## Summary
Adds a GitHub Actions workflow + TypeScript script that posts to Slack
when a merged PR improves unit or E2E test coverage.
## Changes
- **What**: New `coverage-slack-notify.yaml` workflow triggered on push
to main. Compares current coverage against previous baselines, generates
Slack Block Kit payload with progress bars and milestone celebrations,
posts to `#p-frontend-automated-testing`.
- **Script**: `scripts/coverage-slack-notify.ts` — parses lcov files,
computes deltas, detects milestone crossings (every 5%), builds Slack
payload. Pure functions exported for testability.
- **Tests**: 26 unit tests in `scripts/coverage-slack-notify.test.ts`
covering all pure functions including edge cases (malformed lcov, exact
boundaries, zero coverage).
### Security hardening
- All `${{ }}` expressions moved from `run:` blocks to `env:` variables
- `SLACK_BOT_TOKEN` passed via env var, not inline
- Unique heredoc delimiter (timestamp-based) prevents payload injection
- `parseInt` fallback (`|| 0`) guards against malformed lcov
- PR regex anchored to first line of commit message
### Robustness
- `continue-on-error: true` on Slack post step (outage does not fail the
job)
- Baseline save guarded by `steps.unit-tests.outcome == success`
(prevents corrupt baselines on test failure)
- Channel ID commented for maintainability
- Top-level `text` field added for Slack mobile push notifications
- Author linked to GitHub profile instead of bare `@username`
## Review Focus
- Workflow step ordering and conditional logic
- Security of expression handling and secret management
- Slack payload structure and Block Kit formatting
┆Issue is synchronized with this [Notion
page](https://www.notion.so/PR-10977-feat-add-Slack-notification-workflow-for-coverage-improvements-33d6d73d3650819c8950f483c83f297c)
by [Unito](https://www.unito.io)
---------
Co-authored-by: GitHub Action <action@github.com>
## Summary
This PR extends **Image Compare** browser tests, adds **stable
`data-testid` hooks** on `WidgetImageCompare.vue`, and aligns slider
interactions with the **same viewport element** used by
`useMouseInElement`, with **layout polling** to reduce flakes.
## What this PR does
- [x] Adds **`data-testid="image-compare-viewport"`** on the compare
area root (the `containerRef` div) so E2E targets the real slider hit
region instead of a long Tailwind class chain or the `<img>` box alone.
- [x] Adds **`data-testid="image-compare-empty"`** on the no-images
branch so the empty state can be asserted **without hard-coded English**
UI text.
- [x] Adds a **smoke** test that the widget **shows both images and the
drag handle** after value injection, with **`waitForImagesLoaded`** (no
extra full-node screenshot to avoid duplicating the default-50 golden).
- [x] Extends slider coverage to **clamp at both edges** (**0%** and
**~100%**) using the viewport locator and **`expect.poll`** until
**`boundingBox()`** is valid before reading coordinates.
- [x] Updates **`moveToPercentage`** to **`expect.poll`** until the
target locator has a **non-empty layout box** before moving the mouse.
- [x] Routes **hover**, **preserve position**, and **25% / 75%
screenshot** mouse moves through **`image-compare-viewport`**
(consistent with `containerRef`).
- [x] Adds an E2E assertion that the **workflow ImageCompare node** size
is **at least 400×350** (matches the widget workflow asset).
- [x] Hardens the **broken image** case: **`page.on('pageerror')`** /
**`page.off`** in **`finally`**, **`http://127.0.0.1:1/...`** URLs for
fast failure, **`expect.soft`** on key UI invariants, and a hard
assertion that **no page errors** were recorded.
- [x] Extends the **large batch (20×20)** test to **page to the last
index** and assert **counters `20 / 20`**, **previous enabled**, and
**next disabled** on both sides.
- [x] Renames the clamp test title to use an **ASCII hyphen** (`0-100%`)
for easier grepping.
## Out of scope (unchanged in this PR)
- [ ] Replacing **`setImageCompareValue`**’s **`page.evaluate`** setup
with a full UI-driven path (would be a larger follow-up).
## Suggested title
`test: expand Image Compare E2E and stabilize widget selectors`
<!-- CURSOR_SUMMARY -->
---
> [!NOTE]
> **Low Risk**
> Primarily test and selector-hook changes; low production risk, with
only minor DOM attribute additions that could affect external test
tooling if relied upon.
>
> **Overview**
> Improves Image Compare Playwright coverage and reduces flakiness by
driving slider interactions through a new stable
`data-testid="image-compare-viewport"` hook and polling for a valid
layout box before mouse moves.
>
> Updates assertions to avoid localized text (new
`data-testid="image-compare-empty"`), adds smoke coverage that
images/handle render after value injection, validates slider clamping at
both 0% and ~100%, and extends screenshot tests to use the same viewport
target.
>
> Hardens edge-case tests by ensuring broken image loads don’t raise
uncaught `pageerror`s, adds a minimum node size assertion, and extends
large-batch navigation checks through the final index and
disabled/enabled nav states.
>
> <sup>Reviewed by [Cursor Bugbot](https://cursor.com/bugbot) for commit
ce3f7fbf8c. Bugbot is set up for automated
code reviews on this repo. Configure
[here](https://www.cursor.com/dashboard/bugbot).</sup>
<!-- /CURSOR_SUMMARY -->
┆Issue is synchronized with this [Notion
page](https://www.notion.so/PR-11196-test-expand-Image-Compare-E2E-and-stabilize-widget-selectors-3416d73d3650814d8d2bf207943d6205)
by [Unito](https://www.unito.io)
## Summary
- Add `browser_tests/tests/bottomPanel.spec.ts` with tests for behaviors
not covered by existing `bottomPanelLogs` and `bottomPanelShortcuts`
specs
- Tests cover: close button (X), tab persistence on re-open, resize
gutter visibility and drag, canvas interaction when panel is closed,
cross-panel switching (terminal <-> shortcuts), and registered tab
enumeration
- Extend `BottomPanel` fixture with `closeButton` and `resizeGutter`
locators
## Test plan
- [ ] `pnpm test:browser:local` passes all new tests in
`bottomPanel.spec.ts`
- [ ] Existing `bottomPanelLogs.spec.ts` and
`bottomPanelShortcuts.spec.ts` are unaffected
- [ ] `pnpm typecheck:browser` passes
- [ ] `pnpm lint` passes
┆Issue is synchronized with this [Notion
page](https://www.notion.so/PR-10814-test-add-E2E-tests-for-bottom-panel-core-behaviors-3366d73d365081ea9b90c643897845fa)
by [Unito](https://www.unito.io)
---------
Co-authored-by: dante <dante@danteui-MacStudio.local>
Co-authored-by: GitHub Action <action@github.com>
Co-authored-by: Alexander Brown <drjkl@comfy.org>
## Summary
- Adds `selectionToolboxMoreActions.spec.ts` with E2E coverage for
previously untested toolbox actions
- Covers: pin/unpin, minimize/expand, adjust size, copy, duplicate,
refresh button, align (top/left), distribute (horizontal/vertical),
alignment options hidden for single selection, multi-node bypass toggle
- Part of the FixIt Burndown test coverage initiative (toolbox actions)
## Test plan
- [ ] All new tests pass in CI
- [ ] No regressions in existing selectionToolbox tests
┆Issue is synchronized with this [Notion
page](https://www.notion.so/PR-10968-test-expand-E2E-coverage-for-toolbox-actions-33c6d73d3650811286cefdd0eb4f5242)
by [Unito](https://www.unito.io)
---------
Co-authored-by: Alexander Brown <drjkl@comfy.org>
## Summary
- Multiple SubgraphNode instances of the same blueprint share inner
nodes, causing promoted widget values to collide — the last configured
instance overwrites all previous values
- Add per-instance value storage (`_instanceWidgetValues`) on
SubgraphNode so each instance preserves its own promoted widget values
independently
- Restore `widgets_values` from serialized data into this per-instance
map after promoted views are created during configure
- Fixes#10146
## Root Cause
When loading a workflow with multiple SubgraphNode instances of the same
blueprint:
1. `LGraph.configure()` creates ONE shared Subgraph per blueprint (line
2625)
2. Each SubgraphNode instance calls `configure(instanceData)`
sequentially
3. `PromotedWidgetView.value` setter writes to the **shared inner node's
widget** (`promotedWidgetView.ts:199`)
4. The last instance's `configure()` overwrites all previous instances'
values
**Regression**: Introduced by PR #8594 (WidgetValueStore, v1.41.3) which
centralized widget state without per-instance scoping for shared
blueprints.
## Fix
- **SubgraphNode**: Add `_instanceWidgetValues` Map and
`_pendingWidgetsValues` for configure-time restoration
- **PromotedWidgetView getter**: Check instance map first before falling
back to widget store / inner node
- **PromotedWidgetView setter**: Write to instance map to avoid shared
inner node mutation
- **_internalConfigureAfterSlots**: Apply serialized `widgets_values` to
per-instance map after promoted views are created
## Red-Green Verification
| Commit | CI Status | Purpose |
|--------|-----------|---------|
| `test: add failing tests for multi-instance subgraph widget value
collision` | 🔴 Red | Proves widget values collide across
instances |
| `fix: store promoted widget values per SubgraphNode instance` |
🟢 Green | Per-instance storage prevents collision |
## Test Plan
- [x] CI red on test-only commit
- [x] CI green on fix commit
- [x] Unit test: `preserves promoted widget values after configure with
different widgets_values`
- [x] All 253 existing subgraph tests pass
- [ ] Manual: load workflow from issue image → verify 3 subgraph
instances produce different results
┆Issue is synchronized with this [Notion
page](https://www.notion.so/PR-10849-fix-store-promoted-widget-values-per-SubgraphNode-instance-3386d73d3650815a8544f54adcc0d504)
by [Unito](https://www.unito.io)
---------
Co-authored-by: dante <dante@danteui-MacStudio.local>
Leveraging the fancy coverage functionality of #10930, this PR aims to
add coverage to missing dialogue models.
This has proven quite constructive as many of the dialogues have since
been shown to be bugged.
- The APINodes sign in dialog that displays when attempting to run a
workflow containing Partner nodes while not logged in was intended to
display a list of nodes required to execute the workflow. The import for
this component was forgotten in the original commit (#3532) and the
backing component was later knipped
- Error dialogs resulting are intended to display the file responsible
for the error, but the prop was accidentally left out during the
refactoring of #3265
- ~~The node library migration (#8548) failed to include the 'Edit
Blueprint' button, and had incorrect sizing and color on the 'Delete
Blueprint' button.~~
- On request, the library button changes were spun out to a separate PR
┆Issue is synchronized with this [Notion
page](https://www.notion.so/PR-11133-Add-missing-dialog-tests-33e6d73d3650812cb142d610461adcd4)
by [Unito](https://www.unito.io)
---------
Co-authored-by: GitHub Action <action@github.com>
The node library migration (#8548) failed to include the 'Edit
Blueprint' button, and had incorrect sizing and color on the 'Delete
Blueprint' button.
- Re-add edit blueprint which was missed in the migration
- Fix incorrect sizing on delete blueprint
- Fix color (lucide uses background, not text)
- Migrate all action buttons use our capital 'B' Button component and
use standardized variants where possible
Spun out of #11133
┆Issue is synchronized with this [Notion
page](https://app.notion.com/p/PR-11232-Fix-node-library-action-buttons-3426d73d365081339cafc7457c0b5421)
by [Unito](https://www.unito.io)
## Summary
Add Claude Code PreToolUse hooks to block agents from bypassing
package.json scripts with raw tool invocations.
## Changes
- **What**: 15 PreToolUse hooks in `.claude/settings.json` that
intercept `npx`/`pnpx`/bare invocations of tsc, vitest, eslint,
prettier, oxlint, stylelint, and knip — redirecting agents to the
correct `pnpm` script (`pnpm typecheck`, `pnpm test:unit`, `pnpm lint`,
etc.)
- Also removes stale `permissions.allow` entries left over from a
debugging session
## Review Focus
- Pattern coverage: are there common agent invocations we're missing?
- The `if` field only supports simple `*` globs (no alternation), so
each pattern needs its own hook entry
┆Issue is synchronized with this [Notion
page](https://www.notion.so/PR-11201-feat-add-PreToolUse-hooks-to-enforce-pnpm-scripts-3416d73d365081a59a38c86ee4669aee)
by [Unito](https://www.unito.io)
Co-authored-by: Amp <amp@ampcode.com>
## Summary
Add `release_type` input (`minor`/`patch`) to the release workflow so
patch releases can target the current production branch instead of
always preferring the next minor.
## Problem
When a patch release is needed for `core/1.42` but `core/1.43` already
exists, the resolver always prefers `1.43`. There was no way to do a
patch release with PyPI publish + ComfyUI PR for the current production
version.
## Changes
- Rename workflow from "Release: Bi-weekly ComfyUI" → "Release: ComfyUI"
(serves both cadences)
- Add `release_type` choice input: `minor` (default, bi-weekly) vs
`patch` (hotfix for current production version)
- Update `resolve-comfyui-release.ts` to read `RELEASE_TYPE` env var for
branch targeting
- Scheduled runs continue to work as before (default to `minor`)
## Usage
```bash
# Bi-weekly minor release (or just let the schedule run)
gh workflow run release-biweekly-comfyui.yaml --ref main
# Patch release for current production version
gh workflow run release-biweekly-comfyui.yaml --ref main --field release_type=patch
```
┆Issue is synchronized with this [Notion
page](https://app.notion.com/p/PR-11223-ops-add-patch-release-support-to-ComfyUI-release-workflow-3426d73d365081498c15ff978a7f1116)
by [Unito](https://www.unito.io)
## Problem
PR #10338 added `useTransformCompatOverlayProps()` to FormDropdown and
FormDropdownMenuActions, which sets `appendTo: 'self'` in graph mode.
This breaks PrimeVue Popover positioning inside CSS-transformed
containers — the dropdown appears at incorrect Y positions.
## Root Cause
PrimeVue Popover with `appendTo: 'self'` renders the overlay inside the
component's DOM, inheriting parent CSS transforms. This causes the
popover to miscalculate its position when the parent has `transform:
scale()` or `translate()`.
## Fix
Remove the `appendTo` override from both FormDropdown and
FormDropdownMenuActions. PrimeVue defaults to `appendTo: 'body'`, which
teleports the popover to `<body>` — correctly positioning it outside any
CSS transform context.
- **Graph mode**: restores pre-#10338 behavior (`appendTo: 'body'`
default)
- **App mode**: unaffected — `'body'` is exactly what app mode needs
(prevents sidebar overflow clipping)
## Testing
- Existing unit tests pass (5/5)
- Typecheck clean
- Lint clean
- **E2E test rationale**: No E2E test added — this is a pure removal of
a prop override (reverting to PrimeVue defaults). The positioning bug
requires CSS transforms at specific viewport scales which are
impractical to assert reliably in Playwright. The existing
`subgraph-dom-widget-clipping` perf test exercises dropdown rendering in
transformed contexts and shows no regression.
Fixes#10499
Supersedes #11001 (temporary hotfix for backport)
---------
Co-authored-by: GitHub Action <action@github.com>
## Summary
Prevent early `progress_text` websocket events from throwing before the
graph canvas is initialized.
## Changes
- **What**: Guard `handleProgressText()` until `canvasStore.canvas`
exists, and add a regression test for a startup-time `progress_text`
event arriving before `GraphCanvas` finishes initialization.
## Review Focus
Confirm this is the right guard point for the startup race between
`GraphView` websocket binding and `GraphCanvas` async setup, and that
progress text behavior is unchanged once the canvas is ready.
## Validation
- `pnpm exec eslint src/stores/executionStore.ts
src/stores/executionStore.test.ts`
- `pnpm exec vitest run src/stores/executionStore.test.ts -t "should
ignore progress_text before the canvas is initialized"`
- `pnpm test:unit -- --run src/stores/executionStore.test.ts` still
reports one unrelated isolated-file failure in
`nodeLocatorIdToExecutionId` on current `main`
┆Issue is synchronized with this [Notion
page](https://www.notion.so/PR-11174-fix-guard-progress_text-before-canvas-init-3406d73d3650813dad23d511fb51add5)
by [Unito](https://www.unito.io)
## Summary
Follow-up cleanup for #11184 — removes redundant test setup calls that
the `@vue-nodes` fixture now handles.
## Changes
- **What**: Remove 40 lines of redundant `setSetting`, `setup()`, and
`waitForNodes()` calls across 11 test files
- `UseNewMenu: 'Top'` calls (already fixture default)
- `setup()` + `waitForNodes()` on default workflow (fixture already does
this for `@vue-nodes`)
- Page reload in `subgraphZeroUuid` (fixture applies VueNodes.Enabled
server-side before navigation)
## Review Focus
Each removal was verified against the fixture's `setupSettings()`
defaults (ComfyPage.ts:420-442) and the `@vue-nodes` auto-setup (lines
454-456). Tests that call `setup()`/`waitForNodes()` after
`loadWorkflow()` or `page.evaluate()` were intentionally kept.
┆Issue is synchronized with this [Notion
page](https://www.notion.so/PR-11195-test-remove-redundant-setup-settings-now-handled-by-vue-nodes-fixture-3416d73d36508154827df116a97e9130)
by [Unito](https://www.unito.io)
Co-authored-by: Amp <amp@ampcode.com>
## Summary
Track GTM `subscription_success` when a workspace subscription completes
synchronously in the dialog. The async billing-operation path already
emitted telemetry; the missing gap was the immediate `subscribed`
response.
