Repository navigation
[backport cloud/1.54] Agent fix gap sweep: 8 merged main fixes without a cloud carrier - #18254
christian-byrne wants to merge 10 commits into
Conversation
…t attachments (#18104) <!-- ccr-slack-attribution --> _Requested by **Christian Byrne** · [Slack thread](https://comfy-organization.slack.com/archives/C0BGH348Z0C/p1789446697386139?thread_ts=1789446697.386139&cid=C0BGH348Z0C)_ ## Summary Replaces #18067 (originally a repro-only tests PR; same base branch could not be rewritten to drop an AI attribution trailer, see the comment there). This now includes the real fix plus the flipped tests. Fixes: dragging a non-first output of a multi-output/batch job (the "N layers" stack expansion in the asset sidebar) into the agent chat composer showed the wrong image once the message was sent. ## Root cause `UserMessage.vue`'s `splitAttachments` (`src/workbench/extensions/agent/components/agent/message/UserMessage.vue`) rebuilt the attachment image URL as `/view?filename={item.ref}&type=input` whenever `item.ref` was set, discarding the already-correct `item.previewUrl`. That's right for a genuine file upload, but the asset-drag-to-attach path (`AgentPanelRoot.vue`'s `attachDroppedAsset`) reuses the same `ref` field for an *existing output asset*'s identifier. A job's primary/representative output has a real content hash and happens to round-trip through `/view?...&type=input`. A non-first output reached via the "layers" stack expansion is synthesized client-side (`outputAssetUtil.ts`'s `mapOutputsToAssetItems`) and, by design, never gets a `.hash` — so its `ref` is a bare output filename, which `/view?type=input` does not resolve to the dragged file. Canvas drop and the job's first/primary output were unaffected, matching the original report. ## Fix Prefer `item.previewUrl` when present, and only fall back to the `/view?...&type=input` reconstruction when there's no `previewUrl`. This matches how `AttachmentChip.vue` already treats the same attachment shape in the draft composer — it renders `previewUrl` directly and never reconstructs a URL from `ref`. The fix is a single, isolated change: `item.ref` is used nowhere else for URL construction (its other uses are for uploading and sending the attachment reference to the backend), so no other call sites needed touching. ## Changes - `UserMessage.vue`: `splitAttachments` now prefers `previewUrl` over the `ref`-based reconstruction. - `UserMessage.test.ts`: the regression test that pinned this bug is flipped from `it.fails(...)` to a plain `it(...)` — it now passes. - `browser_tests/tests/agent/agentBatchOutputAttachment.spec.ts`: the `test.fail()` call is removed so the e2e repro is now a normal, expected-to-pass test. ## Test Plan - [x] `pnpm typecheck` passes - [x] `pnpm lint` (oxlint + eslint) passes on the changed files - [x] `pnpm format:check` (oxfmt) passes on the changed files - [x] Vitest: `UserMessage.test.ts` — 17/17 passed (previously 16 passed + 1 expected fail) - [ ] Playwright e2e — not executable in this sandbox (no backend/dev server); typechecks and lints cleanly, follows existing agent-panel spec conventions 🤖 Generated with [Claude Code](https://claude.com/claude-code) https://claude.ai/code/session_01ADPh3Nvwr5WhoC9HPVm2Ye --------- Co-authored-by: test-bot <test-bot@users.noreply.github.com> Co-authored-by: bymyself <cbyrne@comfy.org> (cherry picked from commit c61d9df)
…tay on their slots through undo/redo (#18110) <!-- ccr-slack-attribution --> _Requested by **Christian Byrne** · [Slack thread](https://comfy-organization.slack.com/archives/C0BGH348Z0C/p1789592320785689?thread_ts=1789592320.785689&cid=C0BGH348Z0C)_ ## ELI5 A Vue node measures where each of its slot dots sits and stores the measurements so the canvas knows where to draw the wire end. Undo and redo rebuild the graph from a snapshot, and that rebuild throws all those measurements away, but the node on screen never changes and never re-measures. With no measurement the canvas falls back to its own guess of where the widget row is, and that guess can be a full row off, so the wire visibly lands on the wrong slot. This PR tells mounted Vue nodes to measure again whenever their stored measurements are dropped. ## Motivation Before: wiring a node into a Vue-rendered widget slot, then pressing undo and redo, could leave the wire end drawn on a different row than the slot dot it is connected to (38 px off on a promoted subgraph host with a multiline text widget above the slot, 10 px off on a plain EmptyLatentImage's third widget). The link stayed correct in the graph, only its rendering was wrong, and it stayed wrong until something else resized the node. After: the wire end stays on the slot dot through undo and redo, for subgraph hosts and plain nodes alike. Why: the graph reload behind undo and redo purges every DOM-measured slot offset from the layout store, and nothing re-measured a Vue node that survived the reload with identical rows and size. ## Provenance - **Authored by:** interactive session - **Verified:** Playwright `subgraphPromotedWidgetExternalLink.spec.ts` drift rows on the test-only commit: 2 failed on the drift assertion (host 38 px, plain batch_size 10 px); with the fix: both pass and the rest of the spec keeps its expected outcomes. Vitest `src/renderer/core/layout` + `src/renderer/extensions/vueNodes/components`: 33 files, 262 passed (the new NodeSlots case fails pre-fix with `expected null to deeply equal { x: 0, y: 20 }`). `vue-tsc --noEmit` and `vue-tsc --project browser_tests/tsconfig.json`: clean. eslint, oxlint, oxfmt on the changed files: clean. `knip --cache`: clean. `changeTracker.spec.ts` fails identically in the local sandbox with and without the fix (legacy-canvas collapse clicks and default-workflow node locators), so those failures are environmental there; CI is the arbiter. - **Deviations:** none ## Reviewer context - **Type:** bug fix, Vue node link rendering after undo/redo - **Slots into:** #18085 — test: add failing repros for promoted widget external-link rendering. This PR ports that spec and supersedes its `test.fail()` drift case (`keeps the wire on the promoted width slot dot after undo and redo`), which becomes a real table-driven regression test here with a plain-node row added; #18085 should drop that case once this lands. The label and textarea cases no longer carry `test.fail()`: the width-label cases assert the label stays visible after wiring, and the promoted text case asserts the textarea is enabled before wiring and visible but disabled afterward, with its input slot dot shown. ## How - `layoutStore` gains a reactive `slotOffsetVersion` that is bumped only when