The bottom composer now edits and reruns the current generation node while canvas-visible actions own downstream node creation. NewApiWG media normalization, video frame extraction, and node quick-add keep generated media flows explicit and connected rather than hidden behind the prompt box.
Constraint: Composer must not call providers directly or create hidden downstream branches when a node is selected
Constraint: NewApiWG video families use inconsistent media schemas, so common media aliases are normalized in the provider adapter
Rejected: Let the bottom composer create next-step nodes from selected outputs | hides graph mutation and conflicts with the PRD source-of-truth rules
Rejected: Put NewApiWG model-family payload rules in the composer | provider-specific schemas would leak into UI state
Confidence: high
Scope-risk: moderate
Directive: Keep provider payload normalization in provider/executor layers; do not move it into GenerationComposer
Tested: npm run test:run -- src/components/__tests__/GenerationComposer.test.tsx
Tested: npm run test:run -- src/components/__tests__/WorkflowCanvas.test.tsx
Tested: npm run test:run -- src/components/__tests__/GenerateVideoNode.test.tsx
Tested: npm run test:run -- src/app/api/generate/providers/__tests__/newapiwg.test.ts src/app/api/generate/__tests__/route.test.ts
Tested: npm run test:run -- src/store/execution/__tests__/videoProcessingExecutors.test.ts
Tested: npm run build
Not-tested: npm run lint fails because next lint is incompatible with the current Next 16 script configuration
Remote and server-hosted deployments cannot rely on the browser user's desktop tools being available inside the Node server process. Directory browsing now checks for native picker commands first and falls back to a configurable server-side workflows directory instead of surfacing command-not-found errors.
Constraint: The browse endpoint runs on the server, not on the Windows client browser
Constraint: Test servers may not have osascript, zenity, kdialog, PowerShell, or an interactive desktop session
Rejected: Keep propagating native picker failures to the modal | successful server deployments without desktop tools showed red command errors
Confidence: high
Scope-risk: narrow
Directive: For true client-local folder selection, use a separate File System Access API flow instead of this server filesystem endpoint
Tested: npx vitest run src/app/api/browse-directory/__tests__/route.test.ts src/components/__tests__/ProjectSetupModal.test.tsx
Tested: npm run build
Not-tested: Interactive Windows Server desktop picker session
NewApiWG image and video models need to work against the PopiArt test gateway, with generated remote media visible in the canvas and UI text localized for Chinese users. This wires the gateway-specific image/video paths, model fallback display, remote media proxying, and related i18n cleanup into one releaseable change.
Constraint: Test gateway exposes image and video models through mixed OpenAI-compatible and Gemini-native response shapes
Constraint: Browser video playback can fail when loading signed third-party media URLs directly
Rejected: Keep remote video URLs as direct <video> sources | CORS and range handling made successful generations appear unloaded
Confidence: high
Scope-risk: moderate
Directive: Do not remove the media proxy without validating remote video playback, trimming, and stitching against signed CDN URLs
Tested: npm run build
Tested: npx vitest run src/components/__tests__/OutputNode.test.tsx src/components/__tests__/GenerateVideoNode.test.tsx src/components/__tests__/VideoStitchNode.test.tsx
Tested: npx vitest run src/components/__tests__/WorkflowCanvas.test.tsx src/store/execution/__tests__/generateVideoExecutor.test.ts
Tested: npx vitest run src/app/api/generate/providers/__tests__/newapiwg.test.ts
Not-tested: Live browser replay after pushing to remote deployment
Kie rejects the call when both reference_image_urls and first_frame_url
(or last_frame_url) are populated — the API treats them as mutually
exclusive scene modes. Two changes:
1. kie.ts submit: skip the legacy image fallback whenever any
schema-driven image key was already handled via dynamicInputs. The
executor passes both `images` and `dynamicInputs`, so the fallback
was duplicating user-provided references into first_frame_url.
2. seedance2 I2V schema: mark first_frame_url optional and note the
mutual exclusivity in the handle descriptions so users see it in
the node UI.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Add last_frame_url, reference_image_urls, reference_video_urls, and
reference_audio_urls handles to Seedance 2.0 and Seedance 2.0 Fast I2V
schemas so users can seed an end frame, supplementary reference images,
reference videos, and reference audio in addition to the first frame.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Video handles/edges/labels were inconsistent — using white, green, blue,
and hardcoded purple across different files. Standardize all to pink and
use the CSS variable var(--handle-color-video) for labels. Also adds the
missing .react-flow__handle[data-handletype="video"] CSS rule.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Add --handle-color-video CSS variable
- Fix tutorial overlay race conditions by reading nodes from store
directly and add keyboard accessibility to click-to-continue overlay
- Fix audio extraction in video stitching: defer silent buffer creation
until reference format is known, and only trust probed duration when
all blobs were successfully probed
- Clear saved filename/path when clearing 3D output
- Retry 429/408 transient errors during task polling
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Resolve conflicts in 4 executor files by combining develop's
runWithFallback structure with feature branch's client-side polling.