## Changes
- **What**: Add the missing GTM success emission to both synchronous
workspace subscribe success branches while preserving the existing
toast, billing refresh, and dialog close behavior.
## Review Focus
Verify the synchronous `response.status === "subscribed"` workspace
dialog paths are the only missing frontend success emissions, while the
async billing-operation telemetry path remains unchanged.
This PR intentionally stays minimal. It does not add new browser
coverage yet; the previous component-level unit test was more
implementation-coupled than this fix justified, and a better long-term
test would be a higher-level workspace billing flow test once we have a
cleaner harness.
## Problem
API node generation status text (sent via `progress_text` WebSocket
binary messages) was not showing on local ComfyUI, but worked on cloud.
## Root Cause
The binary decoder for `progress_text` messages (eventType 3) checked
`getClientFeatureFlags()?.supports_progress_text_metadata` — the
**client's own flags** — to decide whether to parse the new format with
`prompt_id`. Since the client always advertises
`supports_progress_text_metadata: true`, it always tried to parse the
new wire format:
```
[4B event_type][4B prompt_id_len][prompt_id][4B node_id_len][node_id][text]
```
But the backend PR that adds `prompt_id` to the binary message
([ComfyUI#12540](https://github.com/Comfy-Org/ComfyUI/pull/12540)) was
**closed without merging**, so local ComfyUI still sends the legacy
format:
```
[4B event_type][4B node_id_len][node_id][text]
```
The decoder misinterpreted the `node_id_len` as `prompt_id_len`,
consuming the actual node_id bytes as a prompt_id, then producing
garbled `nodeId` and `text` — silently dropping all progress text
updates via the catch handler.
Cloud worked because the cloud backend supports and echoes the feature
flag.
## Fix
One-line change: check `serverFeatureFlags.value` (what the server
echoed back) instead of `getClientFeatureFlags()` (what the client
advertises).
## Tests
Added 3 tests covering:
- Legacy format parsing when server doesn't support the flag
- New format parsing when server does support the flag
- Corruption regression test: client advertises support but server
doesn't
┆Issue is synchronized with this [Notion
page](https://www.notion.so/PR-10996-fix-check-server-feature-flags-for-progress_text-binary-format-33d6d73d365081449a0dc918358799de)
by [Unito](https://www.unito.io)
---------
Co-authored-by: GitHub Action <action@github.com>
## Summary
Fix tests failing lint
## Changes
- **What**:
- Fix relative imports
- Fix test not using comfyPage
## Review Focus
<!-- Critical design decisions or edge cases that need attention -->
<!-- If this PR fixes an issue, uncomment and update the line below -->
<!-- Fixes #ISSUE_NUMBER -->
## Screenshots (if applicable)
<!-- Add screenshots or video recording to help explain your changes -->
┆Issue is synchronized with this [Notion
page](https://www.notion.so/PR-11182-refactor-fix-lint-errors-in-tests-3416d73d3650812fbf1bc88554c57de2)
by [Unito](https://www.unito.io)
## Summary
Muted and bypassed nodes are excluded from execution but were still
triggering missing model/media/node warnings. This PR makes the error
system mode-aware: muted/bypassed nodes no longer produce missing asset
errors, and all error lifecycle events (mode toggle, deletion, paste,
undo, tab switch) are handled consistently.
- Fixes Comfy-Org/ComfyUI#13256
## Behavioral notes
- **Tab switch overlay suppression (intentional)**: Switching back to a
workflow with missing assets no longer re-shows the error overlay. This
reverses the behavior introduced in #10190. The error state is still
restored silently in the errors tab — users can access it via the
properties panel without being interrupted by the overlay on every tab
switch.
## Changes
### 1. Scan filtering
- `scanAllModelCandidates`, `scanAllMediaCandidates`,
`scanMissingNodes`: skip nodes with `mode === NEVER || BYPASS`
- `collectMissingNodes` (serialized data): skip error reporting for
muted/bypassed nodes while still calling `sanitizeNodeName` for safe
`configure()`
- `collectEmbeddedModelsWithSource`: skip muted/bypassed nodes;
workflow-level `graphData.models` only create candidates when active
nodes exist
- `enrichWithEmbeddedMetadata`: filter unmatched workflow-level models
when all referencing nodes are inactive
### 2. Realtime mode change handling
- `useErrorClearingHooks.ts` chains `graph.onTrigger` to detect
`node:property:changed` (mode)
- Deactivation (active → muted/bypassed): remove missing
model/media/node errors for the node
- Activation (muted/bypassed → active): scan the node and add confirmed
errors, show overlay
- Subgraph container deactivation: remove all interior node errors
(execution ID prefix match)
- Subgraph container activation: scan all active interior nodes
recursively
- Subgraph interior mode change: resolve node via
`localGraph.getNodeById()` then compute execution ID from root graph
### 3. Node deletion
- `graph.onNodeRemoved`: remove missing model/media/node errors for the
deleted node
- Handle `node.graph === null` at callback time by using
`String(node.id)` for root-level nodes
### 4. Node paste/duplicate
- `graph.onNodeAdded`: scan via `queueMicrotask` (deferred until after
`node.configure()` restores widget values)
- Guard: skip during `ChangeTracker.isLoadingGraph` (undo/redo/tab
switch handled by pipeline)
- Guard: skip muted/bypassed nodes
### 5. Workflow tab switch optimization
- `skipAssetScans` option in `loadGraphData`: skip full pipeline on tab
switch
- Cache missing model/media/node state per workflow via
`PendingWarnings`
- `beforeLoadNewGraph`: save current store state to outgoing workflow's
`pendingWarnings`
- `showPendingWarnings`: restore cached errors silently (no overlay),
always sync missing nodes store (even when null)
- Preserve UI state (`fileSizes`, `urlInputs`) on tab switch by using
`setMissingModels([])` instead of `clearMissingModels()`
- `MissingModelRow.vue`: fetch file size on mount via
`fetchModelMetadata` memory cache
### 6. Undo/redo overlay suppression
- `silentAssetErrors` option propagated through pipeline →
`surfaceMissingModels`/`surfaceMissingMedia` `{ silent }` option
- `showPendingWarnings` `{ silent }` option for missing nodes overlay
- `changeTracker.ts`: pass `silentAssetErrors: true` on undo/redo
### 7. Error tab node filtering
- Selected node filters missing model/media card contents (not just
group visibility)
- `isAssetErrorInSelection`: resolve execution ID → graph node for
selection matching
- Missing nodes intentionally unfiltered (pack-level scope)
- `hasMissingMediaSelected` added to `RightSidePanel.vue` error tab
visibility
- Download All button: show only when 2+ downloadable models exist
### 8. New store functions
- `missingModelStore`: `addMissingModels`, `removeMissingModelsByNodeId`
- `missingMediaStore`: `addMissingMedia`, `removeMissingMediaByNodeId`
- `missingNodesErrorStore`: `removeMissingNodesByNodeId`
- `missingModelScan`: `scanNodeModelCandidates` (extracted single-node
scan)
- `missingMediaScan`: `scanNodeMediaCandidates` (extracted single-node
scan)
### 9. Test infrastructure improvements
- `data-testid` on `RightSidePanel.vue` tabs (`panel-tab-{value}`)
- Error-related TestIds moved from `dialogs` to `errorsTab` namespace in
`selectors.ts`
- Removed unused `TestIdValue` type
- Extracted `cleanupFakeModel` to shared `ErrorsTabHelper.ts`
- Renamed `openErrorsTabViaSeeErrors` → `loadWorkflowAndOpenErrorsTab`
- Added `aria-label` to pencil edit button and subgraph toggle button
## Test plan
### Unit tests (41 new)
- Store functions: `addMissing*`, `removeMissing*ByNodeId`
- `executionErrorStore`: `surfaceMissing*` silent option
- Scan functions: muted/bypassed filtering, `scanNodeModelCandidates`,
`scanNodeMediaCandidates`
- `workflowService`: `showPendingWarnings` silent, `beforeLoadNewGraph`
caching
### E2E tests (17 new in `errorsTabModeAware.spec.ts`)
**Missing nodes**
- [x] Deleting a missing node removes its error from the errors tab
- [x] Undo after bypass restores error without showing overlay
**Missing models**
- [x] Loading a workflow with all nodes bypassed shows no errors
- [x] Bypassing a node hides its error, un-bypassing restores it
- [x] Deleting a node with missing model removes its error
- [x] Undo after bypass restores error without showing overlay
- [x] Pasting a node with missing model increases referencing node count
- [x] Pasting a bypassed node does not add a new error
- [x] Selecting a node filters errors tab to only that node
**Missing media**
- [x] Loading a workflow with all nodes bypassed shows no errors
- [x] Bypassing a node hides its error, un-bypassing restores it
- [x] Pasting a bypassed node does not add a new error
- [x] Selecting a node filters errors tab to only that node
**Subgraph**
- [x] Bypassing a subgraph hides interior errors, un-bypassing restores
them
- [x] Bypassing a node inside a subgraph hides its error, un-bypassing
restores it
**Workflow switching**
- [x] Does not resurface error overlay when switching back to workflow
with missing nodes
- [x] Restores missing nodes in errors tab when switching back to
workflow
# Screenshots
https://github.com/user-attachments/assets/e0a5bcb8-69ba-4120-ab7f-5c83e4cfc3c5
## Follow-up work
- Extract error-detection computed properties from `RightSidePanel.vue`
into a composable (e.g. `useErrorsTabVisibility`)
---------
Co-authored-by: github-actions <github-actions@github.com>
Co-authored-by: GitHub Action <action@github.com>
## Summary
Adds Playwright E2E tests verifying that
1. the minimap canvas renders node content
2. clears when the graph is empty
3. correctly navigates the main canvas on click
┆Issue is synchronized with this [Notion
page](https://www.notion.so/PR-10738-test-add-minimap-E2E-tests-for-graph-content-and-click-to-navigate-3336d73d365081eb955ce711b3efc57f)
by [Unito](https://www.unito.io)
<!-- CURSOR_SUMMARY -->
---
> [!NOTE]
> **Low Risk**
> Low risk: changes are limited to adding `data-testid` attributes to
the minimap UI and expanding Playwright E2E assertions, with no
production behavior changes expected.
>
> **Overview**
> Strengthens minimap E2E coverage by switching existing assertions from
CSS selectors to new `data-testid`-based selectors and adding helper
utilities for canvas/overlay interactions.
>
> Adds new Playwright tests that verify the minimap canvas renders
content when nodes exist, clears when the graph is emptied, and that
clicking the minimap pans the main canvas (including a
post-`fitViewToSelectionAnimated` tolerance check).
>
> <sup>Reviewed by [Cursor Bugbot](https://cursor.com/bugbot) for commit
06e7542af1. Bugbot is set up for automated
code reviews on this repo. Configure
[here](https://www.cursor.com/dashboard/bugbot).</sup>
<!-- /CURSOR_SUMMARY -->
---------
Co-authored-by: Alexander Brown <drjkl@comfy.org>
Co-authored-by: Alexander Brown <DrJKL0424@gmail.com>
Co-authored-by: GitHub Action <action@github.com>
### Summary of Improvements
* **Custom Test Coverage Extension**: Enhanced the Painter widget E2E
test suite by refactoring logic for better maintainability and
robustness.
* **Stable Component Targeting**: Introduced
`data-testid="painter-dimension-text"` to `WidgetPainter.vue`, providing
a reliable, non-CSS-dependent locator for canvas size verification.
* **Improved Test Organization**: Reorganized existing test scenarios
into logical categories using `test.describe` blocks (Drawing, Brush
Settings, Canvas Size Controls, etc.).
* **Asynchronous Helper Integration**: Converted `hasCanvasContent` to
an asynchronous helper and unified its usage across multiple test cases
to eliminate redundant pixel-checking logic.
* **Locator Resilience**: Updated Reka UI slider interaction logic to
use more precise targeting (`:not([data-slot])`), preventing ambiguity
and improving test stability.
* **Scenario Refinement**: Updated the `pointerup` test logic to
accurately reflect pointer capture behavior when interactions occur
outside the canvas boundaries.
* **Enhanced Verification Feedback**: Added descriptive error messages
to `expect.poll` assertions to provide clearer context on potential
failure points.
* **Standardized Tagging**: Restored the original tagging strategy
(including `@smoke` and `@screenshot` tags) to ensure tests are
categorized correctly for CI environments.
### Red-Green Verification
| Commit | CI Status | Purpose |
| :--- | :--- | :--- |
| `test: refactor painter widget e2e tests and address review findings`
| 🟢 Green | Addresses all E2E test quality and stability issues from
review findings. |
### Test Plan
- [x] **Quality Checks**: `pnpm format`, `pnpm lint`, and `pnpm
typecheck` verified as passing.
- [x] **Component Integration**: `WidgetPainter.vue` `data-testid`
correctly applied and used in tests.
- [x] **Helper Reliability**: `hasCanvasContent` correctly identifies
colored pixels and returns a promise for `expect.poll`.
- [x] **Locator Robustness**: Verified Reka slider locators correctly
exclude internal thumb spans.
- [x] **Boundary Interaction**: Verified `pointerup` correctly ends
strokes when triggered outside the viewport.
- [x] **Tagging Consistency**: Verified `@smoke` and `@screenshot` tags
are present in the final test suite.
┆Issue is synchronized with this [Notion
page](https://www.notion.so/PR-10846-test-Painter-Widget-E2E-Test-Plan-3386d73d365081deb70fe4afbd417efb)
by [Unito](https://www.unito.io)
<!-- CURSOR_SUMMARY -->
---
> [!NOTE]
> **Low Risk**
> Primarily adds/refactors Playwright E2E tests and stable `data-testid`
hooks, with no changes to Painter drawing logic. Risk is limited to
potential test brittleness or minor UI attribute changes.
>
> **Overview**
> Expands the Painter widget Playwright suite with new grouped scenarios
covering drawing/erasing behavior, tool switching, brush inputs, canvas
resizing (including preserving drawings), clear behavior, and
serialization/upload flows (including failure toast).
>
> Refactors the tests to use a shared `@e2e/helpers/painter` module
(`drawStroke`, `hasCanvasContent`, `triggerSerialization`), improves
stability via role/testid-based locators and clearer `expect.poll`
messaging, and adds `data-testid` attributes (e.g.,
`painter-clear-button`, `painter-*-row`, `painter-dimension-text`) to
`WidgetPainter.vue` to avoid CSS-dependent selectors.
>
> <sup>Reviewed by [Cursor Bugbot](https://cursor.com/bugbot) for commit
053a8e9ed2. Bugbot is set up for automated
code reviews on this repo. Configure
[here](https://www.cursor.com/dashboard/bugbot).</sup>
<!-- /CURSOR_SUMMARY -->
---------
Co-authored-by: GitHub Action <action@github.com>
## Summary
Add E2E tests for ImageCompare widget
Covers slider interaction, batch navigation, single-image modes, visual
regression screenshots, and edge cases for the ImageCompare Vue node
widget.
┆Issue is synchronized with this [Notion
page](https://www.notion.so/PR-10767-test-add-E2E-tests-for-ImageCompare-widget-3346d73d365081c6bfc6fbd97fa04e4d)
by [Unito](https://www.unito.io)
<!-- CURSOR_SUMMARY -->
---
> [!NOTE]
> **Low Risk**
> Adds Playwright E2E coverage and screenshot assertions only; main risk
is increased CI runtime/flakiness due to additional image-loading and
hover/position polling.
>
> **Overview**
> Adds a new Playwright E2E suite for the ImageCompare Vue widget
(tagged `@widget`) that programmatically sets widget values and asserts
rendering for empty, single-image, and dual-image states.
>
> Expands coverage to **slider behavior** (default 50%, hover movement,
clamping, persistence) using polling on inline `clip-path`/handle
position, and adds **batch navigation** tests for multi-image
before/after sets.
>
> Introduces **visual regression screenshots** at default and specific
slider positions, plus edge-case tests for broken URLs, rapid updates
resetting batch index, legacy string values, and custom alt text.
>
> <sup>Reviewed by [Cursor Bugbot](https://cursor.com/bugbot) for commit
2c65440384. Bugbot is set up for automated
code reviews on this repo. Configure
[here](https://www.cursor.com/dashboard/bugbot).</sup>
<!-- /CURSOR_SUMMARY -->
---------
Co-authored-by: GitHub Action <action@github.com>
## Summary
Adds Playwright E2E tests for the ImageCropV2 widget covering
1. the empty state (no source image)
2. default control rendering
3. source image display with crop overlay
4. drag-to-reposition behavior.
┆Issue is synchronized with this [Notion
page](https://www.notion.so/PR-10737-test-add-E2E-tests-for-ImageCropV2-widget-3336d73d365081b28ed9db63e5df383e)
by [Unito](https://www.unito.io)
<!-- CURSOR_SUMMARY -->
---
> [!NOTE]
> **Low Risk**
> Low risk: primarily adds Playwright E2E coverage and introduces
`data-testid` attributes for more stable selectors, with no changes to
core crop behavior.
>
> **Overview**
> Adds new Playwright E2E coverage for the `ImageCropV2` Vue-node
widget, including workflows/fixtures for a disconnected input and a
`LoadImage -> ImageCropV2 -> PreviewImage` pipeline.