measured slot offsets are dropped in bulk (`clearGraph`, `clearViewGeometry`), never on a measurement write, so re-measuring cannot feed back into itself. - `NodeSlots.vue` adds that version to its existing `syncSlotOffsets` watch. Every mounted Vue node renders one `NodeSlots`, and `syncSlotOffsets` measures the whole `[data-node-id]` element, so one trigger per node re-measures every slot (inputs, outputs and widget rows). Purging on clear stays as is; keeping stale offsets would be wrong when the reload does change rows. ## Red-green verification | Commit | CI | Purpose | |---|---|---| | `test: cover the wire drifting off a Vue widget slot after undo and redo` | red | Playwright drift rows (host + plain node) and the NodeSlots re-measure case fail | | `fix: re-measure Vue node slot offsets after the layout store drops them` | green | same tests pass | ## Test plan - [ ] CI red on the test-only commit, green on the fix commit - [ ] Load a workflow with a Vue node whose widget rows differ in height from litegraph's layout (a multiline text widget above a number widget), wire a primitive into the number widget, click the canvas, Ctrl+Z, click elsewhere, Ctrl+Y: the wire end stays on the slot dot - [ ] Enter and leave a subgraph, then check links to widget slots still land on their dots --- _Generated by [Claude Code](https://claude.ai/code/session_01JwD8PkozfznCoUwAG46WAg)_ --------- Co-authored-by: Claude <noreply@anthropic.com> Co-authored-by: bymyself <cbyrne@comfy.org> (cherry picked from commit 38cc119)
…#18117) <!-- ccr-slack-attribution --> _Requested by **Christian Byrne** · [Slack thread](https://comfy-organization.slack.com/archives/C0BGH348Z0C/p1789592320785689?thread_ts=1789592320.785689&cid=C0BGH348Z0C)_ ## ELI5 Dragging a wire out of a slot on a Vue node and dropping it on another slot connects the two nodes, but the undo history never hears about it. Press Ctrl+Z straight after the drop and nothing happens; you first have to click somewhere on the canvas, and that click is what finally writes the undo entry. This PR writes the entry the moment the drop finishes, the same way the classic canvas already does. ## Motivation Before: connect two Vue node slots by dragging, press Ctrl+Z: the link stays. Only a later canvas click, key press or widget edit records the connection, so undo looks like it needs two steps or does nothing. After: Ctrl+Z right after the drop removes the link. Why: the change tracker records canvas edits from the window `mouseup` event and from `LGraphCanvas.processMouseUp`. The Vue slot drag calls `preventDefault()` on `pointerdown`, which suppresses the compatibility `mouseup`, and the drop is resolved entirely inside the Vue composable, so neither hook runs and `captureCanvasState()` is never called. ## Provenance - **Authored by:** interactive session - **Verified:** Playwright `linkInteraction.spec.ts`, new case `undo right after dropping a link on a slot removes that link`: fails on the test-only commit (`expected 0, received 1` links after undo) and passes with the fix. The rest of the spec has the same outcome with and without the fix (9 cases fail in the local sandbox either way: 8 `toHaveScreenshot` baselines and one minimap overlay intercepting the subgraph button; a control run with the fix stashed reproduced all 9). A local probe confirmed the undo queue holds exactly one entry right after the drop. Root and `browser_tests` typecheck, full `pnpm lint`, `pnpm format:check`, eslint, oxlint and oxfmt on the changed files: clean. - **Deviations:** none ## Reviewer context - **Type:** bug fix, undo/redo for Vue node link drags - **Slots into:** #18085 (repro and isolation PR). The "undo needs an extra click" symptom surfaced while isolating the link drift there. The spurious undo entry after a redo is a separate tracker bug, fixed in #18118. - No unit test: the composable's pointer flow has no seam short of the ~200-line module-mock harness in `useSlotLinkInteraction.autoPan.test.ts`, so the e2e is the lowest level that proves the wiring. ## How `useSlotLinkInteraction.finishInteraction()` calls `captureCanvasState()` on the active workflow's change tracker after the drop is resolved and the canvas marked dirty, mirroring the `LGraphCanvas.processMouseUp` wrapper the tracker installs for the legacy canvas. `captureCanvasState()` compares against the last snapshot, so a drop that connects nothing pushes no entry, a second capture from any other hook is a no-op, and the legacy path is untouched. ## Red-green verification | Commit | CI | Purpose | |---|---|---| | `test: cover undo right after a Vue node slot drag-connect` (d8f3af7) | red | the new Playwright case fails: the link survives Ctrl+Z (e2e-status failed on shard 14 and the new-tests video job, everything else green) | | `fix: capture the canvas state when a Vue node slot link drag finishes` (2f9f6df) | green | same case passes, all checks green | ## Test plan - [x] CI red on the test-only commit (e2e-status failed on shard 14 and the new-tests video job) - [x] CI green on the fix commit - [ ] Drag a link between two Vue node slots, press Ctrl+Z immediately: the link is removed - [ ] Drag a link out of a slot and drop it on empty canvas, dismiss the search box: the undo queue does not grow --- _Generated by [Claude Code](https://claude.ai/code/session_01JwD8PkozfznCoUwAG46WAg)_ --------- Co-authored-by: Claude <noreply@anthropic.com> (cherry picked from commit d08de57)
#18118) <!-- ccr-slack-attribution --> _Requested by **Christian Byrne** · [Slack thread](https://comfy-organization.slack.com/archives/C0BGH348Z0C/p1789592320785689?thread_ts=1789592320.785689&cid=C0BGH348Z0C)_ ## ELI5 Redo rebuilds the graph from a saved snapshot. While rebuilding, the graph works out afresh the order in which nodes should run and stamps each node with a new number. The undo history then compares the live graph with its snapshot, sees those numbers differ, and concludes you edited something, so the next plain click on the canvas records a phantom "change". Pressing Ctrl+Z after that restores a snapshot that looks identical, so undo appears to do nothing and you have to press it twice. This PR makes the comparison ignore that run-order number, which the user never edits. ## Motivation Before: connect two nodes, undo, redo, click the canvas, press Ctrl+Z: the link stays. The click after the redo pushed an undo entry that differs from the live graph only in node `order`, so the first Ctrl+Z is a visible no-op and the redo queue is wiped. After: the click after a redo records nothing, and Ctrl+Z removes the redone link. Why: `ChangeTracker.graphEqual` deep-compares node objects, including `order`. `order` is the execution index that `LGraph.updateExecutionOrder()` recomputes on every configure, add or remove; a snapshot captured before a reload and the graph rebuilt from it legitimately disagree on it. The auto-queue comparison in the same file already lists `order` among the non-execution node properties it ignores. ## Provenance - **Authored by:** interactive session - **Verified:** Vitest `changeTracker.test.ts`, new case `does not push when only the recomputed node execution order differs`: fails on the test-only commit (`expected [ Array(1) ] to have a length of +0 but got 1`), and the whole file passes with the fix (63/63). Playwright `changeTracker.spec.ts`, new case `Undo after a redo and a canvas click removes the redone link`: fails on the test-only commit at its final assertion (`expected 0, received 1` links) with every earlier step passing, and passes with the fix. The other 6 failing cases in that spec fail identically in the local sandbox with the fix stashed (legacy-menu and subgraph-entry environment issues) and are green in CI. Full `pnpm lint`, `pnpm format:check`, typecheck, eslint, oxlint and oxfmt on the changed files: clean. - **Deviations:** none ## Reviewer context - **Type:** bug fix, undo/redo change detection - **Slots into:** #18085 (repro and isolation PR). The phantom entry was found while isolating the link drift there. The missing undo entry for a Vue node slot drag-connect is a separate bug, fixed in #18117. - The e2e clicks two different empty spots on purpose: two same-spot clicks inside the double-click window open node search, which swallows Ctrl+Z and Ctrl+Y. ## How `graphEqual` compares the two node lists with `order` omitted from every node. Nothing else in the comparison changes. ## Red-green verification | Commit | CI | Purpose | |---|---|---| | `test: cover the spurious undo entry pushed by a canvas click after redo` (e7539e0) | red | the unit case pushes an entry for an order-only difference (unit `test` job failed); the Playwright case leaves the link in place after the post-redo Ctrl+Z (e2e-status failed on shard 2 and the new-tests video job) | | `fix: ignore recomputed node execution order when comparing graph snapshots` (c2cffd8) | green | same cases pass, all checks green | ## Test plan - [x] CI red on the test-only commit (unit `test` job and e2e-status failed, shard 2 and the new-tests video job) - [x] CI green on the fix commit - [ ] Connect two nodes, Ctrl+Z, Ctrl+Y, click empty canvas, Ctrl+Z: the link is removed on the first press - [ ] Move a node, click the canvas, Ctrl+Z: the move is still undone (a real change is still detected) --- _Generated by [Claude Code](https://claude.ai/code/session_01JwD8PkozfznCoUwAG46WAg)_ --------- Co-authored-by: Claude <noreply@anthropic.com> (cherry picked from commit 5347eda)
<!-- ccr-slack-attribution --> _Requested by **Jo Zhang** · [Slack thread](https://comfy-organization.slack.com/archives/C0BGH348Z0C/p1789788571314109?thread_ts=1789788571.314109&cid=C0BGH348Z0C)_ ## Before `showSubscriptionRequiredDialog` — the seam every "Subscribe" CTA routes through, including the in-app agent panel's paywall card (`AgentPaywallCard.vue` → `AgentPanelRoot.vue`'s `onPaywallAction` → `useAccountPreconditionDialog`) — returned early whenever `window.__CONFIG__?.subscription_required` read falsy: ```ts if (!isCloud || !window.__CONFIG__?.subscription_required) { return } ``` `window.__CONFIG__` is only populated once the bootstrap `/features` fetch resolves (`refreshRemoteConfig()`). A caller that fires before that fetch resolves — plausible right after a fresh page load, e.g. a user landing on a cloud surface and immediately triggering the agent paywall — hit a **completely silent no-op**: no dialog, no toast, no telemetry, no retry. The click just did nothing. ## After - If the config hasn't loaded yet (`remoteConfigState` is `'unloaded'`), await one fresh `/features` fetch before deciding. - If the gate is still closed afterward, report it via `reportError()` instead of returning silently, so this is visible in telemetry instead of only surfacing as a user bug report. ## Test plan - `src/services/dialogService.subscriptionRequired.test.ts` (unit, deterministic): covers the non-cloud no-op, the already-loaded happy path, the retry-on-unloaded-config path (the exact timing race this fix targets, via a mocked `remoteConfigState`), and the reported-failure path. Confirmed red against the pre-fix code, green after. - `browser_tests/tests/agent/agentPaywallSubscribe.spec.ts` (new, e2e): drives the real chat panel end to end through the agent-integration fixtures in `browser_tests/tests/agent/` (`agentPanelMocks.ts`'s `agentTest`/`agentPanel`). A mocked `no_funds` turn rejection surfaces the real `AgentPaywallCard`, and clicking its real "Subscribe" button is asserted to: - open the real subscription dialog (`[aria-labelledby="subscription-required"]`) when the config gate is open — with a screenshot of the paywall card and the opened dialog attached as test evidence for visual proof; - otherwise report the failure via the new `reportError()` call (asserted via a captured `console.error` containing `error_opening_subscription_dialog_gate_closed`) instead of silently doing nothing, when the gate stays durably closed. - Note: the literal race this fix guards against (`remoteConfigState === 'unloaded'` at click time) is only reachable in this app while the whole UI — including the agent panel itself, which is gated behind the same `/features`-driven PostHog flag — has not yet booted, so it cannot be driven from a live page through the DOM; that exact branch is what the unit test covers deterministically. The e2e spec instead proves the two real, reachable end states (`subscription_required` true vs. durably false) drive the correct real-browser outcome for this exact regression. - `pnpm exec vitest run src/services/dialogService` — 41/41 passing. - `pnpm exec eslint`, `pnpm exec oxlint --type-aware`, `pnpm exec oxfmt --check` on the new spec — clean. - `pnpm exec vue-tsc --noEmit` — clean (full project, includes `browser_tests`). - The new Playwright spec itself was not run in this sandbox (no browser/backend available here); it was verified via lint/format/typecheck and by tracing every fixture and route it depends on against the existing, passing `browser_tests/tests/agent/agentPanel.spec.ts` for identical patterns (workflow selection, composer/send locators, `agentPanelMocks.ts` route overrides, `subscription.spec.ts`'s remock-then-reload pattern, and `consoleErrorCollector.ts`). 🤖 Generated with [Claude Code](https://claude.com/claude-code) https://claude.ai/code/session_01JbnJ4RxoJA8aRva1EA7g5K --- _Generated by [Claude Code](https://claude.ai/code/session_01JbnJ4RxoJA8aRva1EA7g5K)_ --------- Co-authored-by: Claude <noreply@anthropic.com> (cherry picked from commit 1fc4bdb)