Both capabilities are now integrated: executors use runWithFallback
for model redundancy, and pollGenerateTask for long-running Kie tasks.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
New endpoint accepts POST with taskId/provider/modelId and returns
either a polling-continue response or the final media result. Each
request is short-lived (~1s) instead of holding a connection open.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
The /api/generate route now submits Kie tasks and returns immediately
with polling metadata (taskId, pollProvider, etc.) instead of holding
the HTTP connection open for 2-10 minutes. Export buildMediaResponse
for reuse by the poll endpoint.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Split the monolithic generateWithKie into three reusable lifecycle functions
for client-side polling support. generateWithKie now composes them internally
for backward compatibility. Added polling fields to GenerateResponse type.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Wrap fetch and JSON parsing in try-catch to handle network errors,
add 429/5xx to retryable status codes, and implement exponential
backoff (2s → 10s) to reduce API load during long-running generations.
Caps consecutive errors at 5 before failing permanently.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
The Kie API can return 404/422 or HTTP 200 with non-200 code when
polling starts before the task is fully registered. Previously these
were treated as fatal errors, causing the client to see a network error
even though the generation was still processing on Kie's side. Now
these transient responses are retried instead of aborting.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
maxDuration was 5 minutes which caused Next.js to kill the route handler
before Seedance 2.0 video generation polling could complete, resulting
in a network error on the client despite successful generation.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
The Kie API expects "bytedance/seedance-2" as the model value, but we
were sending our internal ID "bytedance/seedance-2/text-to-video" which
includes the capability suffix. Strip the suffix for Seedance models
before sending to the API.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Gemini image models (nano-banana, nano-banana-pro, nano-banana-2) were
not showing parameters when set as fallback models. Three root causes:
1. Schema API returned empty for Gemini image models — added
getGeminiImageSchema() with aspect ratio, resolution, and search
grounding parameters per model
2. ModelParameters had an early-return guard skipping all non-Veo Gemini
models — removed to make the component provider-agnostic
3. nanoBananaExecutor ignored parametersOverride for Gemini-specific
fields (aspectRatio, resolution, useGoogleSearch, useImageSearch) —
now reads from override when available
Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>
- Restore FTUX production logic (remove testing override)
- Fix generate3dExecutor clearOutput field and add modelId validation
- Remap startLevel index after loop partitioning in executeWorkflow
- Add missing mockTutorialExecution dep in FloatingActionBar
- Reset settingsTab to "primary" when fallbackModel is removed
- Clear orphaned fallbackParameters when removing fallback model
- Fix OutputGalleryNode removeMedia to splice single item, not all duplicates
- Add data-tutorial="save-button" to untitled save button in Header
- Move data-tutorial to layout-producing element in BaseNode
- Hydrate FTUXApiKeysStep localKeys from providerSettings
- Remove broken docs URL from ftuxStore tutorial steps
- Add error handling in loadTutorialSampleImage
- Fix mediaStorage ref array misalignment in outputGallery hydration
- Make OutputGalleryNodeData.videos optional to match usage
- Remove unused fitView destructuring from WorkflowCanvas
Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>
The Kie docs have contradictory info — the parameter description says
"std"/"pro" but the example request body uses "720p"/"1080p". The API
actually accepts "720p"/"1080p" (same as Kling 2.6 Motion Control),
rejecting "std"/"pro" with "mode is not within the range of allowed
options."
Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>
Per Kie docs, background_source accepts "input_video"/"input_image",
not "image"/"video". The previous wrong values were being sent to the
API after schema defaults pre-population, causing request rejection.
Also added background_source to getKieModelDefaults() for consistency.
Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>
The dynamic input upload logic only matched data:image URLs, so video
data (data:video/mp4) was passed as raw base64 instead of being
uploaded to get a hosted URL. This caused "File type not supported"
errors for models like Kling 3.0 Motion Control that accept video.
- Rename uploadImageToKie → uploadMediaToKie (keep alias for compat)
- Rename detectImageType → detectMediaType with video magic bytes and
fallback to declared MIME type for non-image content
- Broaden data URL checks from data:image to data: in generateWithKie
- Set uploadPath to videos/audio/images based on detected MIME type
Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>
Add Seedance 2.0 and 2.0 Fast with text-to-video and image-to-video
variants. Both support audio generation, web search, and 4-15 second
duration. Image input uses first_frame_url (singular string).
Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>
Add first_frame_url, last_frame_url, and first_clip_url to the
singular-key checks in generateWithKie() so these params receive a
single string URL instead of being wrapped in an array.
Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>
* feat(49-01): add FTUX types and localStorage helpers
- Create src/types/ftux.ts with FTUXStep, FTUXModalProps, FTUXStepProps
- Add FTUX_COMPLETED_KEY to localStorage keys
- Add getFTUXCompleted() and setFTUXCompleted() helpers
- Follow existing localStorage pattern with server-safe checks
* feat(49-02): create tutorial state management with Zustand
- TutorialStep interface with id, message, highlightSelector, requiredAction
- FTUXState with tutorialActive, currentTutorialStep, tutorialSteps, lockedFeatures
- Actions: startTutorial, skipTutorial, completeCurrentStep, nextTutorialStep, resetTutorial
- setFTUXCompleted/getFTUXCompleted localStorage helpers
- Initial tutorial steps (5 steps, placeholders for Plan 03)
* feat(49-01): build FTUXModal container and step components
- Create FTUXModal.tsx: 4-step modal with navigation, skip confirmation
- Create FTUXWelcomeStep.tsx: Welcome screen with project intro
- Create FTUXApiKeysStep.tsx: Provider list with .env status checking
- Create FTUXModelDefaultsStep.tsx: Model selection via ModelSearchDialog
- Create FTUXReadyStep.tsx: Completion screen with tutorial option
- Follow WelcomeModal and ProjectSetupModal patterns
- Use Tailwind neutral-800 theme throughout
- Step indicators show progress (1-4)
- Back button hidden on step 1
- Skip confirmation shows nested dialog overlay
* feat(49-02): build typewriter hook and tutorial UI components
- useTypewriter hook with 50ms/char animation and isComplete flag
- TutorialMessage component with typewriter animation and dark backdrop
- ElementHighlight component with pulsing blue ring (2px border, 8px blur)
- Three-layer highlight: dimmed overlay (bg-black/60), blue ring (z-91), clickable window (z-92)
- TutorialOverlay coordinator with action detection (add-image-node, connect-nodes, run-workflow)
- Skip tutorial button in top-right corner
- pulse-ring keyframes added to globals.css
- Uses React portals for correct z-index stacking
* feat(49-03): replace WelcomeModal with FTUXModal in app entry point
- Import FTUXModal, getFTUXCompleted, setFTUXCompleted, useFTUXStore
- Add SSR-safe showFTUX state (initialized via useEffect)
- Render FTUXModal when showFTUX=true
- onComplete callback sets showFTUX=false and marks FTUX complete
- onStartTutorial callback hides modal and starts tutorial via ftuxStore
- WelcomeModal still available via logo click (not removed, just not on first visit)
* feat(49-03): integrate tutorial overlay and add data-tutorial attributes
WorkflowCanvas.tsx:
- Import TutorialOverlay and useFTUXStore
- Get tutorialActive and lockedFeatures from ftuxStore
- Apply dimming (opacity-30 + pointer-events-none) to Background, Controls, MiniMap when tutorial active
- Render TutorialOverlay component after AnnotationModal
FloatingActionBar.tsx:
- Add optional dataTutorial prop to NodeButton interface
- Add data-tutorial="image-button" to Image input button
- Add data-tutorial="output-button" to Output button
- These attributes allow ElementHighlight to target buttons during tutorial
* fix(49-03): resolve Zustand SSR hydration issue in WorkflowCanvas
- Replace direct useFTUXStore hook call with useState + useEffect pattern
- Subscribe to ftuxStore on client-side only via useEffect
- Initialize state with getState() after subscription
- Prevents 'getServerSnapshot should be cached' error during SSR
- Prevents 'Maximum update depth exceeded' infinite loop
* chore(49-03): add FTUX testing override to always show modal
- Force setShowFTUX(true) to bypass localStorage check
- Allows repeated testing without clearing localStorage each time
- Added clear TODO comment to revert before production
- Original production code commented out for easy restoration
* ux(49-03): improve FTUX modal copy and layout based on user testing
FTUXModal.tsx:
- Increase modal width from max-w-[520px] to max-w-[640px] for better readability
FTUXWelcomeStep.tsx:
- Remove large banana icon (w-20 h-20) for cleaner look
- Change headline from "Welcome to Node Banana" to "Let's get set up."
- Keep workflow description, update subtext to "This will only take a few quick steps."
FTUXApiKeysStep.tsx:
- Simplify description to emphasize .env.local security benefit
- New: "Add your API keys here, or save them to .env.local to keep them secure across sessions."
- Remove mention of settings (cleaner messaging)
FTUXModelDefaultsStep.tsx:
- Remove redundant "You can always change these later" from main description
- Remove "These defaults are applied when creating nodes" from footer (redundant)
- Keep single mention: "You can skip this step and configure them later in settings."
FTUXReadyStep.tsx:
- Remove large green checkmark icon (w-16 h-16 circular bg)
- Keep "All Set Up!" headline and description
- Keep both tutorial buttons unchanged
* ux(49-03): remove redundant .env.local text below API keys list
- Remove duplicate messaging at bottom of FTUXApiKeysStep
- Top description already mentions .env.local security benefit
- Cleaner, less repetitive UI
* copy(49-03): rewrite FTUX copy for clarity and conciseness
Applied principles: Clear, Concise, Consistent, Useful, Human
FTUXWelcomeStep:
- Headline: "Let's get set up." → "Let's get started."
