Release v2.0.1

Previous: v2.0.0
Commits: 62

Changes

• docs: make Pages and dev docs feature-centric
• docs: drop version-listing What's New section from README
• fix: sync runtime test inventory
• docs: add CDP UI persistence evidence
• docs: close QA evidence gaps
• docs: record QA execution evidence
• fix: preserve video metadata in sequence history
• fix: avoid injection-like agy pipeline prompt
• fix: document Gemini providers in CLI help
• fix: harden provider paths and CLI metadata
• fix: preserve video thumbnail fallbacks in history UI
• feat: centralize recursive thumbnail backfill
• docs: add Computer Use QA plan for v2.0.0 audit
• feat: add Grok video model picker (V / V1.5) to video controls
• security: atomic config.json writes in keys routes
• security: atomic token write + scrub codex env + harden vertex init
• fix: reap orphaned codex device-auth child on abandoned flow
• fix: advertise Gemini models in capabilities + document Vertex size limit
• fix: sync Gemini auth-mode dropdown + surface key-delete errors
• security: cap sharp input pixels to prevent decompression bombs
• feat: provider plumbing, CLI parity, prompt studio + contract tests
• feat: Grok model/size pickers + adapter updates
• feat: image/video thumbnails + history sidebar cards
• feat: Gemini/Vertex API key management
• feat: add open-again + copy-link buttons to Switch Account waiting state
• fix: skip GPT pixel-limit size confirm for Grok/Gemini providers
• fix: correct Gemini image cost to official pricing
• fix: aspect ratio 4 rows full-width matching GPT SIZE layout
• fix: aspect ratio 4-col grid (3 rows) for spacious layout
• fix: resolution buttons 2-col 2-row grid layout
• fix: restore original Nano/Banana model button labels
• fix: aspect ratio buttons 5-col grid with proper sizing
• fix: unify Gemini API controls styling with GPT/Grok format
• fix: Vertex AI integration — auth mode persistence, role field, camelCase config
• fix: skip response_format for Vertex AI (unsupported field)
• fix: respect last-saved Gemini auth mode (apikey vs vertex)
• fix: prefer Vertex AI over API key when both configured
• security: audit fixes — crypto session IDs, session cap, double-click guard, key in header
• fix: split Gemini model trigger into two lines (nb2 / api)
• feat: add Gemini API model picker (Nano Banana 2 / Nano Banana Pro)
• fix: split Gemini cost estimation by model (Flash vs Pro)
• fix: Gemini API controls polish — grid layout, NxN preview, cost, model labels
• fix: display Gemini model+provider as single label in history
• feat: add Gemini API aspect ratio + resolution controls
• fix: use protobuf enum values for Gemini API aspectRatio/imageSize
• feat: pass aspectRatio + imageSize to Gemini API generateContent
• fix: provider→model auto-switch for agy/gemini-api
• fix: split Gemini models into 3 lines — agy / api / api pro
• fix: allow API key override even when loaded from env
• feat: add Switch Account buttons for Grok + Codex