## Summary Restore the spare autogrow input after Agent reconciliation so dynamic-node wires remain aligned with their user-facing widgets. ## Changes - **What**: Re-run autogrow reconciliation for live Agent-owned nodes and add a regression test that reproduces the omitted spare input. ## Review Focus The repair uses the existing autogrow growth path and preserves the connected input and link while restoring only the missing spare slot. Fixes [PM-1155](https://linear.app/comfyorg/issue/PM-1155/fix-autogrow-widget-wiring-displays-on-the-wrong-widget-despite-a) ## Screenshots (if applicable) Manually verified with the real OpenAI GPT Image 2 node: the Load Image wire lands on `image_1`, `image_2` remains available, and the node retains its normal labels and controls. ## Test Plan - `pnpm test:unit src/workbench/extensions/agent/crdt/agentNodeMaterializer.test.ts` - `pnpm test:unit src/core/graph/widgets/dynamicWidgets.test.ts` - `pnpm typecheck` - targeted oxfmt, oxlint, and ESLint checks --------- Co-authored-by: Amp <amp@ampcode.com> (cherry picked from commit 3c20214)
…18191) <!-- ccr-slack-attribution --> _Requested by **Jo Zhang** · [Slack thread](https://comfy-organization.slack.com/archives/C0BGH348Z0C/p1789890737989329?thread_ts=1789890737.989329&cid=C0BGH348Z0C)_ Three commits, red then green, so CI proves the test catches the bug: - `5203929` — the failing repro alone (red) - `b57144d` — the minimal fix (green) - `0c9010e` — coverage for the mint-side branch Rebased onto `81fc722`. The merge queue caught a semantic conflict that neither this branch nor main showed alone: `GraphMutationsDeps` gained a required `placement` port while this was open, so the new test's `createGraphMutations` call no longer compiled once merged. Fixed in the commit that introduces the file, so every commit still typechecks. ## Before A node added through the v2 search box with FollowCursor on renders at 60% opacity with `pointer-events-none` forever. It cannot be clicked, selected, moved, or deleted. A browser reload does not clear it, because the state lives in the document rather than in the tab. ## After Placing a node leaves it fully opaque and interactive, and it stays that way across any number of follower reconciles. Documents already carrying the flag are healed on the next read. ## What was happening `flags.ghost` is transient placement state: "semi-transparent, following cursor" (`src/lib/litegraph/src/interfaces.ts:351`). It should live from the search-box add until the click that drops the node. It leaked into the CRDT document instead: 1. `LGraph.add` sets `node.flags.ghost = true` (`LGraph.ts:1382`) **before** `attachNodeLayout` (`:1404`), and only starts placement afterwards (`:1411`). 2. `attachNodeLayout` fires the `createNode` layout change that `layoutMintPort` turns into `add_node`, whose payload is `serializeNode(...)` — taken while placement is still live, so it carries `flags: { ghost: true }`. 3. Clicking to place runs `finalizeGhostPlacement(false)`, which does `delete node.flags.ghost` and fires `node:property:changed` (`LGraphCanvas.ts:3778-3784`). **Nothing in the CRDT layer listens to that event**, and the three mint ports only cover createNode/deleteNode/clearGraph, links and widgets. No wire op carries the clear, so the document kept `ghost: true` permanently. 4. `flags` is in `RESYNCED_NODE_FIELDS` (`ecsFollowerAdapter.ts:38`). Any full reconcile re-reads every node, `prepareNode` cloned the document's flags verbatim, and `assignNodeFields` replaced `state.flags` wholesale (`nodeDataStore.ts`). 5. `LGraphNode.vue` then rendered it dead: `globalOpacity * 0.6` (`:375`) and `pointer-events-none` (`:24-27`). Neither depends on `isGhostPlacing`, so with placement long finished nothing would ever call `finalizeGhostPlacement` again. The trigger is any full reconcile — a resubscribe after a doc gap, a tab switch, or a rejected batch re-arming. The window is the gap between the clear and the next reconcile, which is why only some placed nodes showed it while the rest sat on the same latent `ghost: true`. ## The fix The document is authoritative data; `ghost` is presentation logic, so it should never have been in there (ADR-CRDT-LAYOUT-0003). Stripped at both boundaries: - **`graphMutations.ts`** — `prepareNode` drops `ghost` from the flags it builds, and from `lastSerialization` (which feeds `serialize()` for unregistered node types, so the flag cannot round-trip into a saved workflow either). A reconcile can no longer resurrect it, and documents already carrying it are healed. - **`layoutMintPort.ts`** — `add_node` omits `ghost` from its mid-placement snapshot, so new documents never record it. Read-side alone would leave every document accumulating the flag; mint-side alone would not help anyone whose document already has it. Both are needed. The local placement preview is untouched: `LGraph.add` sets the flag directly on the node's own state, never through `prepareNode`. Interaction during placement is gated by `isGhostPlacing`, not by the flag, so only the document's copy is affected. ## How it is proven `ghostPlacementFlag.test.ts` drives the real `createGraphMutations` seam: add a node whose payload carries `flags.ghost`, delete the flag the way placement does, reconcile from the document, assert the live node is no longer ghosted. Asserted at the invariant rather than at a chosen fix site. `layoutMintPort.test.ts` covers the mint side: a node serialized mid-placement mints an `add_node` with `ghost` gone and its other flags intact. Reverting the two source files puts both back to red. Patch coverage 100%. Related symptoms are collected in the internal agent-feedback tracker; this PR covers the ghost-node half only, not the separate links-not-rendering report filed alongside it. 🤖 Generated with [Claude Code](https://claude.com/claude-code) https://claude.ai/code/session_016jorVkGmycSkD6j7A6sdJV --------- Co-authored-by: Claude <noreply@anthropic.com> (cherry picked from commit dad05c5)