- Body: Simplified from wordy workflow/pipeline jargon to clear action
Old: "Build AI workflows visually with nodes and connections. Create complex image and video generation pipelines by connecting simple building blocks."
New: "Connect AI models like building blocks to generate images, videos, and more."
- Removed defensive subtext ("This will only take a few quick steps")
FTUXApiKeysStep:
- Description: Clearer security benefit, less technical
Old: "Add your API keys here, or save them to .env.local to keep them secure across sessions."
New: "Add keys to use AI providers. For security, save them to your .env.local file instead of the browser."
FTUXModelDefaultsStep:
- Headline: "Default Models (Optional)" → "Choose Your Models (Optional)"
- Description: Removed redundant "by default when creating new nodes"
Old: "Choose which AI models to use by default when creating new nodes."
New: "Pick which AI models to use for images and videos."
- Removed defensive footer text ("You can skip...")
FTUXReadyStep:
- Headline: "All Set Up!" → "You're ready!"
- Body: Removed jargon and redundancy
Old: "You're ready to start building workflows. Would you like a quick tutorial to get started?"
New: "Want a quick tutorial?"
- Buttons: "Skip Tutorial" → "Skip", "Start Tutorial" → "Show me how"
* fix(49-03): remove double scrollbar in API Keys step
- Remove max-h-[380px] overflow-y-auto from provider list container
- FTUXModal already has overflow-y-auto on content area (line 112)
- Provider list now scrolls with rest of modal content, not separately
- Fixes nested scroll container causing double scrollbar
* copy(49-03): clarify browser storage in API Keys step
- Make it explicit that adding keys in UI = stored in browser
- Parenthetical "(stored in browser)" right after action
- "for better security" instead of "For security" flows better
- Clearer tradeoff between convenience (browser) and security (.env.local)
Old: "Add keys to use AI providers. For security, save them to your .env.local file instead of the browser."
New: "Add keys here to use AI providers (stored in browser), or save them to your .env.local file for better security."
* copy(49-03): add "across sessions" to API Keys security benefit
- Change "better security" to "better security across sessions"
- Makes the persistence benefit of .env.local more explicit
- Clarifies why file storage is better than browser storage
New: "Add keys here to use AI providers (stored in browser), or save them to your .env.local file for better security across sessions."
* fix(49-03): save API keys entered during FTUX onboarding
Problem: API keys entered in FTUX modal were stored in local state only
and disappeared when modal closed, requiring users to re-enter them.
Original plan decision (49-01): "API key inputs not persisted during FTUX flow"
Rationale: "Setup is informational only, actual persistence happens in ProjectSetupModal"
This created broken UX - users entered keys expecting them to work.
Solution:
- Import useWorkflowStore and updateProviderApiKey function
- Create handleKeyChange that saves to localStorage immediately via updateProviderApiKey
- Replace direct setLocalKeys with handleKeyChange in input onChange
- Keys now persist automatically as user types (saved as null if empty)
Impact: Users can now enter API keys during onboarding and they'll be
available immediately in the app without re-entry.
Deviation: Auto-fix bug (Rule 1) - keys not persisting broke user expectations
* fix(49-03): close FTUX modal and center tutorial message
Issue 1 - FTUX modal still visible after starting tutorial:
- Add setFTUXCompleted(true) to handleStartTutorial
- Ensures modal won't re-render even with testing override
- Modal now properly closes before tutorial starts
Issue 2 - Tutorial message overlapping floating action bar:
- Change TutorialMessage position from bottom-4 to center of screen
- Use top-1/2 left-1/2 -translate-x-1/2 -translate-y-1/2 for true center
- Message now displays prominently in center, not at bottom
- Update comment: "contextual placement" → "center of screen for maximum visibility"
Before: bottom-4 left-1/2 -translate-x-1/2 (bottom center)
After: top-1/2 left-1/2 -translate-x-1/2 -translate-y-1/2 (true center)
* fix(49-03): close WelcomeModal when starting tutorial
Problem: WelcomeModal (original quickstart modal with "New Project", "Load Workflow",
etc.) remained visible when user clicked "Show me how" to start tutorial, blocking the
tutorial UI.
Root cause: WelcomeModal is controlled by workflowStore.showQuickstart state and was not
being closed when tutorial started.
Solution:
- Import setShowQuickstart from useWorkflowStore
- Add setShowQuickstart(false) to handleStartTutorial
- WelcomeModal now closes before tutorial overlay appears
Now when tutorial starts, both FTUXModal and WelcomeModal are closed, allowing tutorial
overlay to display without obstruction.