>
> Tests validate the empty state and default controls, verify the crop
overlay renders after execution with screenshot assertions, and exercise
drag-to-reposition by dispatching pointer events and asserting the
widget’s crop value updates.
>
> Updates `WidgetImageCrop.vue` to add `data-testid` hooks (empty
state/icon and crop overlay) to make the E2E selectors stable.
>
> <sup>Reviewed by [Cursor Bugbot](https://cursor.com/bugbot) for commit
9f29272742. Bugbot is set up for automated
code reviews on this repo. Configure
[here](https://www.cursor.com/dashboard/bugbot).</sup>
<!-- /CURSOR_SUMMARY -->
## Summary
Adds `HDRIManager` to load `.hdr/.exr` files as equirectangular
environment maps via **three.js** `RGBELoader/EXRLoader`
- Uploads HDRI files to the server via `/upload/image` API so they
persist across page reloads
- Restores HDRI state (enabled, **intensity**, **background**) from node
properties on reload
- Auto-enables "**Show as Background**" on successful upload for
immediate visual feedback
- Hides standard directional lights when HDRI is active; restores them
when disabled
- Hides the Light Intensity control while HDRI is active (lights have no
effect when HDRI overrides scene lighting)
- Limits HDRI availability to PBR-capable formats (.gltf, .glb, .fbx,
.obj); automatically disables when switching to an incompatible model
- Adds intensity slider and "**Show as Background**" toggle to the HDRI
panel
## How to Use HDRI Environment Lighting
1. Load a 3D model using a Load3D or Load3DViewer node (supported
formats: .gltf, .glb, .fbx, .obj)
2. Open the control panel → go to the Light tab
3. Click the globe icon to open the **HDRI panel**
4. Click Upload HDRI and select a` .hdr` or `.exr` file
5. The environment lighting applies automatically — the scene background
also updates to preview the panorama
6. Use the intensity slider to adjust the strength of the environment
lighting
7. Toggle Show as Background to show or hide the HDRI panorama behind
the model
## Screenshots
https://github.com/user-attachments/assets/1ec56ef0-853e-452f-ae2b-2474c9d0d781
┆Issue is synchronized with this [Notion
page](https://www.notion.so/PR-10818-feat-load3d-add-optional-HDRI-environment-lighting-to-3D-preview-nodes-3366d73d365081ea8c7ad9226b8b1e2f)
by [Unito](https://www.unito.io)
<!-- CURSOR_SUMMARY -->
---
> [!NOTE]
> **Medium Risk**
> Adds new HDRI loading/rendering path and persists new
`LightConfig.hdri` state, touching Three.js rendering, file uploads, and
node property restoration. Risk is moderate due to new async flows and
potential compatibility/performance issues with model switching and
renderer settings.
>
> **Overview**
> Adds optional **HDRI environment lighting** to Load3D previews,
including a new `HDRIManager` that loads `.hdr`/`.exr` files into
Three.js environment/background and exposes controls for enable/disable,
background display, and intensity.
>
> Extends `LightConfig` with an `hdri` block that is persisted on nodes
and restored on reload; `useLoad3d` now uploads HDRI files, loads them
into `Load3d`, maps scene light intensity to HDRI intensity, and
auto-disables HDRI when the current model format doesn’t support it.
>
> Updates the UI to include embedded HDRI controls under the Light panel
(with dismissable overlays and icon updates), adjusts light intensity
behavior when HDRI is active, and adds tests/strings/utilities
(`getFilenameExtension`, `mapSceneLightIntensityToHdri`, new constants)
to support the feature.
>
> <sup>Reviewed by [Cursor Bugbot](https://cursor.com/bugbot) for commit
b12c9722dc. Bugbot is set up for automated
code reviews on this repo. Configure
[here](https://www.cursor.com/dashboard/bugbot).</sup>
<!-- /CURSOR_SUMMARY -->
---------
Co-authored-by: Terry Jia <terryjia88@gmail.com>
## Summary
Consolidate duplicate `rgbToHSL` implementation — mask editor now uses
the shared `colorUtil.ts` version instead of its own copy.
## Changes
- Export `rgbToHsl` from `src/utils/colorUtil.ts` (was private)
- Replace 30-line local `rgbToHSL` in `useCanvasTools.ts` with a 2-line
wrapper that imports from `colorUtil.ts` and scales the return values
from 0-1 to degree/percentage
## Testing
### Automated
- All 176 existing tests pass (`colorUtil.test.ts` + `maskeditor/`
suite)
- No new tests needed — behavior is identical
### E2E Verification Steps
1. Open any image in the mask editor
2. Select the magic wand / color picker tool
3. Use HSL-based color matching — results should be identical to before
## Review Focus
The canonical `rgbToHsl` returns normalized 0-1 values while the mask
editor needs degree/percentage scale (h: 0-360, s: 0-100, l: 0-100). The
local wrapper handles this conversion.
Fixes#11080
┆Issue is synchronized with this [Notion
page](https://www.notion.so/PR-11134-chore-11080-consolidate-duplicate-rgbToHSL-use-shared-colorUtil-33e6d73d36508120bbd8f444f5cc94b6)
by [Unito](https://www.unito.io)
Co-authored-by: Terry Jia <terryjia88@gmail.com>
## Summary
Enable the previously disabled `playwright/no-force-option` lint rule at
error level and resolve all 29 violations across 10 files.
## Changes
### Lint rule
- `.oxlintrc.json`: `playwright/no-force-option` changed from `off` to
`error`
### Shared utility
- `CanvasHelper.ts`: Add `mouseClickAt()` and `mouseDblclickAt()`
methods that convert canvas-element-relative positions to absolute page
coordinates and use `page.mouse` APIs, avoiding Playwright's locator
actionability checks that fail when Vue DOM overlays sit above the
`<canvas>` element
### Force removal (20 violations)
- `selectionToolboxActions.spec.ts`: Remove `force: true` from 8 toolbox
button clicks (the `pointer-events: none` splitter overlay does not
intercept `elementFromPoint()`)
- `selectionToolboxSubmenus.spec.ts`: Remove `force: true` from 2
popover menu item clicks
- `BuilderSelectHelper.ts`: Remove `force: true` from 2 widget/node
clicks (builder mode does not disable pointer events)
- `linkInteraction.spec.ts`: Remove `force: true` from 3 slot `dragTo()`
calls (`::after` pseudo-elements do not intercept `elementFromPoint()`)
- `SidebarTab.ts`: Remove `force: true` from toast dismissal (`.catch()`
already handles failures)
- `nodeHelp.spec.ts`: Remove `force: true` from info button click
(preceding `toBeVisible()` assertion is sufficient)
### Rewrites (3 violations)
- `integerWidget.spec.ts`: Replace force-clicking disabled buttons with
`toBeDisabled()` assertions
- `Topbar.ts`: Replace force-click with `waitFor({ state: 'visible' })`
after hover
### Canvas coordinate clicks (9 violations)
- `litegraphUtils.ts`: Convert `NodeReference.click()` and
`navigateIntoSubgraph()` to use
`canvasOps.mouseClickAt()`/`mouseDblclickAt()`
- `subgraphPromotion.spec.ts`: Convert 3 right-click canvas calls to
`canvasOps.mouseClickAt()`
- `selectionToolboxSubmenus.spec.ts`: Convert 1 canvas dismiss-click to
`canvasOps.mouseClickAt()`
## Rationale
The original `force: true` usages were added defensively based on
incorrect assumptions about the `z-999 pointer-events: none` splitter
overlay intercepting Playwright's actionability checks. In reality,
`elementFromPoint()` skips elements with `pointer-events: none`, so the
overlay is transparent to Playwright's hit-test.
For canvas coordinate clicks, `force: true` on a locator does not tunnel
through DOM overlays — it only skips Playwright's preflight checks.
`page.mouse.click()` is the correct API for coordinate-based canvas
interactions.
┆Issue is synchronized with this [Notion
page](https://www.notion.so/PR-11164-fix-enable-playwright-no-force-option-lint-rule-33f6d73d365081e78601c6114121d272)
by [Unito](https://www.unito.io)
---------
Co-authored-by: Amp <amp@ampcode.com>
## Summary
Cloned/pasted nodes in Node 2.0 (Vue renderer) mode now appear above the
original node instead of behind it.
## Root Cause
The legacy LiteGraph canvas renderer uses array ordering for z-ordering:
nodes are stored in `graph._nodes` and drawn sequentially, so newly
added nodes (appended to the end) are automatically drawn on top. There
is no explicit z-index.
The Vue renderer (Node 2.0) uses explicit CSS `z-index` for node
ordering. New nodes default to `zIndex: 0` in `layoutMutations.ts`. When
a node has been interacted with, `bringNodeToFront` raises its z-index.
A cloned node at z-index 0 therefore appears behind any previously
interacted node.
The alt-click clone path in `LGraphNode.vue` already handles this
correctly by calling `bringNodeToFront()` after cloning. However, the
menu clone and keyboard paste paths go through `_deserializeItems` in
`LGraphCanvas.ts`, which does not set z-index for new nodes.
| Clone method | Legacy renderer | Vue renderer (before fix) | Vue
renderer (after fix) |
|---|---|---|---|
| Alt-click drag | On top (array order) | On top (`bringNodeToFront`
called) | On top |
| Right-click menu Clone | On top (array order) | Behind original
(z-index 0) | On top |
| Ctrl+C / Ctrl+V | On top (array order) | Behind original (z-index 0) |
On top |
## Steps to Reproduce
1. Enable Node 2.0 mode (Vue renderer) in settings
2. Add any node to the canvas
3. Click or drag the node (raises its z-index via `bringNodeToFront`)
4. Right-click the node and select "Clone"
5. **Expected**: Cloned node appears above the original, immediately
draggable
6. **Actual**: Cloned node appears behind the original; user must move
the original to access the clone
## Changes
After `batchUpdateNodeBounds` in `_deserializeItems`, calls
`bringNodeToFront` for each newly created node so they receive a z-index
above all existing nodes.
## Side Effect Analysis
Checked all call sites of `_deserializeItems`:
1. **Initial graph load / workflow open**: `loadGraphData` in `app.ts`
does NOT call `_deserializeItems`. Workflow loading goes through
`LGraph.configure()` which directly adds nodes and links. The layout
store is initialized separately via `initializeFromLiteGraph`. No side
effect.
2. **Paste from clipboard (Ctrl+V)**: Both `usePaste.ts` (line 52) and
`pasteFromClipboard` (line 4080) call `_deserializeItems`. Pasted nodes
appearing on top is the correct and desired behavior. No issue.
3. **Undo/Redo**: `ChangeTracker.updateState()` calls
`app.loadGraphData()`, which does a full graph reconfigure -- it does
NOT go through `_deserializeItems`. No side effect.
4. **Subgraph blueprint addition**: `litegraphService.ts` (line 906)
calls `_deserializeItems` when adding subgraph blueprints from the node
library. These are freshly placed nodes that should appear on top.
Desired behavior.
5. **Alt-click clone in LGraphNode.vue**: This path calls
`LGraphCanvas.cloneNodes()` -> `_deserializeItems()`, then separately
calls `bringNodeToFront()` again on line 433 of `LGraphNode.vue`. The
second call is now redundant (the node is already at max z-index), but
harmless -- `bringNodeToFront` finds the current max, adds 1, and sets.
The z-index will increment from N to N+1 on the second call. This is a
minor redundancy, not a bug.
6. **Performance**: `bringNodeToFront` iterates all nodes in the layout
store once per call (O(m)) to find max z-index. For n new nodes, the
total cost is O(n*m). In practice, clone/paste operations involve a
small number of nodes (typically 1-10), so this is negligible. For
extremely large pastes (100+ nodes), each call also increments the max
by 1, so z-indices will be sequential (which is actually a reasonable
stacking order).
7. **layoutStore availability**: `layoutStore` is a module-level
singleton (`new LayoutStoreImpl()`) -- not a Pinia store -- so it is
always available. The `useLayoutMutations()` composable is a plain
function returning an object of closures over `layoutStore`. It does not
require Vue component context. No risk of runtime errors.
8. **Legacy renderer (non-Vue mode)**: When Node 2.0 mode is disabled,
the layout store still exists but is not used for rendering. Calling
`bringNodeToFront` will update z-index values in the Yjs document that
are never read. This is harmless.
## Red-Green Verification
| Commit | Result | Description |
|---|---|---|
| `6894b99` `test:` | RED | Test asserts cloned node z-index > original.
Fails with `expected 0 to be greater than 5`. |
| `3567469` `fix:` | GREEN | Calls `bringNodeToFront` for each new node
in `_deserializeItems`. Test passes. |
Fixes#10307
---------
Co-authored-by: Alexander Brown <drjkl@comfy.org>
## Summary
Extract internal logic from the 2 remaining VTU holdout components into
composables, enabling full VTL migration.
## Changes
- **What**: Extract `useProcessedWidgets` from `NodeWidgets.vue`
(486→135 LOC) and `useWidgetSelectItems`/`useWidgetSelectActions` from
`WidgetSelectDropdown.vue` (563→170 LOC). Rewrite both component test
files as composable unit tests + slim behavioral VTL tests. Remove
`@vue/test-utils` devDependency.
- **Dependencies**: Removes `@vue/test-utils`
## Review Focus
- Composable extraction is mechanical — no logic changes, just moving
code into testable units
- `useProcessedWidgets` handles widget deduplication, promotion border
styling, error detection, and identity resolution (~290 LOC)
- `useWidgetSelectItems` handles the full computed chain from widget
values → dropdown items including cloud asset mode and multi-output job
resolution (~350 LOC)
- `useWidgetSelectActions` handles selection resolution and file upload
(~120 LOC)
- 40 new composable-level unit tests replace 13 `wrapper.vm.*` accesses
across the 2 holdout files
┆Issue is synchronized with this [Notion
page](https://www.notion.so/PR-10966-refactor-extract-composables-from-VTU-holdout-components-complete-VTL-migration-33c6d73d36508148a3a4ccf346722d6d)
by [Unito](https://www.unito.io)
---------
Co-authored-by: Amp <amp@ampcode.com>
## Summary
Add eslint-plugin-playwright as an oxlint JS plugin scoped to
browser_tests/, enforcing Playwright best practices at lint time.
## Changes
- **What**: Configure eslint-plugin-playwright@2.10.1 via oxlint's alpha
`jsPlugins` field (`.oxlintrc.json` override scoped to
`browser_tests/**/*.ts`). 18 recommended rules +
`prefer-native-locators` + `require-to-pass-timeout` at error severity.
All 173 initial violations resolved (config, auto-fix, manual fixes).
`no-force-option` set to off — 28 violations need triage (canvas overlay
workarounds vs unnecessary force) in a dedicated PR.
- **Dependencies**: `eslint-plugin-playwright@^2.10.1` (devDependency,
required by oxlint jsPlugins at runtime)
## Review Focus
- `.oxlintrc.json` override structure — this is the first use of
oxlint's JS plugins alpha feature in this repo
- Manual fixes in spec files: `waitForSelector` → `locator.waitFor`,
deprecated page methods → locator equivalents, `toPass()` timeout
additions
- Compound CSS selectors replaced with `.and()` (Playwright native
locator composition) to avoid `prefer-native-locators` suppressions
- Lint script changes in `package.json` to include `browser_tests/` in
oxlint targets
---------
Co-authored-by: Amp <amp@ampcode.com>
Co-authored-by: GitHub Action <action@github.com>
## Summary
Auto-fit viewport to subgraph content on first entry so interior nodes
are immediately visible.
## Changes
- **What**: On cache miss in `restoreViewport()`, call `fitView()` via
`requestAnimationFrame` instead of silently returning. Existing
cache-hit path (revisiting a subgraph) is unchanged.
## Review Focus
The `anyItemOverlapsRect` guard in `app.ts` (workflow load path) is
intentionally **not** touched — it serves a different purpose
(respecting `extra.ds` on workflow load). This fix only affects subgraph
navigation transitions where there is no saved viewport to respect.
Fixes#8173
## Screenshots (if applicable)
N/A — viewport positioning fix; before: empty canvas on subgraph entry,
after: nodes visible.
┆Issue is synchronized with this [Notion
page](https://www.notion.so/PR-10995-fix-auto-fit-to-view-on-first-subgraph-entry-33d6d73d365081f3a9b3cc2124979624)
by [Unito](https://www.unito.io)
## Summary
Debounce the reconnecting toast with a 1.5s grace period so brief
WebSocket blips don't flash a persistent error banner.
## Problem
After #9543 made all error toasts sticky (no auto-dismiss), brief
WebSocket disconnections (<1s) would show a persistent "Reconnecting..."
error banner that stays until the `reconnected` event fires. On staging,
users see this banner even though jobs are actively executing.
## Changes
- Extract reconnection toast logic from `GraphView.vue` into
`useReconnectingNotification` composable
- Add 1.5s delay (via `useTimeoutFn` from VueUse) before showing the
reconnecting toast
- If `reconnected` fires within the delay window, suppress both the
error and success toasts entirely
- Clean up unused `useToast`/`useI18n` imports from `GraphView.vue`
## Testing
- Sub-1.5s disconnections: no toast shown
- Longer disconnections (>1.5s): error toast shown, cleared with success
toast on reconnect
- Setting `Comfy.Toast.DisableReconnectingToast`: no toasts shown at all
- Multiple rapid `reconnecting` events: only one toast shown
6 unit tests covering all scenarios.