<!-- ccr-slack-attribution --> _Requested by **Jo Zhang** · [Slack thread](https://comfy-organization.slack.com/archives/C0BGH348Z0C/p1789893497566839)_ ## Summary **Before:** Starting a new chat in the agent panel kept targeting whatever workflow tab the *previous* chat was pointed at, even after you had switched to a different tab (or opened a brand new one). Telling the agent to do something "in this workflow" could silently act on a tab that was no longer on screen. **After:** Starting a new chat now targets whichever workflow tab is on screen at that moment - whether you got there by switching to an already-open tab or by creating a fresh one. ## Changes - **What**: `onNewChat()` in `AgentPanelRoot.vue` now sets the agent panel's workflow target to `workflowStore.activeWorkflow` before starting the new chat, instead of leaving the previous chat's target in place. This mirrors `onSelectHistory()`, which already resets the target before loading a different thread. - **Breaking**: none ## Review Focus - `AgentPanelRoot.test.ts`: updated `retains an explicitly chosen target across panel reopen and New Chat` (renamed to `... across a plain panel reopen`) because it asserted the stale-target carryover as intentional; a plain panel reopen (no New Chat) still keeps the explicit target, which is unchanged. - Added tests: New Chat retargets to a tab switched to, to a freshly created/unsaved tab, and clears to "no target" when no tab is open. - Added a Playwright e2e spec, `browser_tests/tests/agent/agentNewChatTargetWorkflow.spec.ts`, covering the same tab-switch and tab-creation flows end to end (with screenshot attachments so a video is captured for this area). - Supersedes the repro-only PR #18080. That PR's test asserted a blank "select a workflow" state after New Chat; the accepted PM-1321 acceptance criteria call for the tab on screen instead, so this PR ships its own e2e spec rather than extending that one. Fixes PM-1321, PM-1322, PM-1415 ## Tests run - `pnpm typecheck` - clean - `pnpm lint` (eslint/oxlint/oxfmt scoped to the changed files, plus the repo's pre-commit/pre-push hooks) - clean - `pnpm test:unit -- src/workbench/extensions/agent/AgentPanelRoot.test.ts` - 226/226 passed - `pnpm test:unit -- src/workbench/extensions/agent/stores/agent/agentPanelTarget.test.ts` - 2/2 passed - Full unit suite (`pnpm test:unit`) - 22367 passed, 14 expected fail, 7 skipped, 0 regressions ## Screenshots (if applicable) See the attached screenshots/video on the new Playwright spec's test run in CI. 🤖 Generated with [Claude Code](https://claude.com/claude-code) https://claude.ai/code/session_015KZNRdWz5cXQPvyefA7ChC Co-authored-by: Claude <noreply@anthropic.com> (cherry picked from commit ef479c2)
|
The latest updates on your projects. Learn more about Vercel for GitHub. 1 Skipped Deployment
|
|
Important Review skippedIgnore keyword(s) in the title. ⛔ Ignored keywords (2)
Please check the settings in the CodeRabbit UI or the ⚙️ Run configurationConfiguration used: Repository: Comfy-Org/ComfyUI_frontend/.coderabbit.yaml Review profile: ASSERTIVE Plan: Team Run ID: You can disable this status message by setting the Use the checkbox below for a quick retry:
Comment |
🎨 Storybook: ✅ Built — View Storybook🎭 Playwright: ❌ 1939 passed, 1 failed❌ Failed Tests📊 Browser Reports
📦 Bundle Size
⚡ Performance Report
Absolute values
Raw data{
"timestamp": "2026-09-21T07:25:39.639Z",
"gitSha": "a81045c1531138db99602134afa9051b07713f52",
"branch": "backport-cloud-1.54-ms215-gap-20260921",
"measurements": [
{
"name": "canvas-idle",
"durationMs": 2006.9070000000124,
"styleRecalcs": 9,
"styleRecalcDurationMs": 6.9079999999999995,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 416.697,
"heapDeltaBytes": -4745712,
"heapUsedBytes": 65542076,
"domNodes": -281,
"jsHeapTotalBytes": 4186112,
"scriptDurationMs": 7.674,
"eventListeners": -194,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "canvas-idle",
"durationMs": 2030.5270000000064,
"styleRecalcs": 10,
"styleRecalcDurationMs": 8.191000000000003,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 440.37500000000006,
"heapDeltaBytes": 4520296,
"heapUsedBytes": 74648748,
"domNodes": -276,
"jsHeapTotalBytes": 4710400,
"scriptDurationMs": 9.021999999999998,
"eventListeners": -194,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.66333333333335,
"p95FrameDurationMs": 16.799999999999272
},
{
"name": "canvas-mouse-sweep",
"durationMs": 1842.1129999999835,
"styleRecalcs": 74,
"styleRecalcDurationMs": 26.473,
"layouts": 14,
"layoutDurationMs": 3.689,
"taskDurationMs": 816.4259999999999,
"heapDeltaBytes": 8928580,
"heapUsedBytes": 78827388,
"domNodes": -280,
"jsHeapTotalBytes": 5496832,
"scriptDurationMs": 134.78300000000002,
"eventListeners": -196,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.800000000000182
},
{
"name": "canvas-mouse-sweep",
"durationMs": 1783.7749999999346,
"styleRecalcs": 76,
"styleRecalcDurationMs": 32.004,
"layouts": 14,
"layoutDurationMs": 4.787999999999999,
"taskDurationMs": 788.5380000000001,
"heapDeltaBytes": 12749088,
"heapUsedBytes": 83222040,
"domNodes": -279,
"jsHeapTotalBytes": 5758976,
"scriptDurationMs": 131.07,
"eventListeners": -164,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "canvas-zoom-sweep",
"durationMs": 1706.4649999999801,
"styleRecalcs": 31,
"styleRecalcDurationMs": 13.438000000000002,
"layouts": 6,
"layoutDurationMs": 0.97,
"taskDurationMs": 293.17400000000004,
"heapDeltaBytes": 4660744,
"heapUsedBytes": 74557056,
"domNodes": 74,
"jsHeapTotalBytes": 4980736,
"scriptDurationMs": 10.273000000000001,
"eventListeners": 19,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.800000000000182
},
{
"name": "canvas-zoom-sweep",
"durationMs": 1656.6510000000108,
"styleRecalcs": 29,
"styleRecalcDurationMs": 12.813999999999998,
"layouts": 6,
"layoutDurationMs": 0.9510000000000001,
"taskDurationMs": 295.844,
"heapDeltaBytes": 4677524,
"heapUsedBytes": 75805704,
"domNodes": 76,
"jsHeapTotalBytes": 5505024,
"scriptDurationMs": 10.633999999999999,
"eventListeners": 19,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.66333333333332,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "dom-widget-clipping",
"durationMs": 542.222000000038,
"styleRecalcs": 12,
"styleRecalcDurationMs": 6.6339999999999995,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 366.80400000000003,
"heapDeltaBytes": 18572296,
"heapUsedBytes": 88589884,
"domNodes": 20,
"jsHeapTotalBytes": 4718592,
"scriptDurationMs": 128.06599999999997,
"eventListeners": 2,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.66333333333332,
"p95FrameDurationMs": 16.800000000000182
},
{
"name": "dom-widget-clipping",
"durationMs": 546.6800000000376,
"styleRecalcs": 12,
"styleRecalcDurationMs": 7.261,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 369.178,
"heapDeltaBytes": -8198188,
"heapUsedBytes": 63007376,
"domNodes": 20,
"jsHeapTotalBytes": 7602176,
"scriptDurationMs": 125.84199999999998,
"eventListeners": 0,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666682,
"p95FrameDurationMs": 16.700000000000273
},
{
"name": "large-graph-idle",
"durationMs": 2021.366999999998,