* fix(49-03): implement complete tutorial progression flow
Problem: Tutorial stuck on "Let's cook." message without progressing.
Root causes:
1. Step 1 (welcome) had no requiredAction, so action detection never triggered
2. Steps 3-5 were placeholder steps with no content or actions
3. No auto-advance logic for informational steps
Solution:
ftuxStore.ts - Replace placeholder steps with real tutorial:
- Step 1: "Let's cook." (welcome, auto-advances after 3s)
- Step 2: "Click the Image button to add an image node." (highlights image button, requires add-image-node)
- Step 3: "Now add an Output node to see your results." (highlights output button, requires add-output-node)
- Step 4: "You're all set! Connect nodes to build workflows, or press Cmd+Enter to run." (completion, auto-advances)
- Add "add-output-node" to TutorialStep requiredAction type
TutorialOverlay.tsx - Add auto-advance and output detection:
- Steps without requiredAction now auto-advance after 3 seconds (typewriter completes ~2s)
- Add detection for "add-output-node" action (checks for node.type === "output")
- Early return cleanup timer to prevent memory leaks
Tutorial now flows: Welcome → Add Image → Add Output → Complete
Each step progresses based on user action or auto-advances for info steps.
* fix(49-03): make tutorial highlighted elements clickable and visible
Problem 1: Floating action bar was darkened during tutorial even though it
contains the highlighted image button that user needs to see and click.
Problem 2: Image button was not clickable - Layer 3 "clickable window" was
blocking clicks instead of allowing them through.
Root cause: ElementHighlight had full-screen dimmed overlay (Layer 1) blocking
clicks, and Layer 3 created an empty div that blocked the actual element.
Solution - Use box-shadow cutout technique:
- Remove solid dimmed overlay (bg-black/60)
- Remove Layer 3 entirely (not needed)
- Use box-shadow: "0 0 0 9999px rgba(0, 0, 0, 0.6)" on positioned div
- Creates dimmed overlay with cutout hole where highlighted element is
- Highlighted element remains fully visible (not dimmed)
- Highlighted element is fully clickable (no overlay blocking it)
- Everything else gets dimmed
FloatingActionBar.tsx:
- Import useFTUXStore (prepared for future use)
- Add tutorial state awareness (not currently used but ready for expansion)
ElementHighlight.tsx:
- Replace full-screen overlay + clickable window with box-shadow cutout
- Simpler 2-layer approach: dimmed background with hole + blue ring
- All pointer-events-none, so nothing blocks the actual DOM element
* fix(49-03): resolve Zustand SSR hydration issue in FloatingActionBar
Error: "The result of getServerSnapshot should be cached to avoid an infinite loop"
at FloatingActionBar (src/components/FloatingActionBar.tsx:345:58)
Same issue as WorkflowCanvas - direct useFTUXStore hook call during SSR.
Solution (identical pattern to WorkflowCanvas fix):
- Replace direct useFTUXStore hook with useState + useEffect pattern
- Initialize with false defaults (SSR-safe, no hydration mismatch)
- Subscribe to ftuxStore on client-side only via useEffect
- Initialize state with getState() after subscription
- Return cleanup function to unsubscribe on unmount
Before: const { tutorialActive, lockedFeatures } = useFTUXStore(...)
After: useState + useEffect subscription (client-side only)
This prevents the SSR hydration error while maintaining reactive updates.
* chore(49-03): restore production FTUX localStorage check
- Remove testing override that forced FTUX to show every time
- Restore production code: if (!getFTUXCompleted()) { setShowFTUX(true); }
- Remove TODO comment about reverting
- FTUX now only shows on first visit as intended
* feat(tutorial): add connection drop menu workflow steps
Extend the FTUX tutorial with 4 new steps that teach users how to use the
connection drop menu workflow pattern:
Step 6: Drag from output handle
- Prompts user to drag from the image node's output handle
- Tracks drag start via onConnectStart handler
Step 7: Release in empty space
- Prompts user to release to show the connection drop menu
- Tracks menu appearance via handleConnectEnd
Step 8: Explain connection menu
- Highlights the drop menu with blue ring
- Explains it shows compatible nodes
- Requires click to continue
Step 9: Select Generate Image
- Highlights "Generate Image" button in menu
- Tracks nanoBanana selection via handleMenuSelect
- Only advances when selected from menu (not FloatingActionBar)
Step 10: Tutorial complete
- Shows completion message with workflow tips
- Auto-advances after 3s and marks FTUX done
Implementation details:
- Add 3 new requiredAction types to TutorialStep interface
- Add 3 state tracking flags to FTUXState (connectionDragStarted, etc.)