┆Issue is synchronized with this [Notion
page](https://www.notion.so/PR-10997-fix-debounce-reconnecting-toast-to-prevent-false-positive-banner-33d6d73d3650810289e8f57c046972f1)
by [Unito](https://www.unito.io)
---------
Co-authored-by: GitHub Action <action@github.com>
Co-authored-by: Alexander Brown <drjkl@comfy.org>
## Summary
Add Playwright coverage for two previously-untested canvas gestures:
Alt+drag to duplicate a regular node, and holding Meta across
click-click-click-drag to move multiple selected Vue nodes together.
## Changes
- **What**:
- `browser_tests/tests/interaction.spec.ts` — new `Node Duplication`
sub-describe with `Can duplicate a regular node via Alt+drag`. Asserts
CLIPTextEncode count goes 2 → 3 via poll, original node still exists.
Exercises the legacy canvas path at
`src/lib/litegraph/src/LGraphCanvas.ts:2434-2458`, which was only tested
for subgraph nodes before.
- `browser_tests/tests/vueNodes/interactions/node/move.spec.ts` — new
`should move all selected nodes together when dragging one with Meta
held`. Holds Meta for the entire sequence (3× `click({ modifiers:
['Meta'] })` + drag), asserts selection stays at 3 and all three nodes
move by the same delta. Exercises
`src/renderer/extensions/vueNodes/layout/useNodeDrag.ts:54-191`.
- Small refactor: `getLoadCheckpointHeaderPos` now delegates to a
reusable `getHeaderPos(comfyPage, title)` helper. Added `deltaBetween`
and `expectSameDelta` utilities (stricter than `expectPosChanged`, which
only checks `> 0`).
┆Issue is synchronized with this [Notion
page](https://www.notion.so/PR-10994-test-add-e2e-coverage-for-alt-drag-duplicate-and-meta-multi-select-drag-33d6d73d3650812dbf15c7053f44f0fd)
by [Unito](https://www.unito.io)
---------
Co-authored-by: Alexander Brown <drjkl@comfy.org>
## Summary
Disables the `vitest/require-mock-type-parameters` oxlint rule,
eliminating all 2,813 lint warnings in the codebase.
## Details
Every warning came from this single rule requiring explicit type
parameters on `vi.fn()` calls. Investigation showed:
- **85% are bare `vi.fn()`** — no type info available without manual
cross-referencing
- The rule's auto-fixer is **declared but not implemented** — `lint:fix`
doesn't help
- No existing codemods exist for this
- A full manual sweep would take 3–5 days across ~210 test files
## Test Plan
- `pnpm lint` passes with 0 warnings, 0 errors
┆Issue is synchronized with this [Notion
page](https://www.notion.so/PR-11146-fix-disable-vitest-require-mock-type-parameters-oxlint-rule-33e6d73d36508186bf1cdc2ce6d2ba57)
by [Unito](https://www.unito.io)
Co-authored-by: Amp <amp@ampcode.com>
## Summary
Add `windowsHide: true` to the `execSync('git rev-parse HEAD')` call in
`vite.config.mts` to prevent a console window from flashing on Windows
during builds.
## Changes
- **What**: Pass `windowsHide: true` option to `execSync` when fetching
the git commit hash at build time. This suppresses the transient cmd.exe
popup that appears on Windows.
## Review Focus
Minimal, single-option change. `windowsHide` is a Node.js built-in
option for `child_process` methods — no-op on non-Windows platforms.
┆Issue is synchronized with this [Notion
page](https://www.notion.so/PR-11144-fix-vite-hide-git-rev-parse-window-on-Windows-33e6d73d365081ed9a14da5f47ccac4d)
by [Unito](https://www.unito.io)
Co-authored-by: Amp <amp@ampcode.com>
## Summary
Adds tests that simulate the execution flow and output feed
## Changes
- **What**:
- Add ExecutionHelper for mocking network activity
- Refactor ws fixture to use Playwright websocket helper instead of
patching window
-
## Review Focus
<!-- Critical design decisions or edge cases that need attention -->
<!-- If this PR fixes an issue, uncomment and update the line below -->
<!-- Fixes #ISSUE_NUMBER -->
## Screenshots (if applicable)
<!-- Add screenshots or video recording to help explain your changes -->
┆Issue is synchronized with this [Notion
page](https://www.notion.so/PR-10801-test-App-mode-Execution-tests-3356d73d365081e4acf0c34378600031)
by [Unito](https://www.unito.io)
## Summary
Fix undersized delete and edit icons on user blueprint items in the node
library sidebar.
## Changes
- **What**: Changed blueprint action icons (trash, edit) from `size-3.5`
(14px) to `size-4` (16px), matching the standard icon size used across
the codebase.
## Review Focus
Trivial sizing fix — `size-4` is the codebase-wide convention for
iconify icons in buttons, and what the button base styles default SVGs
to.
┆Issue is synchronized with this [Notion
page](https://www.notion.so/PR-10992-fix-use-standard-size-4-for-blueprint-action-icons-33d6d73d365081be8c65f9e2a7b1d6ec)
by [Unito](https://www.unito.io)
## Summary
Alright, alright, alright. These e2e tests have been runnin' around like
they're late for somethin', settin' tight little timeouts like the
world's gonna end in 250 milliseconds. Man, you gotta *breathe*. Let the
framework do its thing. Go slow to go fast, that's what I always say.
## Changes
- **What**: Removed ~120 redundant timeout overrides from auto-retrying
Playwright assertions (`toBeVisible`, `toBeHidden`, `toHaveCount`,
`toBeEnabled`, `toHaveAttribute`, `toContainText`, `expect.poll`) where
5000ms is already the default. Also removed sub-5s timeouts (1s, 2s, 3s)
that were just *begging* for flaky failures — like wearin' a belt and
suspenders and also holdin' your pants up with both hands. Raised the
absurdly short timeouts in `customMatchers.ts` (250ms `toPass` → 5000ms,
256ms poll → default). Kept `timeout: 5000` on `.toPass()` calls
(defaults to 0), `.waitFor()`, `waitForRequest`, `waitForFunction`,
intentionally-short timeouts inside retry loops, and conditional
`.isVisible()/.catch()` checks — those fellas actually need the help.
## Review Focus
Every remaining timeout in the diff is there for a *reason*. The ones on
`.toPass()` stay because that API defaults to zero — it won't retry at
all without one. The ones on `.waitFor()` and `waitForRequest` stay
because those are locator actions, not auto-retrying assertions. The
intentionally-short ones inside `toPass` retry loops
(`interaction.spec.ts`) and the negative assertions (`actionbar.spec.ts`
confirming no response arrives) — those are *supposed* to be tight.
The short timeouts on regular assertions were actively *encouragin'*
flaky failures. That's like settin' your alarm for 4 AM and then gettin'
mad you're tired. Just... don't do that, man. Let things take the time
they need.
38 files, net -115 lines. Less code, more chill. That's livin'.
---------
Co-authored-by: Amp <amp@ampcode.com>
## Problem
When a node with DOM widget overlays (e.g. CLIPTextEncode) is collapsed,
the overlay elements can intercept pointer events intended for the
canvas collapse toggler, making click-to-expand unreliable.
## Root Cause
`updateWidgets()` runs during `onDrawForeground` (canvas render cycle)
and sets `widgetState.visible = false` for collapsed nodes. `v-show`
then hides the element with `display: none`. However, there is a timing
gap between the canvas state change and Vue's DOM update — during this
gap the widget overlay still intercepts pointer events.
## Fix
Add `!widgetState.visible` to the `pointerEvents` condition in
`composeStyle()`. This immediately sets `pointer-events: none` when the
widget becomes invisible, preventing event interception before `v-show`
applies `display: none`.
Also restores click-to-expand in the E2E test, removing the programmatic
`node.collapse()` workaround from PR #10967.
- Fixes#11006
┆Issue is synchronized with this [Notion
page](https://www.notion.so/PR-11063-fix-disable-pointer-events-on-non-visible-DOM-widget-overlays-33e6d73d36508179a83cd47121cf933f)
by [Unito](https://www.unito.io)
## Summary
Three issues caused GLSL preview to diverge from backend results:
1. Uniform source resolution always read widgets[0] instead of using
link.origin_slot to select the correct widget. Added directValue
fallback for widgets not registered in widgetValueStore.
2. Hex color strings (e.g. "#45edf5") were coerced to 0 by Number().
Added hexToInt to colorUtil and used it in toNumber coercion.
3. Custom size_mode was ignored — preview always used upstream image
dimensions. Now checks size_mode widget first and respects "custom".
┆Issue is synchronized with this [Notion
page](https://www.notion.so/PR-11010-fix-resolve-incorrect-GLSL-live-preview-for-non-primitive-widget-types-33e6d73d36508101a76bfe8383c0c6ab)
by [Unito](https://www.unito.io)
## Summary
The node "More Options" (⋮) context menu had `md:max-h-none
md:overflow-y-visible` responsive overrides that removed the height
constraint and scrollability on desktop (768px+). When the menu had many
items (e.g., KSampler), items below the viewport fold were inaccessible
with no scrollbar.
Removed the desktop overrides so `max-h-[80vh] overflow-y-auto` applies
at all screen sizes.
- Fixes#10824
## Red-Green Verification
| Commit | CI Status | Purpose |
|--------|-----------|---------|
| `test: add failing test for node context menu viewport overflow` |
🔴 Red | Proves the test catches the bug |
| `fix: prevent node context menu from overflowing viewport on desktop`
| 🟢 Green | Proves the fix resolves the bug |
## Test Plan
- [ ] CI red on test-only commit
- [ ] CI green on fix commit
- [ ] Manual verification: zoom out to 50%, open node More Options menu,
verify last item ("Remove") is scrollable
┆Issue is synchronized with this [Notion
page](https://www.notion.so/PR-10854-fix-prevent-node-context-menu-from-overflowing-viewport-on-desktop-3396d73d365081989403c981847aeda6)
by [Unito](https://www.unito.io)
## Summary
- Add Playwright E2E tests for the ManagerDialog component which had
zero test coverage
- Covers dialog opening, pack browsing, search filtering, tab
navigation, sort controls, search mode switching, error states, install
action, info panel, and close behavior
- All API calls (Algolia search, Manager installed packs, queue) are
mocked with typed responses
Part of the FixIt Burndown test coverage initiative.
## Test plan
- [ ] CI browser tests pass
- [ ] Tests validate core ManagerDialog user flows with mocked APIs
- [ ] No regressions in existing tests
┆Issue is synchronized with this [Notion
page](https://www.notion.so/PR-10970-test-add-E2E-tests-for-ManagerDialog-33c6d73d365081468a4ad8fc1894f05b)
by [Unito](https://www.unito.io)
Adds `browser_tests/tests/canvasModeSelector.spec.ts`, covering the
canvas toolbar mode-selector component that was introduced with no E2E
coverage.
Covers:
- Trigger button: toolbar visibility, `aria-expanded` state, icon
reflects active mode
- Popover lifecycle: open on click, close on re-click / item selection /
Escape
- Mode switching: clicking Hand/Select drives `canvas.state.readOnly`;
clicking the active item is a no-op
- ARIA state: `aria-checked` and roving `tabindex` track active mode,
including state driven by external commands
- Keyboard navigation: ArrowDown/Up with wraparound, Escape restores
focus to trigger — all using `toBeFocused()` retrying assertions
- Focus management: popover auto-focuses the checked item on open
- Keybinding integration: `H` / `V` keys update both
`canvas.state.readOnly` and the trigger icon
- Shortcut hint display: both menu items render non-empty key-sequence
hints
22 tests across 7 `describe` groups. All selectors are ARIA-driven.
┆Issue is synchronized with this [Notion
page](https://www.notion.so/PR-10934-test-add-E2E-tests-for-CanvasModeSelector-toolbar-component-33b6d73d3650819cb2cfdca22bf0b9a5)
by [Unito](https://www.unito.io)
## Description
Fixes the WebcamCapture image upload path construction that was still
broken on cloud environments after #10220.
### Root cause
The cloud `/upload/image` endpoint returns:
```json
{ "name": "hash.png", "subfolder": "", "type": "input" }
```
The previous fix used `??` (nullish coalescing), which doesn't catch
empty strings:
- `subfolder: ""` → `"" ?? "webcam"` = `""` → path becomes `/hash.png`
(wrong)
- `type` was hardcoded as `[temp]` but cloud stores as `input` → file
not found
### Fix
- `??` → `||` so empty strings fall back to defaults
- Use `data.type` from server response instead of hardcoding `[temp]`
### QA evidence
Prod (cloud/1.42): `ImageDownloadError: the input file
'webcam/1775685296883.png [temp]' doesn't exist`
Staging (cloud/1.42): `ImageDownloadError: Failed to validate images`
### Related
- Fixes the remaining issue from #10220
┆Issue is synchronized with this [Notion
page](https://www.notion.so/PR-11000-fix-use-instead-of-and-server-type-in-WebcamCapture-upload-path-33d6d73d36508156b93cfce0aae8e017)
by [Unito](https://www.unito.io)
## Automated Ingest API Type Update
This PR updates the Ingest API TypeScript types and Zod schemas from the
latest cloud OpenAPI specification.
- Cloud commit: 48d94b7
- Generated using @hey-api/openapi-ts with Zod plugin
These types cover cloud-only endpoints (workspaces, billing, secrets,
assets, tasks, etc.).
Overlapping endpoints shared with the local ComfyUI Python backend are
excluded.
---------
Co-authored-by: MillerMedia <7741082+MillerMedia@users.noreply.github.com>
Co-authored-by: GitHub Action <action@github.com>
## Summary
Replace Zendesk feedback URLs with Typeform survey (`q7azbWPi`) in the
action bar feedback button and Help Center menu for Cloud/Nightly
distributions.
## Changes
- **What**:
- `cloudFeedbackTopbarButton.ts`: Replace `buildFeedbackUrl()` (Zendesk)
with direct Typeform survey URL. Remove unused Zendesk import.
- `HelpCenterMenuContent.vue`: Feedback menu item now opens Typeform URL
for Cloud/Nightly builds; falls back to `Comfy.ContactSupport` (Zendesk)
for other distributions. Added external link icon for Cloud/Nightly.
- Help menu item and `Comfy.ContactSupport` command unchanged — support
flows still route to Zendesk.
## Review Focus
- Gating logic: `isCloud || isNightly` correctly limits Typeform
redirect to intended distributions
- Help item intentionally unchanged (support ≠ feedback)
Ticket: COM-17992
┆Issue is synchronized with this [Notion
page](https://www.notion.so/PR-10890-feat-integrate-Typeform-survey-into-feedback-button-33a6d73d36508185abbfe57e7a36b5f7)
by [Unito](https://www.unito.io)
## Summary
In cloud mode, asset-supported nodes (e.g. CheckpointLoaderSimple) used
the server's `object_info` combo options as their default widget value.
These options list local files on the backend which may not exist in the
user's cloud asset library. When the missing-model pipeline runs (on
undo, reload, or tab switch), it checks widget values against cloud
assets and correctly flags these local-only files as missing — producing
errors that appear to be false positives but are actually valid
detections of unusable defaults.
This PR changes the default value source from server combo options to
the cloud assets store.
## Default Value Behavior (Before → After)
### Cloud + asset-supported widgets (changed)
| Condition | Before | After |
|-----------|--------|-------|
| Assets cached, `inputSpec.default` in assets | `inputSpec.default` |
`inputSpec.default` |
| Assets cached, `inputSpec.default` not in assets | `inputSpec.default`
| `assets[0]` |
| Assets cached, no `inputSpec.default`, `options` exist | `options[0]`
| `assets[0]` |
| Assets not cached, `inputSpec.default` exists | `inputSpec.default` |
`undefined` → "Select model" |
| Assets not cached, no `inputSpec.default`, `options` exist |
`options[0]` | `undefined` → "Select model" |
| Assets not cached, no `inputSpec.default`, no `options` | `undefined`
→ "Select model" | `undefined` → "Select model" |
### Cloud + non-asset widgets (unchanged)
| Condition | Behavior |
|-----------|----------|
| `inputSpec.default` exists | `inputSpec.default` |
| `options` exist | `options[0]` |
| `remote` input | `"Loading..."` |
| None | `undefined` |
### OSS (unchanged)
| Condition | Behavior |
|-----------|----------|
| `inputSpec.default` exists | `inputSpec.default` |
| `options` exist | `options[0]` |
| `remote` input | `"Loading..."` |
| None | `undefined` |
## Root Cause
1. `addComboWidget` called `getDefaultValue(inputSpec)` which returns
`inputSpec.options[0]` — a local file from `object_info`
2. In cloud mode, `shouldUseAssetBrowser()` creates an asset widget with
this local filename as default
3. The model (e.g.
`dynamicrafter/controlnet/dc-sketch_encoder_fp16.safetensors`) exists on
the server but not in the user's cloud asset library
4. On undo/reload, `verifyAssetSupportedCandidates()` checks the widget
value against cloud assets → not found → marked as missing
## Changes
### Production (`useComboWidget.ts`)
- New `resolveCloudDefault(nodeType, specDefault)` function encapsulates
cloud default resolution
- Default priority: `inputSpec.default` (if found in cloud assets) →
first cloud asset → `undefined` (shows "Select model" placeholder)
- Edge case guards: `!= null` check for falsy defaults, `|| undefined`
for empty `getAssetFilename` return
- Server combo options (`object_info`) are no longer used as defaults
for asset widgets
### Unit Tests (`useComboWidget.test.ts`)
- 6 scenarios covering all default value paths:
- Cloud assets loaded, no `inputSpec.default` → `assets[0]`
- Cloud assets loaded, `inputSpec.default` in assets → uses default
- Cloud assets loaded, `inputSpec.default` not in assets → `assets[0]`
- No cloud assets, with `inputSpec.default` → placeholder
- No cloud assets, with server options → placeholder
- Asset widget creation verification
- Test helper refactored: assertions moved from helper to each test for
clarity
### E2E Test (`cloud-asset-default.spec.ts`)
- New `@cloud` tagged test verifying CheckpointLoaderSimple uses first
cloud asset, not server default
- Fixture extension stubs `/api/assets` before app loads (local backend
returns 503 for this endpoint)
- Uses typed mock data from existing `assetFixtures.ts`
## Scope
- **Cloud only**: All changes gated behind `isCloud` +
`shouldUseAssetBrowser()`
- **OSS impact**: None — code path is not entered in non-cloud builds
- **Breaking changes**: None — `useComboWidget` export signature
unchanged
## Review Focus
- Should the `/api/assets` stub in the E2E fixture extension be moved
into `ComfyPage` for all `@cloud` tests?