"styleRecalcs": 6,
"styleRecalcDurationMs": 4.216999999999999,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 546.1339999999999,
"heapDeltaBytes": -15971904,
"heapUsedBytes": 87344944,
"domNodes": -265,
"jsHeapTotalBytes": -1380352,
"scriptDurationMs": 29.555999999999997,
"eventListeners": -160,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.670000000000012,
"p95FrameDurationMs": 16.800000000000182
},
{
"name": "large-graph-idle",
"durationMs": 2006.6390000000638,
"styleRecalcs": 7,
"styleRecalcDurationMs": 5.513000000000001,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 530.624,
"heapDeltaBytes": 5124312,
"heapUsedBytes": 81689864,
"domNodes": -275,
"jsHeapTotalBytes": -630784,
"scriptDurationMs": 32.153,
"eventListeners": -194,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.66333333333332,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "large-graph-pan",
"durationMs": 2183.9390000000094,
"styleRecalcs": 68,
"styleRecalcDurationMs": 12.181999999999999,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 1355.127,
"heapDeltaBytes": 8569948,
"heapUsedBytes": 85524248,
"domNodes": -244,
"jsHeapTotalBytes": -405504,
"scriptDurationMs": 651.258,
"eventListeners": -190,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.800000000000182
},
{
"name": "large-graph-pan",
"durationMs": 2244.8530000000346,
"styleRecalcs": 68,
"styleRecalcDurationMs": 12.995999999999999,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 1415.891,
"heapDeltaBytes": 4283600,
"heapUsedBytes": 81379256,
"domNodes": -242,
"jsHeapTotalBytes": 643072,
"scriptDurationMs": 696.099,
"eventListeners": -190,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.699999999999818
},
{
"name": "large-graph-zoom",
"durationMs": 3195.245,
"styleRecalcs": 62,
"styleRecalcDurationMs": 10.584000000000003,
"layouts": 60,
"layoutDurationMs": 11.378,
"taskDurationMs": 1367.87,
"heapDeltaBytes": 8435180,
"heapUsedBytes": 75629452,
"domNodes": 4,
"jsHeapTotalBytes": 0,
"scriptDurationMs": 615.004,
"eventListeners": 8,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.699999999999818
},
{
"name": "large-graph-zoom",
"durationMs": 3241.1260000000084,
"styleRecalcs": 62,
"styleRecalcDurationMs": 10.654,
"layouts": 60,
"layoutDurationMs": 11.452,
"taskDurationMs": 1401.274,
"heapDeltaBytes": 8568944,
"heapUsedBytes": 76407052,
"domNodes": 4,
"jsHeapTotalBytes": 0,
"scriptDurationMs": 642.2800000000001,
"eventListeners": 8,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.66333333333335,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "legacy-node-drag",
"durationMs": 2286.231999999984,
"styleRecalcs": 42,
"styleRecalcDurationMs": 5.934000000000002,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 1531.6009999999999,
"heapDeltaBytes": 18685500,
"heapUsedBytes": 88290816,
"domNodes": 4,
"jsHeapTotalBytes": 1728512,
"scriptDurationMs": 745.676,
"eventListeners": 186,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.66333333333335,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "legacy-node-drag",
"durationMs": 2145.4619999999522,
"styleRecalcs": 42,
"styleRecalcDurationMs": 6.318999999999998,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 1543.104,
"heapDeltaBytes": 16560104,
"heapUsedBytes": 84740772,
"domNodes": 4,
"jsHeapTotalBytes": 417792,
"scriptDurationMs": 748.001,
"eventListeners": 186,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.66333333333332,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "minimap-idle",
"durationMs": 2020.604999999989,
"styleRecalcs": 6,
"styleRecalcDurationMs": 4.8229999999999995,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 504.22,
"heapDeltaBytes": 5099160,
"heapUsedBytes": 83334628,
"domNodes": -279,
"jsHeapTotalBytes": -1417216,
"scriptDurationMs": 28.526000000000003,
"eventListeners": -190,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.66333333333335,
"p95FrameDurationMs": 16.800000000000182
},
{
"name": "minimap-idle",
"durationMs": 1997.953999999936,
"styleRecalcs": 4,
"styleRecalcDurationMs": 3.0960000000000014,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 456.5349999999999,
"heapDeltaBytes": 21803856,
"heapUsedBytes": 88196392,
"domNodes": -269,
"jsHeapTotalBytes": 0,
"scriptDurationMs": 27.904000000000003,
"eventListeners": 6,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.66333333333335,
"p95FrameDurationMs": 16.799999999999272
},
{
"name": "subgraph-dom-widget-clipping",
"durationMs": 519.7959999999853,
"styleRecalcs": 47,
"styleRecalcDurationMs": 9.074999999999998,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 325.62499999999994,
"heapDeltaBytes": 15144828,
"heapUsedBytes": 87958420,
"domNodes": 20,
"jsHeapTotalBytes": 6029312,
"scriptDurationMs": 110.474,
"eventListeners": 6,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666682,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "subgraph-dom-widget-clipping",
"durationMs": 545.4739999998992,
"styleRecalcs": 47,
"styleRecalcDurationMs": 9.789,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 350.11800000000005,
"heapDeltaBytes": 15224072,
"heapUsedBytes": 88060132,
"domNodes": 20,
"jsHeapTotalBytes": 6029312,
"scriptDurationMs": 117.72000000000001,
"eventListeners": 8,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "subgraph-idle",
"durationMs": 2018.287999999984,
"styleRecalcs": 9,
"styleRecalcDurationMs": 6.290000000000002,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 424.607,
"heapDeltaBytes": -14269204,
"heapUsedBytes": 58462744,
"domNodes": -296,
"jsHeapTotalBytes": 5234688,
"scriptDurationMs": 8.273,
"eventListeners": -210,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "subgraph-idle",
"durationMs": 2025.9459999999763,
"styleRecalcs": 11,
"styleRecalcDurationMs": 6.716999999999999,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 422.68500000000006,
"heapDeltaBytes": -3572340,
"heapUsedBytes": 68441720,
"domNodes": -295,
"jsHeapTotalBytes": 5496832,
"scriptDurationMs": 8.110999999999999,
"eventListeners": -212,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.799999999999272
},
{
"name": "subgraph-mouse-sweep",
"durationMs": 1743.8730000000078,
"styleRecalcs": 77,
"styleRecalcDurationMs": 23.171999999999997,
"layouts": 18,
"layoutDurationMs": 4.136,
"taskDurationMs": 702.108,
"heapDeltaBytes": -13741348,
"heapUsedBytes": 59067628,
"domNodes": -296,
"jsHeapTotalBytes": 4972544,
"scriptDurationMs": 90.961,
"eventListeners": -210,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "subgraph-mouse-sweep",
"durationMs": 1689.5729999999958,
"styleRecalcs": 77,
"styleRecalcDurationMs": 23.588,
"layouts": 18,
"layoutDurationMs": 4.328,
"taskDurationMs": 688.694,