- Add tutorial tracking in WorkflowCanvas connection lifecycle
- Add data-tutorial attributes for highlighting menu elements
- Import OnConnectStart type and create handleConnectStart handler
- Track menu appearance and nanoBanana selection with tutorialActive guards
Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>
* refactor(tutorial): simplify drag-and-drop flow
Combine drag and drop instructions into a single step and immediately
highlight the Generate Image button when the context menu appears:
Changes:
- Combine step 6 (drag) and step 7 (drop) into one: "Drag from the output
handle and drop into empty space"
- Remove step 8 (explain menu) - go straight to highlighting Generate Image
- Step 6 now waits for menu to appear (show-connection-menu)
- Step 7 highlights Generate Image button immediately (add-nanoBanana-from-menu)
Removed unused tracking:
- Remove start-connection-drag requiredAction type
- Remove connectionDragStarted state flag and setter
- Remove handleConnectStart callback from WorkflowCanvas
- Remove onConnectStart prop from ReactFlow
- Remove OnConnectStart import
New flow is cleaner and more direct:
Step 5: Explain outputs (click to continue)
Step 6: Drag and drop to show menu (auto-advance)
Step 7: Select Generate Image (auto-advance)
Step 8: Complete
Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>
* fix(tutorial): highlight Generate Image button immediately
Combine drag-and-drop and selection steps into one to ensure the Generate
Image button is highlighted as soon as the connection menu appears.
Problem:
- Previous flow had two separate steps (drag-drop, then select)
- 1 second delay between menu appearing and Generate Image highlighting
- Output handle was highlighted while menu was already visible (confusing)
Solution:
- Single step: "Drag from the output handle and drop into empty space. Then
select 'Generate Image'."
- highlightSelector targets the Generate Image button from the start
- ElementHighlight polls every 500ms, so highlights button immediately when
menu appears (no delay)
- Cleaner UX: instruction → action → immediate visual feedback
Removed:
- show-connection-menu requiredAction (no longer needed)
- connectionMenuShown state flag and setter
- Tutorial tracking in handleConnectEnd
New flow:
- Step 5: Explain outputs (click to continue)
- Step 6: Drag, drop, select Generate Image (highlight on button, auto-advance)
- Step 7: Complete
Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>
* fix(tutorial): increase highlight z-index above connection menu
The ElementHighlight component was rendering at z-index 91, but the
ConnectionDropMenu is at z-index 100, causing the highlight ring to
appear behind the menu.
Change: Increase ElementHighlight z-index from 91 to 101
This ensures the pulsing blue ring appears on top of the menu while
still allowing clicks through (pointer-events-none).
Z-index stack:
- ElementHighlight: 101 (above menu)
- ConnectionDropMenu: 100
- Skip tutorial button: 94
- TutorialMessage: 93
- Click overlay: 92
Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>
* fix(tutorial): highlight output handle when prompting drag action
Split the combined step back into two sequential steps with proper highlights
and instant transition between them.
Problem:
- Single step told user to "Drag from the output handle" but highlighted
the Generate Image button (which doesn't exist yet)
- Confusing visual feedback - wrong element highlighted for the instruction
Solution:
- Step 6: "Drag from the output handle and drop into empty space."
- Highlights: node-output-handle
- Action: show-connection-menu
- advanceDelay: 0ms (instant transition)
- Step 7: "Select 'Generate Image' to add an AI image generation node."
- Highlights: generate-image-option
- Action: add-nanoBanana-from-menu
- advanceDelay: 1000ms (default)
New feature: advanceDelay property
- Added to TutorialStep interface
- Configurable delay before advancing to next step after action completes
- Default: 1000ms (preserves existing behavior)
- Step 6 uses 0ms for instant transition when menu appears
Flow:
1. User sees output handle highlighted → drags and drops
2. Menu appears → step advances instantly (0ms delay)
3. Generate Image button immediately highlighted
4. User clicks → tutorial advances after 1s
Restored:
- show-connection-menu requiredAction
- connectionMenuShown state and setter
- Tutorial tracking in handleConnectEnd
Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>
* fix(tutorial): subscribe to ftux state for action detection
The action detection useEffect wasn't re-running when connectionMenuShown
or nanoBananaAddedFromMenu changed because these states weren't in the
dependency array.
Problem:
- User drags and drops → menu appears
- setConnectionMenuShown(true) updates ftuxStore
- But action detection useEffect doesn't re-run (missing dependency)
- Step doesn't advance to highlight Generate Image button
Solution:
- Subscribe to connectionMenuShown and nanoBananaAddedFromMenu at component level
- Use these subscribed values instead of getState() calls
- Add them to useEffect dependency array
Now when these states change:
1. Component re-renders with new values
2. useEffect re-runs due to dependency change
3. Action is detected as completed
4. Step advances with configured delay (0ms for instant transition)
Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>
* feat(tutorial): add prompt node and connection steps
Extend tutorial with steps 8-9 to teach users:
- Adding nodes via FloatingActionBar (different from connection menu)
- Connecting nodes by dragging between existing handles
Changes:
- Add data-tutorial="prompt-button" to Prompt button
- Add data-tutorial="prompt-output-handle" to Prompt output handle
- Add "add-prompt-node" to requiredAction types
- Add step 8: Add Prompt node via FloatingActionBar
- Add step 9: Connect Prompt to Generate Image node
- Update step 10 (complete) message for clarity
- Add action detection for add-prompt-node in TutorialOverlay
Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>
* fix(tutorial): wait for prompt node connection in step 9
Step 9 was advancing immediately because it checked for ANY edges,
but an edge already existed from step 7 (Image → Generate Image).