## Record
Before
https://github.com/user-attachments/assets/994162a0-b56a-4e84-9b1c-d0f0068196d5
After
https://github.com/user-attachments/assets/ba299990-9bd3-4565-bd09-bffac3db60a9
## Summary
- Fix `CustomCombo` copy/paste so the combo keeps its option list and
selected value
- Scope the fix to `src/extensions/core/customWidgets.ts` instead of
changing LiteGraph core deserialization
- Replace the previous round-trip test with a regression test that
exercises the actual clone/paste lifecycle
- Fixes#9927
## Root Cause
`CustomCombo` option widgets override `value` to read from
`widgetValueStore`.
During `node.clone()` and clipboard paste, `configure()` restores widget
values before the new node is added to the graph and before those
widgets are registered in the store.
That meant the option widgets read back as empty while `updateCombo()`
was rebuilding the combo state, so `comboWidget.options.values` became
blank on the pasted node.
## Fix
Keep a local fallback value for each generated `option*` widget in
`customWidgets.ts`.
The getter now returns the store-backed value when available and falls
back to the locally restored value before store registration.
This preserves the option list during `clone().serialize()` and paste
without hard-coding `CustomCombo` behavior into
`LGraphNode.configure()`.
## Why No E2E Test
This regression happens in the internal LiteGraph clipboard lifecycle:
`clone() -> serialize() -> createNode() -> configure() -> graph.add()`.
The failing state is the transient pre-add relationship between
`CustomCombo`'s store-backed option widgets and
`comboWidget.options.values`, which is not directly exposed through a
stable DOM assertion in the current Playwright suite.
A focused unit regression test is the most direct way to cover that
lifecycle without depending on brittle canvas interaction timing.
## Test Plan
- [x] Regression test covers `clone().serialize() -> createNode() ->
configure() -> graph.add()` for `CustomCombo`
- [ ] CI on the latest two commits (`81ac6d2ce`, `94147caf1`)
- [ ] Manual: create `CustomCombo` -> add `alpha`, `beta`, `gamma` ->
select `beta` -> copy/paste -> verify the pasted combo still shows all
three options and keeps `beta` selected
## Summary
<!-- One sentence describing what changed and why. -->
## Changes
- **What**: <!-- Core functionality added/modified -->
- **Breaking**: <!-- Any breaking changes (if none, remove this line)
-->
- **Dependencies**: <!-- New dependencies (if none, remove this line)
-->
## Review Focus
<!-- Critical design decisions or edge cases that need attention -->
<!-- If this PR fixes an issue, uncomment and update the line below -->
<!-- Fixes #ISSUE_NUMBER -->
## Screenshots (if applicable)
<!-- Add screenshots or video recording to help explain your changes -->
┆Issue is synchronized with this [Notion
page](https://www.notion.so/PR-10157-feat-website-add-zh-CN-translations-for-homepage-and-secondary-pages-3266d73d3650811f918cc35eca62a4bc)
by [Unito](https://www.unito.io)
---------
Co-authored-by: GitHub Action <action@github.com>
## Summary
Consolidate the `SubscriptionTier` type from 3 independent definitions
into a single source of truth in `tierPricing.ts`.
## Changes
- **What**: Exported `SubscriptionTier` from `tierPricing.ts`. Removed
hand-written unions from `workspaceApi.ts` (lines 80-88),
`PricingTable.vue`, and `PricingTableWorkspace.vue`. All now import from
the canonical location.
- **Files**: 4 files changed (type-only, ~5 net lines)
## Review Focus
- This is a type-only change — `pnpm typecheck` is the primary
validation
- If the OpenAPI schema ever adds tiers, there is now one place to
update
## Stack
PR 5/5: #10483 → #10484 → #10485 → #10486 → **→ This PR**
## Summary
Decompose the 562-line `MembersPanelContent.vue` into focused
single-responsibility components.
## Changes
- **What**: Extracted `RoleBadge.vue`, `MemberListItem.vue`,
`PendingInvitesList.vue`, and `MemberUpsellBanner.vue` from
`MembersPanelContent.vue`. Added `RoleBadge.test.ts`. The parent
component is slimmed from 562 → ~120 lines.
- **Files**: 6 files changed (4 new components + 1 new test + 1
refactored)
## Review Focus
- Component boundaries — each extracted component has a clear single
responsibility
- `MembersPanelContent.vue` still orchestrates all behavior; extracted
components are presentational
- Visual QA needed: workspace settings panel should look and behave
identically
## Stack
PR 4/5: #10483 → #10484 → #10485 → **→ This PR** → #10487
---------
Co-authored-by: Alexander Brown <drjkl@comfy.org>
Co-authored-by: GitHub Action <action@github.com>
## Summary
Removes hardcoded `version: 10` from `pnpm/action-setup` and instead
injects the `packageManager` field into `package.json` when absent
(legacy `core/*` branches).
## Why
PR #10952 re-added `version: 10` to fix old branches lacking
`packageManager`. But `main` now has **both** `version: 10` (workflow)
and `packageManager: pnpm@10.33.0` (`package.json`), causing
`pnpm/action-setup` to error with:
> Multiple versions of pnpm specified
Failed run:
https://github.com/Comfy-Org/ComfyUI_frontend/actions/runs/24158869559
This fix handles both cases:
- **`main`**: has `packageManager` → action reads it directly, no
conflict
- **`core/1.42` etc**: missing `packageManager` → step injects it before
the action runs
E2E test not applicable — this is a CI workflow configuration change
with no user-facing behavior.
┆Issue is synchronized with this [Notion
page](https://www.notion.so/PR-10972-fix-ci-resolve-pnpm-version-conflict-in-version-bump-workflow-33c6d73d36508112802df75c0dd5ea50)
by [Unito](https://www.unito.io)
Closes#9226
## Summary
Image uploads had no timeout or abort mechanism, meaning a stalled
upload could hang indefinitely with no user feedback. This adds a
2-minute timeout using `AbortController` and shows a user-friendly toast
message when the upload times out.
## Changes
- `src/composables/node/useNodeImageUpload.ts`: Added `AbortController`
with a 120-second timeout to the `uploadFile` function. The abort signal
is passed to `fetchApi`. In the `handleUpload` error handler,
`AbortError` is now caught separately to display a localized timeout
message.
- `src/locales/en/main.json`: Added `uploadTimedOut` i18n translation
key.
---
Automated by coderabbit-fixer
┆Issue is synchronized with this [Notion
page](https://www.notion.so/PR-9491-fix-Add-timeout-and-abort-mechanism-for-image-upload-9226-31b6d73d365081d7a7d7f7016f3a71c6)
by [Unito](https://www.unito.io)
---------
Co-authored-by: CodeRabbit Fixer <coderabbit-fixer@automated.bot>
Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>
Co-authored-by: Christian Byrne <cbyrne@comfy.org>
## Summary
Adds Playwright E2E tests for the ShareWorkflow dialog component and its
various states.
## Tests added
- Dialog opens and shows unsaved state for new workflows (save prompt)
- Ready state shows create link button for saved but unpublished
workflows
- Shared state shows copy URL field with share link after publishing
- Stale state shows update link button when workflow modified after
publishing
- Close button dismisses dialog
- Create link transitions dialog from ready to shared state
- Tab switching between share link and publish to hub (when
comfyHubUploadEnabled)
- Tab aria-selected states update correctly on switch
## Approach
- Share dialog is gated behind `isCloud` (compile-time constant), so
tests invoke it directly via `page.evaluate()` importing
`useShareDialog`
- Share service API calls (`/api/userdata/*/publish`,
`/api/assets/from-workflow`) mocked via `page.route()` for deterministic
state testing
- Dialog state (loading → unsaved → ready → shared → stale) controlled
by mock responses
- Feature flags set via `serverFeatureFlags.value` for tab visibility
testing
## Notes
- All pre-existing TS2307 errors are `.vue` module resolution — no new
type errors
- Tests cover the 5 dialog states: loading, unsaved, ready, shared,
stale
## Task
Part of Test Coverage Q2 Overhaul (DLG-05).
## Conventions
- Uses Vue nodes with new menu enabled
- Tests read as user stories
- No full-page screenshots
- Proper waits, no sleeps
- All API calls mocked
┆Issue is synchronized with this [Notion
page](https://www.notion.so/PR-10588-test-add-ShareWorkflow-dialog-E2E-tests-DLG-05-3306d73d365081a0ab15f333707e493b)
by [Unito](https://www.unito.io)
---------
Co-authored-by: GitHub Action <action@github.com>
## Summary
Add design standards instructions so agents consult the canonical [Comfy
Design
Standards](https://www.figma.com/design/QreIv5htUaSICNuO2VBHw0/Comfy-Design-Standards)
Figma file before implementing user-facing features.
### Changes
- **`docs/guidance/design-standards.md`** — Auto-loaded guidance for
`src/components/**/*.vue` and `src/views/**/*.vue` with Figma MCP fetch
instructions, section node IDs, and component set references
- **`AGENTS.md`** — Added Design Standards section and Figma link in
External Resources
### Design
All content is fetched **live from Figma** via the Figma MCP tool —
designers can update the file and agents will always see the latest
version. No hardcoded design rules that can go stale.
Ref: COM-17639
┆Issue is synchronized with this [Notion
page](https://www.notion.so/PR-10696-feat-add-Comfy-Design-Standards-Figma-reference-for-agents-3326d73d36508181844fdcaa5c17cf00)
by [Unito](https://www.unito.io)
---------
Co-authored-by: GitHub Action <action@github.com>
## Summary
Add a "Testing with Cloud & Staging Environments" section to
CONTRIBUTING.md documenting how to test partner/API nodes that require
cloud backend authentication.
## Changes
- **What**: New section in CONTRIBUTING.md between "Dev Server" and
"Access dev server on touch devices" explaining two approaches for
staging/cloud development:
1. Frontend approach: `pnpm dev:cloud` or custom
`DEV_SERVER_COMFYUI_URL` in `.env`
2. Backend approach: `--comfy-api-base https://stagingapi.comfy.org`
flag
## Review Focus
- Accuracy of the `--comfy-api-base` backend flag documentation (sourced
from internal Slack discussion)
- Whether the section placement and level of detail is appropriate
┆Issue is synchronized with this [Notion
page](https://www.notion.so/PR-10775-docs-add-staging-environment-setup-instructions-to-CONTRIBUTING-md-3346d73d36508112bcd4df6ecb7f83c6)
by [Unito](https://www.unito.io)
## Summary
- Adds Playwright E2E tests for the Settings dialog covering behaviors
not tested elsewhere:
- About panel displays version badges (navigates to About, verifies
content)
- Boolean setting toggle through the UI persists the value (uses search
+ ToggleSwitch click, verifies via API)
- Dialog can be closed via the close button (complements existing
Escape-key test)
## Test plan
- [ ] `pnpm test:browser:local -- --grep "Settings dialog"` passes
- [ ] No overlap with existing tests in `dialog.spec.ts` or
`useSettingSearch.spec.ts`
┆Issue is synchronized with this [Notion
page](https://www.notion.so/PR-10797-test-add-E2E-tests-for-settings-dialog-3356d73d365081e4a3adcd3979048444)
by [Unito](https://www.unito.io)
## Summary
Complete the @e2e/ path alias migration started in #10735 by converting
all 354 remaining relative imports and adding a lint rule to prevent
backsliding.
## Changes
- **What**: Migrate all relative imports in browser_tests/ to use
`@e2e/` (intra-directory) and `@/` (src/ imports) path aliases. Add
`no-restricted-imports` ESLint rule banning `./` and `../` imports in
`browser_tests/**/*.ts`. Suppress pre-existing oxlint `no-eval` and
`no-console` warnings exposed by touching those files.
## Review Focus
- ESLint flat-config merging: the `@playwright/test` ban and
relative-import ban are in two separate blocks to avoid last-match-wins
collision with the `useI18n`/`useVirtualList` blocks higher in the
config.
- The `['./**', '../**']` glob patterns (not `['./*', '../*']`) are
needed to catch multi-level relative paths like `../../../src/foo`.
Follows up on #10735
┆Issue is synchronized with this [Notion
page](https://www.notion.so/PR-10958-test-migrate-browser_tests-to-e2e-path-alias-and-add-lint-rule-33c6d73d365081649d1be771eac986fd)
by [Unito](https://www.unito.io)
Co-authored-by: Amp <amp@ampcode.com>
## Summary
Comprehensive Playwright E2E tests for the error systems: ErrorDialog,
ErrorOverlay, and the errors tab (missing nodes, models, media,
execution errors).
## Changes
- **What**:
- **ErrorDialog** (`errorDialog.spec.ts`, 7 tests): configure/prompt
error triggers, Show Report, Copy to Clipboard, Find Issues on GitHub,
Contact Support
- **ErrorOverlay** (`errorOverlay.spec.ts`, 12 tests): error count
labels, per-type button labels (missing nodes/models/media/multiple),
See Errors flow (open panel, dismiss, close), undo/redo persistence
- **Errors tab — common** (`errorsTab.spec.ts`, 3 tests): tab
visibility, search/filter execution errors
- **Errors tab — Missing nodes** (`errorsTabMissingNodes.spec.ts`, 5
tests): MissingNodeCard, packs group, expand/collapse, locate button
- **Errors tab — Missing models** (`errorsTabMissingModels.spec.ts`, 6
tests): group display, model name, expand/referencing nodes, clipboard
copy, OSS Copy URL/Download buttons
- **Errors tab — Missing media** (`errorsTabMissingMedia.spec.ts`, 7
tests): migrated from `missingMedia.spec.ts` with detection,
upload/library/cancel flows, locate
- **Errors tab — Execution** (`errorsTabExecution.spec.ts`, 2 tests):
Find on GitHub/Copy buttons, runtime error panel
- **Shared helpers**: `ErrorsTabHelper.ts` (openErrorsTabViaSeeErrors),
`clipboardSpy.ts` (interceptClipboardWrite/getClipboardText)
- **Component changes**: added `data-testid` to
`ErrorDialogContent.vue`, `FindIssueButton.vue`, `MissingModelRow.vue`,
`MissingModelCard.vue`
- **Selectors**: registered all new test IDs in `selectors.ts`
- **Test assets**: `missing_nodes_and_media.json` (compound errors),
`missing_models_with_nodes.json` (expand/locate)
- **Migrations**: error tests from `dialog.spec.ts` → dedicated files,
`errorOverlaySeeErrors.spec.ts` → `errorOverlay.spec.ts`,
`missingMedia.spec.ts` → `errorsTabMissingMedia.spec.ts`
## Review Focus
- OSS tests (`@oss` tag) verify Download/Copy URL buttons appear for
models with embedded URLs.
- The `missing_models.json` fixture must remain without nodes — adding
`CheckpointLoaderSimple` nodes causes directory mismatch in
`enrichWithEmbeddedMetadata` that prevents URL enrichment. A separate
`missing_models_with_nodes.json` fixture is used for expand/locate
tests.
## Cloud tests deferred
Missing model cloud environment tests (`@cloud` tag — hidden buttons,
import-unsupported notice) are deferred to a follow-up PR. The
`comfyPage` fixture cannot bypass the Firebase auth guard in cloud
builds, causing `window.app` initialization timeout. A separate infra PR
is needed to add cloud auth bypass to the fixture.
## Bug Discovery
During testing, a bug was found where the **Locate button for missing
nodes in subgraphs fails on initial workflow load**.
`collectMissingNodes` in `loadGraphData` captures execution IDs using
pre-`configure()` JSON node IDs, but `configure()` triggers subgraph
node ID deduplication (PR #8762, always-on since PR #9510) which remaps
colliding IDs. This will be addressed in a follow-up PR.