"heapDeltaBytes": 490492,
"heapUsedBytes": 66234768,
"domNodes": -293,
"jsHeapTotalBytes": 4972544,
"scriptDurationMs": 89.754,
"eventListeners": -210,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.699999999999818
},
{
"name": "subgraph-transition-enter",
"durationMs": 900.9459999999763,
"styleRecalcs": 20,
"styleRecalcDurationMs": 25.227000000000007,
"layouts": 16,
"layoutDurationMs": 12.303999999999998,
"taskDurationMs": 744.5689999999998,
"heapDeltaBytes": 21118164,
"heapUsedBytes": 109070300,
"domNodes": 13751,
"jsHeapTotalBytes": 11010048,
"scriptDurationMs": 22.729,
"eventListeners": 2385,
"totalBlockingTimeMs": 89,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.699999999999818
},
{
"name": "viewport-pan-sweep",
"durationMs": 8396.529999999984,
"styleRecalcs": 249,
"styleRecalcDurationMs": 37.425,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 4774.142,
"heapDeltaBytes": -6337184,
"heapUsedBytes": 69786476,
"domNodes": -277,
"jsHeapTotalBytes": 4087808,
"scriptDurationMs": 2178.9429999999998,
"eventListeners": -174,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.799999999999272
},
{
"name": "viewport-pan-sweep",
"durationMs": 8298.969000000056,
"styleRecalcs": 250,
"styleRecalcDurationMs": 37.695,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 4691.8,
"heapDeltaBytes": -16465172,
"heapUsedBytes": 86411292,
"domNodes": -263,
"jsHeapTotalBytes": 417792,
"scriptDurationMs": 2146.08,
"eventListeners": -144,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "vue-large-graph-idle",
"durationMs": 14713.741000000027,
"styleRecalcs": 0,
"styleRecalcDurationMs": 0,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 14378.431,
"heapDeltaBytes": -48799732,
"heapUsedBytes": 176544352,
"domNodes": -8068,
"jsHeapTotalBytes": -12361728,
"scriptDurationMs": 211.48999999999998,
"eventListeners": -14933,
"totalBlockingTimeMs": 0,
"frameDurationMs": 17.220000000000073,
"p95FrameDurationMs": 16.799999999999272
},
{
"name": "vue-large-graph-idle",
"durationMs": 15352.146000000062,
"styleRecalcs": 0,
"styleRecalcDurationMs": 0,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 14853.894999999999,
"heapDeltaBytes": -44628584,
"heapUsedBytes": 182512092,
"domNodes": -8068,
"jsHeapTotalBytes": -19054592,
"scriptDurationMs": 238.06199999999998,
"eventListeners": -14935,
"totalBlockingTimeMs": 0,
"frameDurationMs": 17.223333333333358,
"p95FrameDurationMs": 16.799999999999272
},
{
"name": "vue-large-graph-pan",
"durationMs": 17283.532999999978,
"styleRecalcs": 132,
"styleRecalcDurationMs": 14.760999999999996,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 16997.296,
"heapDeltaBytes": -55572408,
"heapUsedBytes": 168408292,
"domNodes": -8068,
"jsHeapTotalBytes": -15970304,
"scriptDurationMs": 747.897,
"eventListeners": -14931,
"totalBlockingTimeMs": 31,
"frameDurationMs": 17.776666666666642,
"p95FrameDurationMs": 16.799999999999272
},
{
"name": "vue-large-graph-pan",
"durationMs": 17638.048000000024,
"styleRecalcs": 133,
"styleRecalcDurationMs": 16.628000000000004,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 17393.18,
"heapDeltaBytes": -44217428,
"heapUsedBytes": 183347560,
"domNodes": -8068,
"jsHeapTotalBytes": -21729280,
"scriptDurationMs": 791.8199999999999,
"eventListeners": -14931,
"totalBlockingTimeMs": 17,
"frameDurationMs": 17.776666666666763,
"p95FrameDurationMs": 16.799999999999272
},
{
"name": "vue-node-resize-workload",
"durationMs": 5317.843999999979,
"styleRecalcs": 20,
"styleRecalcDurationMs": 61.01900000000002,
"layouts": 20,
"layoutDurationMs": 21.715,
"taskDurationMs": 5256.473,
"heapDeltaBytes": -18532240,
"heapUsedBytes": 199159748,
"domNodes": -8068,
"jsHeapTotalBytes": 18354176,
"scriptDurationMs": 95.30700000000002,
"eventListeners": -14933,
"totalBlockingTimeMs": 231,
"frameDurationMs": 17.220000000000013,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "vue-node-resize-workload",
"durationMs": 5457.218000000012,
"styleRecalcs": 20,
"styleRecalcDurationMs": 67.748,
"layouts": 20,
"layoutDurationMs": 25.291999999999994,
"taskDurationMs": 5356.394,
"heapDeltaBytes": -47943408,
"heapUsedBytes": 181797864,
"domNodes": -8068,
"jsHeapTotalBytes": 16764928,
"scriptDurationMs": 97.32700000000001,
"eventListeners": -14931,
"totalBlockingTimeMs": 235,
"frameDurationMs": 17.223333333333297,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "vue-zoom-out-idle",
"durationMs": 7329.073999999991,
"styleRecalcs": 0,
"styleRecalcDurationMs": 0,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 7172.5740000000005,
"heapDeltaBytes": 646296,
"heapUsedBytes": 176786628,
"domNodes": -1,
"jsHeapTotalBytes": -19660800,
"scriptDurationMs": 108.47300000000004,
"eventListeners": -12,
"totalBlockingTimeMs": 0,
"frameDurationMs": 17.220000000000073,
"p95FrameDurationMs": 16.799999999999272
},
{
"name": "vue-zoom-out-idle",
"durationMs": 7411.4680000000135,
"styleRecalcs": 0,
"styleRecalcDurationMs": 0,
"layouts": 0,
"layoutDurationMs": 0,
"taskDurationMs": 7275.446000000001,
"heapDeltaBytes": 23625372,
"heapUsedBytes": 195083912,
"domNodes": 0,
"jsHeapTotalBytes": -27525120,
"scriptDurationMs": 111.549,
"eventListeners": -12,
"totalBlockingTimeMs": 0,
"frameDurationMs": 17.223333333333358,
"p95FrameDurationMs": 16.700000000000728
},
{
"name": "workflow-execution",
"durationMs": 462.4109999999746,
"styleRecalcs": 20,
"styleRecalcDurationMs": 23.918999999999997,
"layouts": 6,
"layoutDurationMs": 2.104,
"taskDurationMs": 111.39500000000001,
"heapDeltaBytes": 5191676,
"heapUsedBytes": 73494064,
"domNodes": 151,
"jsHeapTotalBytes": 262144,
"scriptDurationMs": 13.561,
"eventListeners": 67,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.666666666666668,
"p95FrameDurationMs": 16.700000000000273
},
{
"name": "workflow-execution",
"durationMs": 81.32800000009865,
"styleRecalcs": 6,
"styleRecalcDurationMs": 9.982,
"layouts": 2,
"layoutDurationMs": 0.6779999999999999,
"taskDurationMs": 52.73799999999999,
"heapDeltaBytes": 3008916,
"heapUsedBytes": 67482568,
"domNodes": 104,
"jsHeapTotalBytes": 262144,
"scriptDurationMs": 4.580000000000001,
"eventListeners": 33,
"totalBlockingTimeMs": 0,
"frameDurationMs": 16.663333333333338,
"p95FrameDurationMs": 16.700000000000273
}
]
}🔬 E2E Coverage
Top 10 files by uncovered lines
|
Codecov Report✅ All modified and coverable lines are covered by tests. @@ Coverage Diff @@
## cloud/1.54 #18254 +/- ##
=============================================
Coverage ? 66.74%
=============================================
Files ? 1669
Lines ? 75190
Branches ? 0
=============================================
Hits ? 50184
Misses ? 25006
Partials ? 0
Flags with carried forward coverage won't be shown. Click here to find out more.