Changes:
- Add "connect-prompt-node" requiredAction type
- Update step 9 to use "connect-prompt-node" instead of "connect-nodes"
- Add detection logic that specifically checks for edges FROM prompt nodes
- Track edges in TutorialOverlay component state for reactive detection
- Add edges to useEffect dependency array
Now step 9 correctly waits for user to connect the prompt node
before advancing to the completion step.
Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>
* feat: prevent node overlap on creation with automatic collision detection
Implements AABB collision detection to automatically place new nodes in collision-free positions. When users add nodes via FloatingActionBar or keyboard shortcuts, the system now finds the nearest free position using a spiral search pattern (right, down, left, up, diagonals).
Changes:
- Add spatialLayout.ts with rectanglesIntersect() and findNearestFreePosition()
- Update addNode() to use collision detection before placing nodes
- Remove random offset from FloatingActionBar for more deterministic placement
- Use consistent 20px gap (COLLISION_GAP) matching existing STACK_GAP pattern
Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>
* feat(tutorial): extend FTUX with content setup and resource guidance
Add four new tutorial steps to guide users through completing their first workflow:
- Prompt users to add text to the Prompt node
- Prompt users to add an image to the Image node
- Instruct users to run the workflow with Cmd/Ctrl+Enter
- Inform users about saving projects, templates, and Discord community
This ensures users complete a full workflow cycle during onboarding and know where to find help and resources.
Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>
* fix(tutorial): remove emojis from tutorial messages
Remove decorative emojis from the save-and-resources and completion steps for a cleaner, more professional appearance.
Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>
* fix(tutorial): improve viewport zoom timing with auto-populated nodes
Increase timeout from 400ms to 600ms and improve viewport setting logic
to ensure proper 0.7 zoom level when tutorial image node is added with
pre-loaded base64 data. Fetch fresh nodes from store and extend animation
duration for smoother transition.
Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>
* feat(tutorial): delay auto-population until after prompt connection
Change tutorial flow to create empty nodes initially and auto-populate
with sample content only after the prompt node is connected to generate.
This maintains hands-on learning while still removing friction.
Changes:
- Nodes created empty during tutorial
- Auto-populate after "connect-prompt-node" step completes
- Switch sample image: model.png → owl.jpg
- Update prompt: "wearing aviator sunglasses and a leather jacket"
- Add nodesPopulated ref to prevent duplicate population
Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>
* feat(tutorial): improve messaging, dual highlights, and pulsing animation
- Change welcome message: "Let's cook" → "Let's go over the basics"
- Add dual highlight for connect-prompt step: both prompt output and
generate text input handles are now highlighted simultaneously
- Position connect-prompt message on left side of screen
- Update ElementHighlight to support array of selectors
- Add data-tutorial attribute to generate node's text input handle
- Pulsing animation already present via pulse-ring animation
Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>
* feat(tutorial): implement mock execution and organized node demonstration
- Add mock workflow execution to save API costs during tutorial
- Mock execution simulates 5-second generation time with edge animations
- Loads pre-made owl aviator image instead of calling Gemini API
- Provides realistic experience without incurring charges
- Implement visible workflow tree demonstration
- Automatically builds branching workflow after mock execution
- Video branch: Prompt → Generate Video → Output
- LLM branch: Prompt → LLM → Generate Image #2 → Output
- Clean horizontal layout with 350px spacing between nodes
- Vertical separation of ±350px for clear branch distinction
- Improve user experience
- Auto-close run menu when advancing past tutorial steps
- Unlock canvas navigation after demonstration completes
- Zoom to 0.35 with centered viewport for complete tree overview
- Updated LLM prompt for nightclub scene generation
- Add tutorial assets
- Add owl aviator image (public/tutorial/owl-aviator.png)
Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>
* fix: restore .env.example file
Restore .env.example that was accidentally deleted in commit 9be881c.
This file is needed as a reference for users to set up their environment.
Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>
---------
Co-authored-by: Claude Sonnet 4.5 <noreply@anthropic.com>
The handle dot color is set via data-handletype attribute selector in CSS.
