔ `"work"`, or `"agent:staging:staging"` ↔ `"staging"`) would fail the match and silently drop events.
   
   If the alias pattern is always `agent:<name>:<name>` ↔ `<name>`, a general rule would be more resilient:
   ```swift
   // Generalize: "agent:<name>:<name>" is the canonical form of alias "<name>"
   let parts = incomingNormalized.split(separator: ":", maxSplits: 2)
   if parts.count == 3 && parts[0] == "agent" && parts[1] == parts[2] && parts[2] == currentNormalized {
       return true
   }
   let currentParts = currentNormalized.split(separator: ":", maxSplits: 2)
   if currentParts.count == 3 && currentParts[0] == "agent" && currentParts[1] == currentParts[2] && currentParts[2] == incomingNormalized {
       return true
   }
   ```
   If `"main"` truly is the only valid alias in production today, a comment to that effect would help future maintainers understand the constraint.
   
   How can I resolve this? If you propose a fix, please make it concise.

Best possible solution:

Land the narrowed Chat UI fix after adding a short simulator or device streaming proof that shows visible streamed text replacing the Writing state without flicker.

Do we have a high-confidence way to reproduce the issue?

Source-reproducible: current main gives assistant text segments fresh UUIDs on each parse and renders the typing indicator solely from blocking run activity, so the reported visual instability path is clear from source. I did not run a live iOS streaming session.

Is this the best way to solve the issue?

Yes, this is the best narrow fix I found: it changes parser identity and duplicate-state rendering at the SwiftUI surface where the churn originates, while leaving current-main run-state and scroll ownership untouched.

AGENTS.md: found and applied where relevant.

Codex review notes: model internal, reasoning high; reviewed against c067802cda3a.

Label changes:

• add rating: 🦪 silver shellfish: Overall readiness is 🦪 silver shellfish; proof is 🦪 silver shellfish and patch quality is 🐚 platinum hermit.
• add status: 📣 needs proof: The PR needs real behavior proof before ClawSweeper can clear the contributor ask. Needs real behavior proof before merge: The PR has exact-head CI/Testbox build and parser test proof, but no after-fix real chat streaming recording, logs, or transcript showing the visual behavior; add proof with private details redacted, and updating the PR body should trigger a fresh ClawSweeper review or a maintainer can comment @clawsweeper re-review. After adding proof, update the PR body; ClawSweeper should re-review automatically. If it does not, the PR author or someone with repository write access can comment @clawsweeper re-review.

Label justifications:

• P2: This is a normal-priority iOS/macOS chat UI stability bugfix with limited blast radius and no evidence of data loss, security impact, or core runtime outage.
• rating: 🦪 silver shellfish: Overall readiness is 🦪 silver shellfish; proof is 🦪 silver shellfish and patch quality is 🐚 platinum hermit.
• status: 📣 needs proof: The PR needs real behavior proof before ClawSweeper can clear the contributor ask. Needs real behavior proof before merge: The PR has exact-head CI/Testbox build and parser test proof, but no after-fix real chat streaming recording, logs, or transcript showing the visual behavior; add proof with private details redacted, and updating the PR body should trigger a fresh ClawSweeper review or a maintainer can comment @clawsweeper re-review. After adding proof, update the PR body; ClawSweeper should re-review automatically. If it does not, the PR author or someone with repository write access can comment @clawsweeper re-review.

PR surface:

Other +10. Total +10 across 4 files.

What I checked:

• Repository policy read: Root AGENTS.md was read fully and applied as OpenClaw review policy, including SwiftUI, release-owned changelog, and real-proof guidance. (AGENTS.md:1, c067802cda3a)
• Current main segment identity: Current main constructs AssistantTextSegment with a new UUID, so repeated parses of unchanged completed assistant text produce new identities for SwiftUI ForEach consumers. (apps/shared/OpenClawKit/Sources/OpenClawChatUI/AssistantTextParser.swift:9, c067802cda3a)
• PR head segment identity: The PR head replaces the UUID with an Int and appends segments with the current positional count, giving repeated parses deterministic IDs for unchanged segment positions. (apps/shared/OpenClawKit/Sources/OpenClawChatUI/AssistantTextParser.swift:9, 72e1ab863487)
• Current main duplicate visible state: Current main renders ChatTypingIndicatorBubble whenever hasBlockingRunActivity is true, then independently renders ChatStreamingAssistantBubble when visible streaming text exists. (apps/shared/OpenClawKit/Sources/OpenClawChatUI/ChatView.swift:373, c067802cda3a)
• PR head duplicate-state fix: The PR head suppresses the typing indicator when hasVisibleStreamingAssistantText is true and reuses that helper for stream bubble, transient-content, and empty-state decisions. (apps/shared/OpenClawKit/Sources/OpenClawChatUI/ChatView.swift:373, 72e1ab863487)
• Parser regression coverage: The PR adds a focused test proving repeated parses return the same deterministic IDs for thinking and response segments. (apps/shared/OpenClawKit/Tests/OpenClawKitTests/AssistantTextParserTests.swift:52, 72e1ab863487)

Likely related people:

• mahmoud: Current-main blame attributes the relevant OpenClawChatUI parser and rendering files in this checkout to commit 32510a3. (role: recent area contributor; confidence: medium; commits: 32510a3225b8; files: apps/shared/OpenClawKit/Sources/OpenClawChatUI/AssistantTextParser.swift, apps/shared/OpenClawKit/Sources/OpenClawChatUI/ChatView.swift, apps/shared/OpenClawKit/Sources/OpenClawChatUI/ChatMessageViews.swift)
• Peter Steinberger: History shows prior OpenClawKit Chat UI work in d15b6af and the current PR head was rewritten by this reviewer with contributor credit preserved. (role: feature-history contributor and current reviewer; confidence: high; commits: d15b6af77b49, e01f23abd50c, 72e1ab863487; files: apps/shared/OpenClawKit/Sources/OpenClawChatUI/AssistantTextParser.swift, apps/shared/OpenClawKit/Sources/OpenClawChatUI/ChatView.swift, apps/shared/OpenClawKit/Sources/OpenClawChatUI/ChatMessageViews.swift)
• Mariano: Git history for ChatViewModel shows multiple nearby iOS chat streaming and session-key fixes in February 2026, which are adjacent to the older session-routing parts intentionally removed from this PR. (role: adjacent stream/session-state contributor; confidence: medium; commits: 42d11a3ec5f4, fe3f0759b5c4, 6effcdb551a8; files: apps/shared/OpenClawKit/Sources/OpenClawChatUI/ChatViewModel.swift)

• 🦀 challenger crab: rare, exceptional readiness with strong proof, clean implementation, and convincing validation.
• 🦞 diamond lobster: very strong readiness with only minor maintainer review expected.
• 🐚 platinum hermit: good normal PR, likely mergeable with ordinary maintainer review.
• 🦐 gold shrimp: useful signal, but proof or patch confidence is still limited.
• 🦪 silver shellfish: thin signal; proof, validation, or implementation needs work.
• 🧂 unranked krab: not merge-ready because proof is missing/unusable or there are serious correctness or safety concerns.
• 🌊 off-meta tidepool: rating does not apply to this item.

Shiny media proof means a screenshot, video, or linked artifact directly shows the changed behavior. Runtime, network, CSP, and security claims still need visible diagnostics.

• ClawSweeper keeps one durable marker-backed review comment per issue or PR.
• Re-runs edit this comment so the latest verdict, findings, and automation markers stay together instead of adding duplicate bot comments.
• A fresh review can be triggered by eligible @clawsweeper re-review comments, exact-item GitHub events, scheduled/background review runs, or manual workflow dispatch.
• PR/issue authors and users with repository write access can comment @clawsweeper re-review or @clawsweeper re-run on an open PR or issue to request a fresh review only.
• Maintainers can also comment @clawsweeper review to request a fresh review only.
• Fresh-review commands do not start repair, autofix, rebase, CI repair, or automerge.
• Maintainer-only repair and merge flows require explicit commands such as @clawsweeper autofix, @clawsweeper automerge, @clawsweeper fix ci, or @clawsweeper address review.
• Maintainers can comment @clawsweeper explain to ask for more context, or @clawsweeper stop to stop active automation.

• reviewed 2026-07-04T01:04:06.272Z sha 2896ccf :: needs real behavior proof before merge. :: [P1] Pass agentId through agent event routing | [P1] Update the agent event schema with routing fields | [P3] Remove the release-owned changelog entry
• reviewed 2026-07-05T05:01:11.280Z sha 2896ccf :: needs real behavior proof before merge. :: [P1] Carry agentId through session-key agent routing | [P1] Rebase away the stale raw-agent session flow | [P3] Remove the release-owned changelog entry
• reviewed 2026-07-05T07:35:57.648Z sha 2896ccf :: needs real behavior proof before merge. :: [P1] Carry agentId through session-key agent routing | [P1] Rebase away the stale raw-agent session flow | [P3] Remove the release-owned changelog entry
• reviewed 2026-07-05T15:28:02.977Z sha 2896ccf :: needs real behavior proof before merge. :: [P1] Carry agentId through session-key agent routing | [P1] Rebase raw-agent routing onto current main | [P3] Remove the release-owned changelog entry
• reviewed 2026-07-05T17:30:22.564Z sha 2896ccf :: needs real behavior proof before merge. :: [P1] Carry agentId through agent-event routing | [P1] Rebase the session-stream filter onto current main | [P3] Remove the release-owned changelog entry
• reviewed 2026-07-05T17:53:20.743Z sha 2896ccf :: needs real behavior proof before merge. :: [P1] Carry agentId through agent stream matching | [P3] Remove the release-owned changelog entry
• reviewed 2026-07-09T06:17:01.498Z sha 2896ccf :: needs real behavior proof before merge. :: [P1] Carry agentId through agent stream matching | [P3] Remove the release-owned changelog entry

• The egg game starts only after the PR passes the real-behavior proof check.
• Before that, no creature or rarity is rolled. The treat waits for real proof.
• This is still just collectible flavor: proof affects review readiness, not creature quality.