🚀 New features to boost your workflow:
|
|
Closing in favor of #18257: same tree, commit messages fixed so the AI co-author check passes. Full context for agent readersSix of the eight |
…t a cloud carrier (#18257) ## Summary Eight merged `main` Agent fixes that had no `cloud/1.54` carrier are cherry-picked onto the current release tip. Two sources need hand re-expression and are excluded. Runtime source is unmodified from the eight sources; seven follow-up commits are test-only adaptations to the release's module surface. Verification at the current head `2991cfc3f8` is hosted CI (77 success, 16 skipped, 0 failed) plus DrJKL's local run (see below); the picked browser specs had not been run against the target before this PR. Open gates: release-scope sign-off on the three non-Agent carries and exact-head human re-approval on `2991cfc3f8`, both tracked in [blocked-on-christian 825](christian-byrne/blocked-on-christian#825). Replaces [18254](#18254): six picks there inherited `Co-authored-by: Claude` trailers from the `main` squash commits and failed the AI co-author check. Runtime source is byte-identical to that PR's head `d1f35ac622`; two test files differ (listed under Adaptations). <details><summary>Full context for agent readers</summary> ## Changes Sources, all `cherry-pick -x -m 1`, clean, in merge order: - [18104](#18104) prefer previewUrl for dragged batch-output attachments - [18110](#18110) re-measure Vue node slot offsets after graph reload - [18117](#18117) record an undo entry when a link is dropped from a Vue node slot - [18118](#18118) stop a canvas click after redo from pushing a spurious undo entry - [18124](#18124) never silently no-op the subscription-required dialog - [18127](#18127) restore autogrow inputs after Agent reconciliation - [18191](#18191) keep the ghost placement flag out of the CRDT document - [18196](#18196) retarget new Agent chat to the workflow tab on screen Adaptations, all test-only, no runtime change: - `NodeSlots` reload unit test uses `createTestingPinia` + `setActivePinia` instead of `getActivePinia()!`, because the shared test Pinia setup from [17237](#17237) is `main`-only. - Browser fixture gains `NodeSlotReference.expectLinkCount` (same body as `main`, from [17403](#17403)); three picked Agent specs open the panel with the label the release UI renders, `agent.askComfyAgent`, because `agent.entryButton` from [17621](#17621) does not exist on this line. - `useSlotLinkInteraction.autoPan.test.ts`: the `globalEnums` mock is dropped (the picked composable now reads `RenderShape` from the real module, as on `main`), and the `@vueuse/core` mock spreads the actual module so release-native transitive imports such as `useStorage` resolve while the event-listener overrides stay focused. - `useSlotLinkInteraction.autoPan.test.ts` (`249123390c`): the workflow store is mocked at its seam instead of a partial module mock, per dante01yoon's review. - `subgraphPromotedWidgetExternalLink.spec.ts` (`2991cfc3f8`): the plain CLIPTextEncode "hides the text control" case and its two constants are deleted, because it asserts [17576](#17576) behavior (WidgetGrid retains labels for connected inputs) that is not on `cloud/1.54`. Re-add when 17576 lands on this line; the 17576 decision is [blocked-on-christian 824](christian-byrne/blocked-on-christian#824). Dependencies: none added; release lockfile pins preserved. Exclusions: [18063](#18063) and [18073](#18073) conflict because they edit `agentCrdtProjection.ts` and `agentCrdtDocLifecycle.ts`, files created by the follower split in [16908](#16908), which `cloud/1.54` never received. They need focused hand re-expression onto the pre-split follower, following [18219](#18219) and [18195](#18195); tracked as separate program rows. ## Review focus - Every source commit is an unmodified cherry-pick; review effort belongs on the test-only adaptation commits and on whether the eight sources are wanted on this line at all. - Sources 18110, 18117, and 18118 touch undo/redo and slot measurement outside the Agent panel, so they change behavior for users who never open the Agent. - This is not behavioral acceptance or a merge-check waiver; no target-native browser acceptance has run outside hosted CI. ## Exact-head verification Head history: `210e191fc9` (approved by DrJKL) received five test-only follow-ups: `a9ab888243` (module-surface alignment), `a63fef35ae` (`test.fixme` instead of `test.skip`, which the release's oxlint config rejects), `118ac5e9c6` (restore VueUse exports in the auto-pan mock, fixing the suite-collection failure DrJKL found at `a63fef35ae`), `249123390c` (workflow-store seam mock), and `2991cfc3f8` (delete the plain CLIPTextEncode browser case). - `210e191fc9`, local on top of `cloud/1.54` tip `1bbcdc66fa`: cherry-pick dry run of all 10 candidates, 8 clean individually and as a sequential stack, 2 conflicting as described above; `vitest run` on the 9 touched unit test files 405 passed; both `vue-tsc` projects exit 0; `oxfmt` and `eslint` clean; lint-staged and knip passed in hooks. - `a63fef35ae`, hosted: `lint-and-format` passed. DrJKL's local focused run: 405 passed plus one suite-collection failure in `useSlotLinkInteraction.autoPan.test.ts` (`useStorage` missing from the total `@vueuse/core` mock). Fixed in `118ac5e9c6`. - `118ac5e9c6`: hosted CI green; the unit `test` job covers the collection fix. - `2991cfc3f8` (current head): hosted CI 77 success, 16 skipped, 0 failed. DrJKL's local run on this head: both typecheck projects, ESLint, Oxlint, format, and Knip clean; Vitest 4 of 4 on the auto-pan suite; Playwright discovery 43 tests across 7 specs; Fallow clean. Author-side local verification is not run on this head (host resource hold); hosted CI plus DrJKL's run are the verification of record. Program receipt: [2026-09-21-ms215-gap-backport-cloud154.md](https://github.com/christian-byrne/in-app-agent-program/blob/main/reports/pr-ownership/2026-09-21-ms215-gap-backport-cloud154.md). </details> --------- Co-authored-by: claude[bot] <209825114+claude[bot]@users.noreply.github.com> Co-authored-by: test-bot <test-bot@users.noreply.github.com> Co-authored-by: Alexander Brown <drjkl@comfy.org> Co-authored-by: Amp <amp@ampcode.com>
Summary
Eight merged
mainAgent fixes that had nocloud/1.54carrier are cherry-picked onto the current release tip. Two sources need hand re-expression and are excluded. Unit tests, typechecks, formatting, and ESLint pass at the exact head; the picked browser specs were not run against the target.Full context for agent readers
Changes
cherry-pick -x -m 1, clean, in merge order): 18104 prefer previewUrl for dragged batch-output attachments; 18110 re-measure Vue node slot offsets after graph reload; 18117 record an undo entry when a link is dropped from a Vue node slot; 18118 stop a canvas click after redo from pushing a spurious undo entry; 18124 never silently no-op the subscription-required dialog; 18127 restore autogrow inputs after Agent reconciliation; 18191 keep the ghost placement flag out of the CRDT document; 18196 retarget new Agent chat to the workflow tab on screen.NodeSlotsreload unit test usescreateTestingPinia+setActivePiniainstead ofgetActivePinia()!, because the shared test Pinia setup from 17237 ismain-only.NodeSlotReference.expectLinkCount(same body asmain, from 17403); three picked Agent specs open the panel with the label the release UI renders,agent.askComfyAgent, becauseagent.entryButtonfrom 17621 does not exist on this line and adding an unrendered locale key would only satisfy the typecheck.agentCrdtProjection.tsandagentCrdtDocLifecycle.ts, files created by the follower split in 16908, whichcloud/1.54never received. They need focused hand re-expression onto the pre-split follower, following 18219 and 18195; tracked as separate program rows.Review focus
Exact-head verification
Local, Node 26.8.2, on top of
cloud/1.54tip1bbcdc66fa(the current staging pin):vitest runon the 9 touched unit test files: 405 passed, 0 failed;vue-tsc --noEmit(main project) andvue-tsc --project browser_tests/tsconfig.json: both exit 0 at this head (browser typecheck failed before the fixture and label adaptations);oxfmtclean andeslintzero errors on all touched files; lint-staged (oxfmt, oxlint, eslint, both typechecks) passed in the pre-commit hook; knip passed in pre-push.changeTracker,linkInteraction,subgraphPromotedWidgetExternalLink,agentBatchOutputAttachment,agentUploadedImageAttachment,agentNewChatTargetWorkflow,agentPaywallSubscribe) against acloud/1.54build. Hosted CI on this PR is the first browser execution of the adapted specs on the target.Program receipt: 2026-09-21-ms215-gap-backport-cloud154.md.