Audio handles had no rule, so they fell back to the default appearance.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
isImageInput() has permissive description heuristics (e.g. "data uri")
that false-positive on audio properties like audio_url whose description
mentions "base64 data URI". Reorder the checks so isAudioInput() — which
uses specific name patterns — runs first.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Add audio-to-video to VIDEO_CAPABILITIES, render audio input handles
from model schema with purple color (--handle-color-audio: #a855f7),
and include backward-compatibility hidden audio handle. Audio handles
are positioned between image and text handle groups.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Add AUDIO_INPUT_PATTERNS, isAudioInput() function modeled after
isImageInput(), and integrate into extractParametersFromSchema() to
detect audio inputs from model schemas. Audio inputs are sorted
between image and text inputs.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Add "audio-to-video" to ModelCapability union, extend ModelInput.type
and ModelInputDef.type with "audio", and include "audio-to-video" in
fal.ai RELEVANT_CATEGORIES for model discovery.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- GroupsOverlay: reset color picker when menu closes, guard locked groups
from drag/resize/rename
- AudioInputNode: add keyboard accessibility to dropzone (role, tabIndex,
onKeyDown)
- list-workflows: fix name regex to handle escaped quotes
- GroupsOverlay.test: add missing NBP Input menu item assertion
- WorkflowBrowserView: make relativePath required to match API contract
- WorkflowBrowserModal: add dialog role, aria-modal, aria-labelledby
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- cleanup-workflow.js: Filter only embedded data: images from outputGallery
instead of wiping entire array; preserve imageRef/video on output nodes
- workflow/route.ts: Validate parsed.version is number, nodes/edges are
arrays; sort candidates by mtime desc for deterministic file selection
- mediaStorage.ts: Guard against undefined mediaData in loadMediaById
before caching/returning
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Add recursive directory scanning (up to 3 levels deep) to the
list-workflows API so projects organized into subdirectories are
visible in the workflow browser. Skips hidden dirs and node_modules.
Returns relativePath field displayed as the subtitle in the UI.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Instead of parsing entire workflow JSON files (which can be many MB with
embedded base64 data), read only the first 1KB to extract the name via
regex. Drop nodeCount from the listing. Probe all directories in parallel.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Scans subdirectories of a given parent path for valid workflow JSON files,
returning name, path, node count, and last modified time.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Replace browser file picker with the existing /api/browse-directory native
OS folder picker so loadWorkflow receives the actual filesystem path,
enabling media hydration from sibling inputs/ and generations/ folders.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Use API response imageId instead of locally generated ID in
saveVideoAndGetRef/saveAudioAndGetRef (filename mismatch)
- Read result.video/result.audio instead of result.data in
loadMediaById (response field mismatch)
- Add audio extensions and MIME types to load-generation route
so audio files can be found and served
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Critical bug fix - the mediaStorage.ts functions were using incorrect
parameter names when calling the save-generation and load-generation APIs.
Fixed parameter mappings:
- workflowPath → directoryPath (API expects this name)
- id → imageId (API expects this name)
- data + type → video/audio (API expects media in named fields)
- Changed load from GET to POST with correct body structure
Also increased serverActions bodySizeLimit from 50mb to 100mb to handle
large video files during externalization.
This fixes the save failures when externalizing workflows with videos/audio.
Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>
- Add node-skipped CSS class (opacity 0.35) for skip dimming
- Expose skippedNodeIds in Zustand store for reactive CSS updates
- Apply node-skipped class in WorkflowCanvas alongside switch-dimmed
- Clear skippedNodeIds on workflow stop, reset, and completion
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Nodes with `position: relative` on inner wrappers (e.g. image wrapper
needing it for its remove button) caused left-side handles to render
behind the wrapper due to CSS painting order. Adding z-index: 5 to
all handles ensures they always paint above node content.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Browser hit-testing against large base64 <img> elements is expensive
even when no React state updates occur. Two CSS-level fixes:
1. pointer-events: none on all .react-flow__node img — images never
need direct mouse events (parent containers handle clicks)
2. .canvas-interacting class on <html> during pan/drag disables
pointer-events on all node content via CSS descendant selector,
toggled via direct DOM access (zero React overhead)
https://claude.ai/code/session_01MvD1n4QeXutgwUpKJuDGHa
- Add video, audio, text, 3d, easeCurve entries to EDGE_COLORS in SharedEdgeGradients
- Normalize handle type suffixes (e.g. image-0 -> image) in EditableEdge color lookup
- Gate Kie provider filter behind API key check (return 400 if missing)
- Fix getMediaDimensions to check URL pathname for image extensions before defaulting to video
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
fal was unconditionally listed as available, which could surface
fal models to users without a fal API key configured.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
The capabilities prop doesn't exist on ModelSearchDialogProps. Use the
correct initialCapabilityFilter="image" prop instead.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
The `.react-flow__node:not(.switch-dimmed)` rule applied a transition to
every node DOM element unnecessarily. The `.switch-dimmed` rule already
handles the transition for dimming. Removing this makes undimming instant
(snappier) and eliminates a transition declaration from every node.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Add beforeunload handler that prompts the user when they try to close
the tab or navigate away while there are unsaved changes.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
jiajia