- Fixes#10847 (tracked, fix pending in separate PR)
## Testing
- 42 new/migrated E2E tests across 8 spec files
- All OSS tests pass locally and in CI
┆Issue is synchronized with this [Notion
page](https://www.notion.so/PR-10848-test-comprehensive-E2E-tests-for-error-dialog-overlay-and-errors-tab-3386d73d36508137a5e4cec8b12fa2fa)
by [Unito](https://www.unito.io)
---------
Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
## Summary
- Upload Model dialog stays stuck in "Processing" after async download
completes because the wizard never watches for task completion
- `useUploadModelWizard.ts` sets `uploadStatus = 'processing'` but has
no watcher linking it to `assetDownloadStore.lastCompletedDownload`
- Added a `watch` on `lastCompletedDownload` that transitions
`uploadStatus` to `'success'` when the tracked task finishes
- Fixes#10609
## Root Cause
`uploadModel()` (line 249) sets `uploadStatus = 'processing'` when the
async task starts, but control flow ends there. The `assetDownloadStore`
receives WebSocket completion events and updates
`lastCompletedDownload`, but the wizard never watches this reactive
state.
## Fix
Added a `watch` inside the async branch that monitors
`assetDownloadStore.lastCompletedDownload`. When the completed task ID
matches the upload's task ID, it transitions `uploadStatus` from
`'processing'` to `'success'` and refreshes model caches. The watcher
auto-disposes after firing.
## Red-Green Verification
| Commit | CI Status | Purpose |
|--------|-----------|---------|
| `test: add failing test for upload dialog stuck in processing state` |
🔴 Red | Proves the test catches the bug |
| `fix: update upload dialog status when async download completes` |
🟢 Green | Proves the fix resolves the bug |
## Test Plan
- [x] CI red on test-only commit
- [x] CI green on fix commit
- [x] Unit test: `updates uploadStatus to success when async download
completes`
- [ ] Manual: Import model via URL → verify dialog transitions from
Processing to Success (requires `--enable-assets`)
┆Issue is synchronized with this [Notion
page](https://www.notion.so/PR-10838-fix-update-upload-dialog-status-when-async-download-completes-3376d73d365081b6a5b1d2be3804ce4b)
by [Unito](https://www.unito.io)
## Summary
- Adds E2E regression tests for the bug fixed in PR #10745 where closing
an inactive workflow tab would save the active workflow's content into
the closing tab's file
- Three test scenarios covering the full range of the bug surface:
1. Closing an unmodified inactive tab preserves both workflows
2. Closing a modified inactive tab with "Save" preserves its own content
(not the active tab's)
3. Closing an unsaved inactive tab with "Save As" preserves its own
content
## Linked Issues
- Regression coverage for #10745 / Comfy-Org/ComfyUI#13230
## Test plan
- [ ] `pnpm exec playwright test
browser_tests/tests/topbar/workflowTabSave.spec.ts` passes against the
current main (with the fix)
┆Issue is synchronized with this [Notion
page](https://www.notion.so/PR-10815-test-add-E2E-regression-tests-for-workflow-tab-save-bug-3366d73d365081eab409ed303620a959)
by [Unito](https://www.unito.io)
---------
Co-authored-by: GitHub Action <action@github.com>
## Summary
- Add E2E test verifying duplicated subgraphs maintain independent
widget values (convert CLIP node to subgraph, duplicate, set different
text in each, assert no bleed)
- Extract `clickMenuItemExact` and `openForVueNode` into `ContextMenu`
fixture for reuse across Vue node tests
- Refactor `contextMenu.spec.ts` to delegate to the new fixture methods
## Test plan
- [x] `pnpm typecheck:browser` passes
- [x] `pnpm lint` passes
- [x] New test passes locally (`pnpm test:browser:local --
browser_tests/tests/subgraph/subgraphDuplicateIndependentValues.spec.ts`)
┆Issue is synchronized with this [Notion
page](https://www.notion.so/PR-10949-test-add-E2E-test-for-subgraph-duplicate-independent-widget-values-33b6d73d3650818191c1f78ef8db4455)
by [Unito](https://www.unito.io)
## What
Follow-up to #10575. Pass `CODECOV_TOKEN` secret to codecov upload
action and add `codecov.yml` config so Codecov posts coverage diff
comments on PRs.
## Changes
- `ci-tests-unit.yaml`: add `token: ${{ secrets.CODECOV_TOKEN }}`
- `codecov.yml`: configure PR comment layout (header, diff, flags,
files)
## Manual Step Required
An admin needs to add the `CODECOV_TOKEN` secret to the repo:
1. Go to [codecov.io](https://app.codecov.io) → sign in → find
`Comfy-Org/ComfyUI_frontend` → Settings → General → copy the Repository
Upload Token
2. Go to [repo
secrets](https://github.com/Comfy-Org/ComfyUI_frontend/settings/secrets/actions)
→ New repository secret → name: `CODECOV_TOKEN`, value: the token
## Testing
Config-only change. Once the secret is added, the next PR will get a
Codecov coverage comment.
┆Issue is synchronized with this [Notion
page](https://www.notion.so/PR-10774-ci-pass-CODECOV_TOKEN-and-add-codecov-yml-for-PR-comments-3346d73d36508169bac5e61eecc94063)
by [Unito](https://www.unito.io)
---------
Co-authored-by: GitHub Action <action@github.com>
What changed
- Typed the `PUT /assets/:id` mock body in `AssetHelper` as
`AssetUpdatePayload` instead of treating it as an untyped record.
Why
- Keeps the mock aligned with the frontend update contract used by the
asset service.
- Narrows the helper without changing behavior, so follow-up typing work
can build on a smaller base.
Validation
- `pnpm typecheck`
- `pnpm typecheck:browser`
┆Issue is synchronized with this [Notion
page](https://www.notion.so/PR-10751-test-type-asset-helper-update-payload-3336d73d365081fd8f1bc0c7c49c5ddb)
by [Unito](https://www.unito.io)
## Summary
Fix regression from #10247 where template workflows (e.g. LTX2.3) loaded
with a broken viewport.
## Problem
`restoreViewport()` called `fitView()` on every cache miss via rAF. This
raced with `loadGraphData`'s own viewport restore (`extra.ds` for saved
workflows, or its own `fitView()` for templates at line 1266 of app.ts).
The second `fitView()` overwrote the correct viewport, causing templates
with subgraphs to display incorrectly.
## Fix
On cache miss, check if any nodes are already visible in the current
viewport before calling `fitView()`. If `loadGraphData` already
positioned things correctly, we don't override it. Only intervene when
the viewport is genuinely empty (first visit to a subgraph with no prior
cached state AND no loadGraphData restore).
## Review Focus
Single-file change in `subgraphNavigationStore.ts`. The visibility check
mirrors the same pattern used in `app.ts:1272-1281` where loadGraphData
itself checks for visible nodes.
## E2E Regression Test
The existing Playwright tests in
`browser_tests/tests/subgraphViewport.spec.ts` (added in #10247) already
cover viewport restoration after subgraph navigation. The specific
regression (template load viewport race) is not practically testable in
E2E because:
1. Template loading requires the backend's template API which returns
different templates per environment
2. The race condition depends on exact timing between `loadGraphData`'s
viewport restore and the rAF-deferred `restoreViewport` — Playwright
cannot reliably reproduce frame-level timing races
3. The fix is a guard condition (skip fitView if nodes visible) that
makes the behavior idempotent regardless of timing
## Alternative to #10790
This can replace the full revert in #10790 — it preserves the viewport
persistence feature while fixing the template regression.
Fixes regression from #10247
# Summary
* Model Info sidebar panel displays `asset.name` (registry name) instead
of the actual filename from `user_metadata.filename`
* Other UI components (asset cards, widgets, missing model scan)
correctly use `getAssetFilename()` which prefers
`user_metadata.filename` over `asset.name`
* One-line template fix: `{{ asset.name }}` → `{{
getAssetFilename(asset) }}`
* Fixes#10598
# Bug
`ModelInfoPanel.vue:35` used raw `asset.name` for the "File Name" field.
When `user_metadata.filename` differs from `asset.name` (e.g. registry
name vs actual path like `checkpoints/v1-5-pruned.safetensors`), users
see inconsistent filenames across the UI.
# AS-IS / TO-BE
<img width="800" height="600" alt="before-after-10598"
src="https://github.com/user-attachments/assets/15beb6c8-4bad-4ed2-9c85-6f8c7c0b6d3e"
/>
| | File Name field shows |
| :--- | :--- |
| **AS-IS** (bug) | `sdxl-lightning-4step` — raw `asset.name` (registry
display name) |
| **TO-BE** (fix) | `checkpoints/sdxl_lightning_4step.safetensors` —
`getAssetFilename(asset)` (actual file path) |
# Red-Green Verification
| Commit | CI Status | Purpose |
| :--- | :--- | :--- |
| `test: add failing test for ModelInfoPanel showing wrong filename` | 🔴
Red | Proves the test catches the bug |
| `fix: use getAssetFilename in ModelInfoPanel filename field` | 🟢 Green
| Proves the fix resolves the bug |
# Test Plan
- [x] CI red on test-only commit
- [x] CI green on fix commit
- [x] Unit test: `prefers user_metadata.filename over asset.name for
filename field`
- [ ] Manual: open Asset Browser → click a model → verify File Name in
Model Info panel matches the actual file path (requires
`--enable-assets`)
## Summary
Fix subgraph internal node positions being permanently corrupted when
entering a subgraph after a draft workflow reload. The corruption
accumulated across page refreshes, causing nodes to progressively drift
apart or compress together.
## Changes
- **What**: In the `ResizeObserver` callback
(`useVueNodeResizeTracking.ts`), node positions are now read from the
Layout Store (source of truth, initialized from LiteGraph) instead of
reverse-converting DOM screen coordinates via `getBoundingClientRect()`
+ `clientPosToCanvasPos()`. The fallback to DOM-based conversion is
retained only for nodes not yet present in the Layout Store.
## Root Cause
`ResizeObserver` was using `getBoundingClientRect()` to get DOM element
positions, then converting them to canvas coordinates via
`clientPosToCanvasPos()`. This conversion depends on the current
`canvas.ds.scale` and `canvas.ds.offset`.
During graph transitions (e.g., entering a subgraph from a draft-loaded
workflow), the canvas viewport was stale — it still had the **parent
graph's zoom level** because `fitView()` hadn't run yet (it's scheduled
via `requestAnimationFrame`). The `ResizeObserver` callback fired before
`fitView`, converting DOM positions using the wrong scale/offset, and
writing the corrupted positions to the Layout Store. The `useLayoutSync`
writeback then permanently overwrote the LiteGraph node positions.
The corruption accumulated across sessions:
1. Load workflow → enter subgraph → `ResizeObserver` writes corrupted
positions
2. Draft auto-saves the corrupted positions to localStorage
3. Page refresh → draft loads with corrupted positions → enter subgraph
→ positions corrupted further
4. Each cycle amplifies the drift based on the parent graph's zoom level
This is the same class of bug that PR #9121 fixed for **slot** positions
— the DOM→canvas coordinate conversion is inherently fragile during
viewport transitions. This PR applies the same principle to **node**
positions.
## Why This Only Affects `main` (No Backport Needed)
This bug requires two features that only exist on `main`, not on
`core/1.41` or `core/1.42`:
1. **PR #10247** changed `subgraphNavigationStore`'s watcher to `flush:
'sync'` and added `requestAnimationFrame(fitView)` on viewport cache
miss. This creates the timing window where `ResizeObserver` fires before
`fitView` corrects the canvas scale.
2. **PR #6811** added hash-based subgraph auto-entry on page load, which
triggers graph transitions during the draft reload flow.
On 1.41/1.42, `restoreViewport` does nothing on cache miss (no `fitView`
scheduling), and the watcher uses default async flush — so the
`ResizeObserver` never runs with a stale viewport.
## Review Focus
- The core change is small: use `nodeLayout.position` (already in the
Layout Store from `initializeFromLiteGraph`) instead of computing
position from `getBoundingClientRect()`. This eliminates the dependency
on canvas scale/offset being up-to-date during `ResizeObserver`
callbacks.
- The fallback path (`getBoundingClientRect` → `clientPosToCanvasPos`)
is retained for nodes not yet in the Layout Store (e.g., first render of
a newly created node). At that point the canvas transform is stable, so
the conversion is safe.
- Unit tests updated to reflect that position is no longer overwritten
from DOM when Layout Store already has the position.
- E2E test added: load subgraph workflow → enter subgraph → reload
(draft) → verify positions preserved.
## E2E Test Fixes
- `subgraphDraftPositions.spec.ts`: replaced `comfyPage.setup({
clearStorage: false })` with `page.reload()` + explicit draft
persistence polling. The `setup()` method performs a full navigation via
`goto()` which bypassed the draft auto-load flow.
- `SubgraphHelper.packAllInteriorNodes`: replaced `canvas.click()` with
`dispatchEvent('pointerdown'/'pointerup')`. The position fix places
subgraph nodes at their correct locations, which now overlap with DOM
widget textareas that intercept pointer events.
## Test Plan
- [x] Unit tests pass (`useVueNodeResizeTracking.test.ts`)
- [x] E2E test: `subgraphDraftPositions.spec.ts` — draft reload
preserves subgraph node positions
- [x] Manual: load workflow with subgraph, zoom in/out on root graph,
enter subgraph, verify no position drift
- [x] Manual: repeat with page refresh (draft reload) — positions should
be stable across reloads
- [x] Manual: drag nodes inside subgraph — positions should update
correctly
- [x] Manual: create new node inside subgraph — position should be set
correctly (fallback path)
## Screenshots
Before
<img width="1331" height="879" alt="스크린샷 2026-04-03 오전 3 56 48"
src="https://github.com/user-attachments/assets/377d1b2e-6d47-4884-8181-920e22fa6541"
/>
After
<img width="1282" height="715" alt="스크린샷 2026-04-03 오전 3 58 24"
src="https://github.com/user-attachments/assets/34528f6c-0225-4538-9383-227c849bccad"
/>
┆Issue is synchronized with this [Notion
page](https://www.notion.so/PR-10828-fix-prevent-subgraph-node-position-corruption-during-graph-transitions-3366d73d365081418502dbb78da54013)
by [Unito](https://www.unito.io)
---------
Co-authored-by: Alexander Brown <drjkl@comfy.org>
## Summary
Update backport-management skill with learnings from the 2026-04-06
backport session (29 PRs across core/1.42, cloud/1.42, core/1.41,
cloud/1.41).
## Changes
- **What**: Captures operational learnings into the backport skill
reference docs
- Branch scope clarification: app mode and Firebase auth go to both
core+cloud branches, not cloud-only
- Accept-theirs regex warning: produces broken hybrids on component
rewrites (PrimeVue to Reka UI migrations); use `git show SHA:path`
instead
- Missing dependency pattern: cherry-picks can silently bring in code
referencing composables/components not on the target branch
- Fix PRs are expected: plan for 1 fix PR per branch after wave
verification
- `--no-verify` required for worktree commits/pushes (husky hooks fail
in /tmp/ worktrees)
- Automation success varies wildly by branch: core/1.42 got 69%
auto-PRs, cloud/1.42 got 4%
- Test-then-resolve batch pattern for efficient handling of
low-automation branches
- Slack-compatible final deliverables: plain text format replacing
mermaid diagrams (no emojis, tables, headers, or bold)
- Updated conflict triage table with component rewrite, import-only, and
locale/JSON conflict types
- Interactive approval flow replacing static decisions.md for human
review
## Review Focus
Documentation-only change to internal skill files. No production code
affected.
┆Issue is synchronized with this [Notion
page](https://www.notion.so/PR-10927-docs-update-backport-management-skill-with-v1-43-session-learnings-33a6d73d3650811aa7cffed4b2d730b0)
by [Unito](https://www.unito.io)
---------
Co-authored-by: GitHub Action <action@github.com>
## Summary
Removes `api-update-registry-api-types.yaml` — the workflow that cloned
the private `comfy-api` repo using a PAT to generate registry API types.
This is a security risk: a PAT with private repo access stored on a
public repo.
Type generation now happens in `comfy-api` and pushes PRs to this repo
instead.
### Action needed after merge
- Remove the `COMFY_API_PAT` secret from this repo's settings (Settings
→ Secrets → Actions)
### Depends on
- https://github.com/Comfy-Org/comfy-api/pull/937 (must merge first)
- Refs: COM-16785
┆Issue is synchronized with this [Notion
page](https://www.notion.so/PR-9957-chore-remove-pull-model-registry-type-gen-workflow-3246d73d365081cd9dbad03597817f05)
by [Unito](https://www.unito.io)
## Summary
<!-- One sentence describing what changed and why. -->
## Changes
- **What**: <!-- Core functionality added/modified -->
- **Breaking**: <!-- Any breaking changes (if none, remove this line)
-->
- **Dependencies**: <!-- New dependencies (if none, remove this line)
-->
## Review Focus
<!-- Critical design decisions or edge cases that need attention -->
<!-- If this PR fixes an issue, uncomment and update the line below -->
<!-- Fixes #ISSUE_NUMBER -->
## Screenshots (if applicable)
<!-- Add screenshots or video recording to help explain your changes -->
┆Issue is synchronized with this [Notion
page](https://www.notion.so/PR-10156-feat-website-add-SEO-sitemap-redirects-CI-workflow-and-Vercel-config-3266d73d3650816ab9eaebd11072d481)
by [Unito](https://www.unito.io)
---------
Co-authored-by: GitHub Action <action@github.com>
## Summary
Add dedicated auth token priority tests and reconcile workspace token
resolution between `authStore` and `workspaceAuthStore`.
## Changes
- **What**: Created `src/stores/__tests__/authTokenPriority.test.ts`
with 6 scenarios covering the full priority chain: workspace token →
Firebase token → API key → null. Added `getWorkspaceToken()` method to
`workspaceAuthStore`. Replaced raw `sessionStorage` reads in `authStore`
with proper store delegation.
- **Files**: 3 files changed (+ 1 new test file)
## Review Focus
- Token priority chain in `authTokenPriority.test.ts` — validates
workspace > Firebase > API key ordering
- The `getAuthToken()` and `getAuthHeader()` methods now delegate to
`workspaceAuthStore` instead of reading sessionStorage directly
## Stack
PR 3/5: #10483 → #10484 → **→ This PR** → #10486 → #10487
Co-authored-by: Alexander Brown <drjkl@comfy.org>
## Summary
Document acceptable vs avoidable `page.evaluate` patterns in Playwright
E2E tests, with migration candidates for existing offenders.
## Changes
- **What**: Add "When to Use `page.evaluate`" section to
`docs/guidance/playwright.md` with acceptable/avoid/preferred guidance
and 8 migration candidates identified from audit
## Review Focus
Whether the migration candidate list covers the right tests and whether
the acceptable/avoid boundary is drawn correctly.
┆Issue is synchronized with this [Notion
page](https://www.notion.so/PR-10658-docs-document-when-to-use-page-evaluate-vs-user-actions-in-browser-tests-3316d73d36508186be90f263f36daf75)
by [Unito](https://www.unito.io)
## Summary
Replace remaining inline `getByRole('dialog')` calls in E2E tests with
page object accessors, following the pattern from PR #10586.
**Note:** The original scope covered 9 calls across 5 files. Three of
those refactors (ConfirmDialog, MediaLightbox, TemplatesDialog) were
already merged into main via other PRs, so this PR now covers only the
remaining 3 files.
## Changes
- **ComfyNodeSearchBox.ts**: Extract a `root` locator on
`ComfyNodeSearchFilterSelectionPanel` so `header` getter uses
`this.root` instead of an inline `page.getByRole('dialog')`
- **AppModeHelper.ts**: Add `imagePickerPopover` getter that locates the
PrimeVue Popover (`role="dialog"`) filtered by a button named "All"
- **appModeDropdownClipping.spec.ts**: Replace 4-line inline
`getByRole('dialog')` chain with `comfyPage.appMode.imagePickerPopover`
## Review Focus
Pure test infrastructure refactor — no production code changes. Each
page object follows existing conventions.
Fixes#10723
Merge email dataLayer push event into existing signup/login push. Hash
with SHA 256 and deduplicate with a util.
AI summary below
---
## Summary
- attach hashed auth email to the `login` / `sign_up` GTM event payload
instead of pushing a separate standalone `user_data` message
- add a shared telemetry email hashing utility and reuse it for both GTM
(`SHA-256`) and Impact (`SHA-1`)
- extend tests to cover merged auth payloads, both hash algorithms, and
fallback behavior when Web Crypto is unavailable
## Why
The current GTM auth flow was splitting `user_data` and the auth event
into separate dataLayer pushes, which makes GTM preview harder to
interpret and relies on carried-over dataLayer state instead of the
event payload that actually fired.
There was also an implementation gap: the previous GTM change was
intended to hash email client-side, but the provider was still sending
normalized plaintext email into `dataLayer`.
## Impact
Auth events now carry hashed email on the same event payload that GTM
tags consume, so the preview timeline is cleaner and downstream matching
can rely on the auth event directly.
┆Issue is synchronized with this [Notion
page](https://www.notion.so/PR-10778-codex-merge-hashed-auth-user-data-into-GTM-auth-events-3346d73d36508197b57deada345dbeaa)
by [Unito](https://www.unito.io)
## Summary
<!-- One sentence describing what changed and why. -->
## Changes
- **What**: <!-- Core functionality added/modified -->
- **Breaking**: <!-- Any breaking changes (if none, remove this line)
-->
- **Dependencies**: <!-- New dependencies (if none, remove this line)
-->
## Review Focus
<!-- Critical design decisions or edge cases that need attention -->
<!-- If this PR fixes an issue, uncomment and update the line below -->
<!-- Fixes #ISSUE_NUMBER -->
## Screenshots (if applicable)
<!-- Add screenshots or video recording to help explain your changes -->
┆Issue is synchronized with this [Notion
page](https://www.notion.so/PR-10145-feat-website-add-Wave-4-secondary-pages-3266d73d3650818c9101c7d2086c21ba)
by [Unito](https://www.unito.io)
## Summary
Add E2E Playwright tests for the asset delete confirmation dialog flow.
## Changes
- **What**: New `Assets sidebar - delete confirmation` describe block in
`assets.spec.ts` covering right-click delete showing confirmation
dialog, confirming delete removes asset with success toast, and
cancelling delete preserves asset. Added `mockDeleteHistory()` to
`AssetsHelper` to intercept POST `/api/history` delete payloads and
update mock state.
## Review Focus
Tests use existing `ConfirmDialog` page object and `AssetsHelper` mock
infrastructure. The `mockDeleteHistory` handler removes jobs from the
in-memory mock array so subsequent `/api/jobs` fetches reflect the
deletion.
Fixes#10781
┆Issue is synchronized with this [Notion
page](https://www.notion.so/PR-10785-test-assets-add-E2E-tests-for-delete-confirmation-flow-3356d73d365081fb90c8e2a69de3a666)
by [Unito](https://www.unito.io)
## Summary
Extract inline `getByRole('dialog')` calls across E2E tests into
reusable page objects.
## Changes
- **What**: Extract `ConfirmDialog` class from `ComfyPage.ts` into
`browser_tests/fixtures/components/ConfirmDialog.ts` with new `save`
button locator. Add `MediaLightbox` and `TemplatesDialog` page objects.
Refactor 4 test files to use these page objects instead of raw dialog
locators.
- **Skipped**: `appModeDropdownClipping.spec.ts` uses
`getByRole('dialog')` for a PrimeVue Popover (not a true dialog), left
as-is.
## Review Focus
The `ConfirmDialog.click()` method now supports a `save` action used by
`workflowPersistence.spec.ts`, which also waits for the dialog mask to
disappear and workflow service to settle.
Fixes#10723
┆Issue is synchronized with this [Notion
page](https://www.notion.so/PR-10822-refactor-test-extract-dialog-page-objects-from-inline-getByRole-usage-3366d73d365081b3bc0ee7ef0ddce658)
by [Unito](https://www.unito.io)
## Summary
Virtualize the shared job queue history list so opening the jobs panel
does not eagerly mount the full history on cloud.
## Changes
- **What**: Virtualize the shared queue history list used by the overlay
and sidebar, flatten date headers plus job rows into a single virtual
stream, and preserve hover/menu behavior with updated queue list tests.
- **Why `@tanstack/vue-virtual` instead of Reka virtualizers**: the
installed `reka-ui@2.5.0` does not expose a generic list virtualizer. It
only exposes `ListboxVirtualizer`, `ComboboxVirtualizer`, and
`TreeVirtualizer`, and those components inject `ListboxRoot`/`TreeRoot`
context and carry listbox or tree selection/keyboard semantics. The job
history UI is a flat grouped action list, not a selectable listbox or
navigable tree, so this uses the same TanStack virtualizer layer
directly without forcing the wrong semantics onto the component.
## Review Focus
Please verify the virtual row sizing and inter-group spacing behavior
across date headers and the last row in each group.
> [!TIP]
> Diff reads much cleaner through vscode's unified view with show
leading/trailing whitespace differences enabled
Linear: COM-304
https://tanstack.com/virtual/latest/docs/api/virtualizer
┆Issue is synchronized with this [Notion
page](https://www.notion.so/PR-10592-fix-virtualize-cloud-job-queue-history-list-3306d73d3650819d956bf4b2d8b59a40)
by [Unito](https://www.unito.io)
## Summary
Add missing delete and bookmark actions for user blueprints in the V2
node library sidebar, fixing parity with the V1 sidebar.
## Changes
- **What**:
- Add delete button (inline + context menu) for user blueprints in
`TreeExplorerV2Node` and `TreeExplorerV2`
- Extract `isUserBlueprint()` helper in `subgraphStore` for DRY usage
across V1/V2 sidebars

## Review Focus
- `isUserBlueprint` consolidates logic previously duplicated between
`NodeTreeLeaf` and the new V2 components
- Context menu guard `contextMenuNode?.data` prevents showing empty
menus
- Folder `@contextmenu` handler clears stale `contextMenuNode` to
prevent wrong actions
┆Issue is synchronized with this [Notion
page](https://www.notion.so/PR-10827-fix-add-delete-bookmark-actions-for-blueprints-in-V2-node-library-sidebar-3366d73d36508111afd2c2c7d8ff0220)
by [Unito](https://www.unito.io)
## Summary
- Add `ModelLibraryHelper` mock helper for `/experiment/models` and
`/view_metadata` endpoints
- Add `ModelLibrarySidebarTab` page object fixture with search, folder,
and leaf locators
- Add 11 E2E test scenarios covering tab open/close, folder display,
folder expansion, search with debounce, refresh, load all folders, and
empty state
## Test plan
- [ ] CI passes all Playwright shards
- [ ] `pnpm typecheck:browser` passes
- [ ] `pnpm lint` passes
┆Issue is synchronized with this [Notion
page](https://www.notion.so/PR-10789-test-modelLibrary-add-E2E-tests-for-model-library-sidebar-tab-3356d73d365081b49a7ed752512164da)
by [Unito](https://www.unito.io)
---------
Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
## Summary
The existing pagination smoke test only asserts `count >= 1`, which
passes even if the sidebar eagerly loads all items or ignores page
boundaries entirely.
### What changed
**Before:**
- Created 30 mock jobs (less than BATCH_SIZE of 200) — all loaded in one
request, `has_more: false`
- Asserted `count >= 1` — redundant with the grid-render smoke test
**After — two targeted assertions:**
1. **Initial batch < total**: Mock 250 jobs (> BATCH_SIZE 200). First
`/api/jobs?limit=200&offset=0` returns 200 items with `has_more: true`.
Assert `initialCount < 250`.
2. **Scroll triggers second fetch**: Scroll `VirtualGrid` container to
bottom → `approach-end` event → `handleApproachEnd()` →
`assetsStore.loadMoreHistory()` → `/api/jobs?limit=200&offset=200`
fetches remaining 50. Assert `finalCount > initialCount` via
`expect.poll()`.
### Types
Mock data uses `RawJobListItem` from
`src/platform/remote/comfyui/jobs/jobTypes.ts` (Zod-inferred). This is
the correct source-of-truth per `docs/guidance/playwright.md` —
`/api/jobs` is a Python backend endpoint not covered by
`@comfyorg/ingest-types`.
## Test plan
- [ ] CI E2E tests pass
- [ ] `initial batch is smaller than total job count` validates
pagination boundary
- [ ] `scrolling to the end loads additional items` triggers actual
second API call
Fixes#10649
---------
Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
## Summary
Add Playwright E2E tests for workflow tab interactions in the topbar.
## Changes
- **What**: New test file
`browser_tests/tests/topbar/workflowTabs.spec.ts` with 5 tests covering
default tab visibility, tab creation, switching, closing, and context
menu. Added `newWorkflowButton`, `getTab()`, and `getActiveTab()`
locators to `Topbar.ts` fixture.
## Review Focus
Tests are focused on tab UI interactions only (sidebar workflow
operations are already covered in `workflows.spec.ts`). Context menu
assertion uses Reka UI's `data-reka-context-menu-content` attribute.
┆Issue is synchronized with this [Notion
page](https://www.notion.so/PR-10796-test-add-E2E-tests-for-workflow-tab-operations-3356d73d36508170a657ef816e23b71c)
by [Unito](https://www.unito.io)
---------
Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
## Summary
- Adds Playwright E2E tests for the Node Library V2 sidebar tab
(`Comfy.NodeLibrary.NewDesign: true`)
- Adds `NodeLibrarySidebarTabV2` fixture class with V2-specific locators
(search input, tab buttons, node cards)
- Exposes `menu.nodeLibraryTabV2` on `ComfyPage` for test access
- Tests cover: tab visibility, default tab selection, tab switching,
folder expansion, search filtering, and sort button presence
## Test plan
- [ ] Run `pnpm test:browser:local -- --grep "Node library sidebar V2"`
against a running ComfyUI server with the V2 node library
- [ ] Verify tests pass in CI
Fixes#9079
┆Issue is synchronized with this [Notion
page](https://www.notion.so/PR-10798-test-add-E2E-tests-for-Node-Library-V2-sidebar-3356d73d36508185a11feaf95e32225b)
by [Unito](https://www.unito.io)
---------
Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
## Summary
- Replace raw CSS selectors (`.lg-node-header`, `.p-contextmenu`,
`.node-title-editor input`, `.image-preview img`) with centralized
`TestIds` constants and existing fixtures in the context menu E2E spec
- Add `data-testid="title-editor-input"` to TitleEditor overlay for
stable selector targeting
- Use `NodeLibrarySidebarTab` fixture for node library sidebar
interaction
## Changes
- `browser_tests/fixtures/selectors.ts`: add `pinIndicator`,
`innerWrapper`, `titleEditorInput`, `mainImage` to `TestIds.node`
- `browser_tests/fixtures/utils/vueNodeFixtures.ts`: add `pinIndicator`
getter
- `src/components/graph/TitleEditor.vue`: add `data-testid` via
`input-attrs`
- `browser_tests/.../contextMenu.spec.ts`: replace all raw selectors
with TestIds/fixtures
## Test plan
- [x] All 23 context menu E2E tests pass locally
- [x] Typecheck passes
- [x] Lint passes
Fixes#10750Fixes#10749
┆Issue is synchronized with this [Notion
page](https://www.notion.so/PR-10760-test-replace-raw-CSS-selectors-with-TestIds-in-context-menu-spec-3336d73d3650818790c0e32e0b6f1e98)
by [Unito](https://www.unito.io)
2026-04-02 21:04:50 +01:00
936 changed files with 47848 additions and 19459 deletions
**⚠️ NEVER backport cloud-only PRs to `core/*` branches.** Cloud-only changes (app mode, cloud auth, cloud billing UI, cloud-specific API calls) are irrelevant to local users and waste effort. Before backporting any PR to a `core/*` branch, check:
### What Goes Where
- Does the PR title/description mention "app mode", "cloud", or cloud-specific features?
- Does the PR only touch files like `appModeStore.ts`, cloud auth, or cloud-specific components?
-If yes → skip for `core/*` branches (may still apply to `cloud/*` branches)
**Both core + cloud:**
-**App mode** PRs — app mode is NOT cloud-only
- **Firebase auth** PRs — Firebase auth is on core for API nodes
**⚠️ NEVER backport cloud-only PRs to `core/*` branches.** But do NOT assume "app mode" or "Firebase" = cloud-only. Check the actual files changed.
## ⚠️ Gotchas (Learn from Past Sessions)
@@ -67,6 +81,32 @@ The `pr-backport.yaml` action reports more conflicts than reality. `git cherry-p
12 or 27 conflicting files can be trivial (snapshots, new files). **Categorize conflicts first**, then decide. See Conflict Triage below.
### Accept-Theirs Can Produce Broken Hybrids
When a PR **rewrites a component** (e.g., PrimeVue → Reka UI), the accept-theirs regex produces a broken mix of old and new code. The template may reference new APIs while the script still has old imports, or vice versa.
**Detection:** Content conflicts with 4+ conflict markers in a single `.vue` file, especially when imports change between component libraries.
**Fix:** Instead of accept-theirs regex, use `git show MERGE_SHA:path/to/file > path/to/file` to get the complete correct version from the merge commit on main. This bypasses the conflict entirely.
### Cherry-Picks Can Reference Missing Dependencies
When PR A on main depends on code introduced by PR B (which was merged before A), cherry-picking A brings in code that references B's additions. The cherry-pick succeeds but the branch is broken.
**Common pattern:** Composables, component files, or type definitions introduced by an earlier PR and used by the cherry-picked PR.
**Detection:**`pnpm typecheck` fails with "Cannot find module" or "is not defined" errors after cherry-pick.
**Fix:** Use `git show MERGE_SHA:path/to/missing/file > path/to/missing/file` to bring the missing files from main. Always verify with typecheck.
### Use `--no-verify` for Worktree Pushes
Husky hooks fail in worktrees (can't find lint-staged config). Always use `git push --no-verify` and `git commit --no-verify` when working in `/tmp/` worktrees.
### Automation Success Varies Wildly by Branch
In the 2026-04-06 session: core/1.42 got 18/26 auto-PRs, cloud/1.42 got only 1/25. The cloud branch has more divergence. **Always plan for manual fallback** — don't assume automation will handle most PRs.
## Conflict Triage
**Always categorize before deciding to skip. High conflict count ≠ hard conflicts.**
@@ -77,6 +117,8 @@ The `pr-backport.yaml` action reports more conflicts than reality. `git cherry-p
| **Modify/delete (new file)** | PR introduces files not on target | `git add $FILE` — keep the new file |
| **Modify/delete (removed)** | Target removed files the PR modifies | `git rm $FILE` — file no longer relevant |
| **Content conflicts** | Marker-based (`<<<<<<<`) | Accept theirs via python regex (see below) |
| **Component rewrites** | 4+ markers in `.vue`, library change | Use `git show SHA:path > path` — do NOT accept-theirs |
| **Import-only conflicts** | Only import lines differ | Keep both imports if both used; remove unused after |
| **Add/add** | Both sides added same file | Accept theirs, verify no logic conflict |
@@ -103,7 +145,7 @@ Skip these without discussion:
- **Test-only / lint rule changes** — Not user-facing
- **Revert pairs** — If PR A reverted by PR B, skip both. If fixed version (PR C) exists, backport only C.
- **Features not on target branch** — e.g., Painter, GLSLShader, appModeStore on core/1.40
- **Cloud-only PRs on core/\* branches** — App mode, cloud auth, cloud billing. These only affect cloud-hosted ComfyUI.
- **Cloud-only PRs on core/\* branches** — Team workspaces, cloud queue, cloud-only login. (Note: app mode and Firebase auth are NOT cloud-only — see Branch Scope Rules)
If typecheck or tests fail, stop and investigate before continuing. A broken branch after wave N means all subsequent waves will compound the problem.
### Fix PRs Are Normal
Expect to create 1 fix PR per branch after verification. Common issues:
1.**Component rewrite hybrids** — accept-theirs produced broken `.vue` files. Fix: overwrite with correct version from merge commit via `git show SHA:path > path`
2.**Missing dependency files** — cherry-pick brought in code referencing composables/components not on the branch. Fix: add missing files from merge commit
3.**Missing type properties** — cherry-picked code uses interface properties not yet on the branch (e.g., `key` on `ConfirmDialogOptions`). Fix: add the property to the interface
4.**Unused imports** — conflict resolution kept imports that the branch doesn't use. Fix: remove unused imports
5.**Wrong types from conflict resolution** — e.g., `{ top: number; right: number }` vs `{ top: number; left: number }`. Fix: match the return type of the actual function
Create a fix PR on a branch from the target, verify typecheck passes, then merge with `--squash --admin`.
### Never Admin-Merge Without CI
In a previous bulk session, all 69 backport PRs were merged with `gh pr merge --squash --admin`, bypassing required CI checks. This shipped 3 test failures to a release branch. **Lesson: `--admin` skips all branch protection, including required status checks.** Only use `--admin` after confirming CI has passed (e.g., `gh pr checks $PR` shows all green), or rely on auto-merge (`--auto --squash`) which waits for CI by design.
@@ -135,6 +189,43 @@ Large backport sessions (50+ PRs) are expensive and error-prone. Prefer continuo
- Reserve session-style bulk backporting for catching up after gaps
- When a release branch is created, immediately start the continuous process
## Interactive Approval Flow
After analysis, present ALL candidates (MUST, SHOULD, and borderline) to the human for interactive review before execution. Do not write a static decisions.md — collect approvals in conversation.
### Batch Presentation
Present PRs in batches of 5-10, grouped by theme (visual bugs, interaction bugs, cloud/auth, data correctness, etc.). Use this table format:
When many PRs need manual cherry-pick (e.g., cloud branches), test all first:
```bash
cd /tmp/backport-$BRANCH
for pr in "${ORDER[@]}";do
git checkout -b test-$pr origin/$BRANCH
if git cherry-pick -m 1$SHA 2>/dev/null;then
echo"CLEAN: $pr"
else
echo"CONFLICT: $pr"
git cherry-pick --abort
fi
git checkout --detach HEAD
git branch -D test-$pr
done
```
Then process clean PRs in a batch loop, conflicts individually.
### PR Title Convention
```
[backport TARGET_BRANCH] Original Title (#ORIGINAL_PR)
```
## Final Deliverables (Slack-Compatible)
After execution completes, generate two files in `~/temp/backport-session/`. Both must be **Slack-compatible plain text** — no emojis, no markdown tables, no headers (`#`), no bold (`**`), no inline code. Use plain dashes, indentation, and line breaks only.
### 1. Author Accountability Report
File: `backport-author-accountability.md`
Lists all backported PRs grouped by original author (via `gh pr view $PR --json author`). Surfaces who should be self-labeling.
```
Backport Session YYYY-MM-DD -- PRs that should have been labeled by authors
- author-login
- #1234 fix: short title
- #5678 fix: another title
- other-author
- #9012 fix: some other fix
```
Authors sorted alphabetically, 4-space indent for nested items.
### 2. Slack Status Update
File: `slack-status-update.md`
A shareable summary of the session. Structure:
```
Backport session complete -- YYYY-MM-DD
[1-sentence summary: N PRs backported to which branches. All pass typecheck.]
Branches updated:
- core/X.XX: N PRs + N fix PRs (N auto, N manual)
- cloud/X.XX: N PRs + N fix PRs (N auto, N manual)
- ...
N total PRs created and merged (N backports + N fix PRs).
Notable fixes included:
- [category]: [list of fixes]
- ...
Conflict patterns encountered:
- [pattern and how it was resolved]
- ...
N authors had PRs backported. See author accountability list for details.
```
No emojis, no tables, no bold, no headers. Plain text that pastes cleanly into Slack.
| `core/*` | Cloud-only (team workspaces, cloud queue, cloud-only login). Note: app mode and Firebase auth are NOT cloud-only. |
| `cloud/*` | Local-only features not present on cloud branch |
Cloud-only PRs backported to `core/*` are wasted effort — `core/*` branches serve local/self-hosted users who never see cloud features. Check PR titles, descriptions, and files changed for cloud-specific indicators.
@@ -61,8 +61,6 @@ done
## Human Review Checkpoint
Present decisions.md before execution. Include:
Use the Interactive Approval Flow (see SKILL.md) to review all candidates interactively. Do not write a static decisions.md for the human to edit — instead, present batches of 5-10 PRs with context and recommendations, and collect Y/N/? responses in conversation.
1. All MUST/SHOULD/SKIP categorizations with rationale
2. Questions for human (feature existence, scope, deps)
3. Estimated effort per branch
All candidates must be reviewed (MUST, SHOULD, and borderline items), not just a subset.
8.**Always validate JSON** after resolving locale file conflicts
9.**Dep refresh PRs** — skip on stable branches. Risk of transitive dep regressions outweighs audit cleanup. Cherry-pick individual CVE fixes instead.
10.**Verify after each wave** — run `pnpm typecheck && pnpm test:unit` on the target branch after merging a batch. Catching breakage early prevents compounding errors.
11.**Cloud-only PRs don't belong on core/\* branches** — app mode, cloud auth, and cloud-specific UI changes are irrelevant to local users. Always check PR scope against branch scope before backporting.
11.**App mode and Firebase auth are NOT cloud-only** — they go to both core and cloud branches. Only team workspaces, cloud queue, and cloud-specific login are cloud-only.
12.**Never admin-merge without CI** — `--admin` bypasses all branch protections including required status checks. A bulk session of 69 admin-merges shipped 3 test failures. Always wait for CI to pass first, or use `--auto --squash` which waits by design.
13.**Accept-theirs regex breaks component rewrites** — when a PR migrates between component libraries (PrimeVue → Reka UI), the regex produces a broken hybrid. Use `git show SHA:path > path` to get the complete correct version instead.
14.**Cherry-picks can silently bring in missing-dependency code** — if PR A references a composable introduced by PR B, cherry-picking A succeeds but typecheck fails. Always run typecheck after each wave and add missing files from the merge commit.
15.**Fix PRs are expected** — plan for 1 fix PR per branch to resolve typecheck issues from conflict resolutions. This is normal, not a failure.
16.**Use `--no-verify` in worktrees** — husky hooks fail in `/tmp/` worktrees. Always push/commit with `--no-verify`.
17.**Automation success varies by branch** — core/1.42 got 18/26 auto-PRs (69%), cloud/1.42 got 1/25 (4%). Cloud branches diverge more. Plan for manual fallback.
18.**Test-then-resolve pattern** — for branches with low automation success, run a dry-run loop to classify clean vs conflict PRs before processing. This is much faster than resolving conflicts serially.
Track verification results per wave within execution-log.md:
```markdown
## Wave N Verification — TARGET_BRANCH
Wave N Verification -- TARGET_BRANCH
- PRs merged: #A, #B, #C
- Typecheck: ✅ Pass / ❌ Fail
-Unit tests: ✅ Pass / ❌ Fail
- Typecheck: pass / fail
-Fix PR: #YYYY (if needed)
- Issues found: (if any)
- Human review needed: (list any non-trivial conflict resolutions)
```
## Session Report Template
@@ -63,40 +62,42 @@ Track verification results per wave:
- Feature branches that need tracking for future sessions?
```
## Final Deliverable: Visual Summary
## Final Deliverables
At session end, generate a **mermaid diagram** showing all backported PRs organized by target branch and category (MUST/SHOULD), plus a summary table. Present this to the user as the final output.
After all branches are complete and verified, generate these files in `~/temp/backport-session/`:
```mermaid
graph TD
subgraph branch1["☁️ cloud/X.XX — N PRs"]
C1["#XXXX title"]
C2["#XXXX title"]
end
### 1. execution-log.md (internal)
subgraph branch2must["🔴 core/X.XX MUST — N PRs"]
M1["#XXXX title"]
end
Per-branch tables with PR#, title, status, backport PR#, notes. Markdown tables are fine — this is for internal tracking, not Slack.
subgraph branch2should["🟡 core/X.XX SHOULD — N PRs"]
See SKILL.md "Final Deliverables" section. Plain text, no emojis/tables/headers/bold. Authors sorted alphabetically with PRs nested under each.
Use the `mermaid` tool to render this diagram and present it alongside the summary table as the session's final deliverable.
### 3. slack-status-update.md (Slack-compatible)
See SKILL.md "Final Deliverables" section. Plain text summary that pastes cleanly into Slack. Includes branch counts, notable fixes, conflict patterns, author count.
## Slack Formatting Rules
Both shareable files (author accountability + status update) must follow these rules:
- No emojis (no checkmarks, no arrows, no icons)
- No markdown tables (use plain lists with dashes)
- No headers (no # or ##)
- No bold (\*_) or italic (_)
- No inline code backticks
- Use -- instead of em dash
- Use plain dashes (-) for lists with 4-space indent for nesting
- Line breaks between sections for readability
These files should paste directly into a Slack message and look clean.
## Files to Track
-`candidate_list.md` — all candidates per branch
-`decisions.md` — MUST/SHOULD/SKIP with rationale
-`wave-plan.md` — execution order
-`execution-log.md` — real-time status
-`backport-session-report.md` — final summary
All in `~/temp/backport-session/`:
All in `~/temp/backport-session/`.
-`execution-plan.md` -- approved PRs with merge SHAs (input)
-`execution-log.md` -- real-time status with per-branch tables (internal)
-`backport-author-accountability.md` -- PRs grouped by author (Slack-compatible)
description: 'Diagnoses and fixes flaky Playwright e2e tests by replacing race-prone patterns with retry-safe alternatives. Use when triaging CI flakes, hardening spec files, fixing timing races, or asked to stabilize browser tests. Triggers on: flaky, flake, harden, stabilize, race condition in e2e, intermittent failure.'
---
# Hardening Flaky E2E Tests
Fix flaky Playwright specs by identifying race-prone patterns and replacing them with retry-safe alternatives. This skill covers diagnosis, pattern matching, and mechanical transforms — not writing new tests (see `writing-playwright-tests` for that).
## Workflow
### 1. Gather CI Evidence
```bash
gh run list --workflow=ci-test.yaml --limit=5
gh run download <run-id> -n playwright-report
```
- Open `report.json` and search for `"status": "flaky"` entries.
- Collect file paths, test titles, and error messages.
- Do NOT trust green checks alone — flaky tests that passed on retry still need fixing.
- Use `error-context.md`, traces, and page snapshots before editing code.
- Pull the newest run after each push instead of assuming the flaky set is unchanged.
### 2. Classify the Flake
Read the failing assertion and match it against the pattern table. Most flakes fall into one of these categories:
| Screenshot | Ensure nodes are rendered: `waitForNodes()` or poll state |
#### Rule: Expose Locators for Retrying Assertions
When a helper returns a count via `await loc.count()`, callers can't use `toHaveCount()`. Expose the underlying `Locator` as a getter so callers choose between:
```typescript
// Helper exposes locator
getdomWidgets():Locator{
returnthis.page.locator('.dom-widget')
}
// Caller uses retrying assertion
awaitexpect(comfyPage.domWidgets).toHaveCount(2)
```
Replace count methods with locator getters so callers can use retrying assertions directly.
#### Rule: Fix Check-then-Act Races in Helpers
```typescript
// ❌ Race: count can change between check and waitFor
constcount=awaitlocator.count()
if(count>0){
awaitlocator.waitFor({state:'hidden'})
}
// ✅ Direct: waitFor handles both cases
awaitlocator.waitFor({state:'hidden'})
```
#### Rule: Remove force:true from Clicks
`force: true` bypasses actionability checks, hiding real animation/visibility races. Remove it and fix the underlying timing issue.
```typescript
// ❌ Hides the race
awaitcloseButton.click({force: true})
// ✅ Surfaces the real issue — fix with proper wait
awaitcloseButton.click()
awaitdialog.waitForHidden()
```
#### Rule: Handle Non-deterministic Element Order
When `getNodeRefsByType` returns multiple nodes, the order is not guaranteed. Don't use index `[0]` blindly.
description: 'Parse ticket URL (Notion or GitHub), extract all data, initialize pipeline run. Use when starting work on a new ticket or when asked to pick up a ticket.'
---
# Ticket Intake
Parses a ticket URL from supported sources (Notion or GitHub), extracts all relevant information, and creates a ticket in the pipeline API.
> **🚨 CRITICAL REQUIREMENT**: This skill MUST register the ticket in the Pipeline API and update the source (Notion/GitHub). If these steps are skipped, the entire pipeline breaks. See [Mandatory API Calls](#mandatory-api-calls-execute-all-three) below.
**⚠️ These three API calls are the ENTIRE POINT of this skill. Without them, the ticket is invisible to the pipeline, downstream skills will fail, and Notion status won't update.**
**You MUST make these HTTP requests.** Use `curl` from bash — do not just read this as documentation.
**If `state` is not `RESEARCH`, go back to Step 4 and complete the missing calls.**
### Step 6: Output Summary and Handoff
Print a clear summary:
```markdown
## Ticket Intake Complete
**Source:** Notion | GitHub
**Title:** [Ticket title]
**ID:** abc12345
**Status:** In Progress (queued)
**Priority:** High
**Area:** UI
### Description
[Brief description or first 200 chars]
### Acceptance Criteria
- [ ] Criterion 1
- [ ] Criterion 2
### Links
- **Ticket:** [Original URL]
- **Slack:** [Slack thread content fetched via slackdump] (Notion only)
### Pipeline
- **API Ticket ID:** abc12345
- **State:** RESEARCH
- **Verified:** ✅ (via verify-intake.sh or public API)
```
**After printing the summary, immediately handoff** to continue the pipeline. Use the `handoff` tool with all necessary context (ticket ID, source, title, description, slack context if any):
> **Handoff goal:** "Continue pipeline for ticket {ID} ({title}). Ticket is in RESEARCH state. Load skill: `research-orchestrator` to begin research phase. Ticket data: source={source}, notion_page_id={pageId}, priority={priority}. {slack context summary if available}"
**Do NOT wait for human approval to proceed.** The intake phase is complete — handoff immediately.
## Error Handling
### Unsupported URL
```
❌ Unsupported ticket URL format.
Supported formats:
- Notion: https://notion.so/... or https://www.notion.so/...
@@ -179,6 +179,12 @@ This project uses **pnpm**. Always prefer scripts defined in `package.json` (e.g
24. Do not use function expressions if it's possible to use function declarations instead
25. Watch out for [Code Smells](https://wiki.c2.com/?CodeSmell) and refactor to avoid them
## Design Standards
Before implementing any user-facing feature, consult the [Comfy Design Standards](https://www.figma.com/design/QreIv5htUaSICNuO2VBHw0/Comfy-Design-Standards) Figma file. Use the Figma MCP to fetch it live — the file is the single source of truth and may be updated by designers at any time.
See `docs/guidance/design-standards.md` for Figma file keys, section node IDs, and component references.
## Testing Guidelines
See @docs/testing/\*.md for detailed patterns.
@@ -226,6 +232,7 @@ See @docs/testing/\*.md for detailed patterns.
- Run `pnpm dev:electron` to start the dev server with electron API mocked
- Run `pnpm dev:cloud` to start the dev server against the cloud backend (instead of local ComfyUI server)
#### Testing with Cloud & Staging Environments
Some features — particularly **partner/API nodes** (e.g. BFL, OpenAI, Stability AI) — require a cloud backend for authentication and billing. Running these against a local ComfyUI instance will result in permission errors or logged-out states. There are two ways to connect to a cloud/staging backend:
**Option 1: Frontend — `pnpm dev:cloud`**
The simplest approach. This proxies all API requests to the test cloud environment:
```bash
pnpm dev:cloud
```
This sets `DEV_SERVER_COMFYUI_URL` to `https://testcloud.comfy.org/` automatically. You can also set this variable manually in your `.env` file to target a different environment:
Any `*.comfy.org` URL automatically enables cloud mode, which includes the GCS media proxy needed for viewing generated images and videos. See [.env_example](.env_example) for all available cloud URLs.
**Option 2: Backend — `--comfy-api-base`**
Alternatively, launch the ComfyUI backend pointed at the staging API:
Some files were not shown because too many files have changed in this diff
Show More
Reference in New Issue
Block a user
Blocking a user prevents them from interacting with repositories, such as opening or commenting on pull requests or issues. Learn more about blocking a user.