Release v2.0.0

Previous: v1.1.23
Commits: 1

Changes

• dd

Preview Release v1.1.24-preview.20260602110035

Base: 1.1.23 → preview patch 1.1.24
Commits since v2.0.0: 0

Changes

Release v1.1.23

Previous: v1.1.22
Commits: 20

Changes

• Revert "docs(site): add compressed demo video (37MB)"
• docs(site): add compressed demo video (37MB)
• docs(site): add one-click install guide to quickstart and homepage
• fix: remove grok-composer-2.5-fast planner (no image input support)
• docs(site): update GitHub Pages for v1.1.22
• fix(docs): update stale grok-4.3 references → grok-composer-2.5-fast
• docs: production-ready v1.1.22 release notes + feature documentation
• fix: remove reasoning effort from video settings (Grok doesn't support it)
• fix: Active prompt guidance text → English (was Korean in UI)
• fix(ui): video settings — compact pill selectors for planner + effort
• feat(ui): video settings — planner model + reasoning effort + stronger storyboard
• feat: strengthen storyboard prompts with production-grade rules
• fix: add storyboard field to VideoGenerateRequest + NodeGenerateRequest types
• fix: resolve all audit findings — storyboard payload, dead code, CLI flag
• feat: storyboard mode — UI toggle + prompt injection for image and video
• docs: strengthen i2v skill guide — composition, decision tree, storyboard
• feat: planner multi-character dialogue visual-ID rules
• feat: CLI --planner-model flag for ima2 video
• feat: add GPT models to planner model dropdown
• feat: switch Grok planner default to grok-composer-2.5-fast, add settings UI for model selection

Release v1.1.22

Previous: v1.1.21
Commits: 43

Changes

• chore: sync bin/ima2.js — process tree kill, error handling, SIGBREAK
• fix: remove unused isGrok/provider vars (TS6133)
• fix(ui): reasoning effort always visible in model selector
• fix(ui): remove fixed width/height/flex from compact web search toggle
• fix(ui): make save button fill grid cell, fix web search sizing
• fix(ui): show reasoning effort for all models including Grok
• fix(ui): add composer__tool class to compact web search toggle
• fix(ui): composer toolbar as 3-column grid — 2 rows of 3 buttons
• fix(ui): remove forced toolbar break so all composer buttons flow in one row
• fix: Windows SIGBREAK handler + exclude .ts from npm package
• fix(ui): use flex: 1 1 auto for composer toolbar buttons
• fix(ui): stretch composer toolbar buttons to fill row width
• fix(site): replace stale npx in Korean docs index with npm install -g
• chore: sync JS build artifacts for grokProxyLauncher and oauthLauncher
• fix(ui): use DRAG handle for images too — pointer capture blocks inline draggable
• fix(ui): right panel tabs 2-row grid layout
• fix(ui): enlarge video drag handle button
• fix(ui): make right panel tabs flex with equal width and center text
• fix(ui): replace drag handle SVG icon with plain DRAG text button
• fix(ui): add dedicated drag handle for canvas video, fix video drag
• feat(ui): enable prompt builder for default profile, show result actions in prompt studio, add canvas image drag-to-composer
• fix(ui): use video URL for all frame extraction buttons
• fix(ui): fix video frame copy buttons, cap toasts at 5 with 3s dismiss
• fix(ui): save button label — 저장/Save
• fix(ui): bookmark button → text label 프롬/Prom
• fix(ui): composer toolbar two-row layout + hint above buttons
• fix(ui): restore text label on 이어가기/Continue button in composer
• fix(ui): video frame buttons — two-line centered labels
• fix: remove unused extractLastFrame import (TS6133)
• feat(ui): video frame copy buttons — 처음/중간/마지막
• fix: memory leak reapers + global error handlers
• ci: skip SIGINT graceful shutdown test on Windows
• fix: deep audit — path traversal, fetch timeout, atomic sidecar writes
• fix(ui): clear existing refs and prompt before continue-from-item, zero-pad quota reset hours
• fix: cross-platform audit — 5 issues patched
• fix(ui): replace wrong Grok SVG with actual Grok logo from cli-jaw
• fix(ui): add provider SVG icons to quota cards, show full date+time on reset labels
• fix(ui): strip composer toolbar labels to SVG-only, reorganize model selector into Image/Video sections with GPT-IMAGE-2 and Grok Imagine sub-groups
• docs(ko): add one-click install and update sections to Korean README
• ci: add install script sync check + pages trigger for scripts/
• feat: cross-platform install scripts + CI graceful shutdown test
• fix: graceful shutdown releases file handles on Windows (EBUSY fix)
• feat(ui): add Codex quota bars and Grok usage link to settings

Release v1.1.20

Previous: v1.1.19
Commits: 13

Changes

• docs(cli): update grok login help text for --manual-paste default
• docs: video provider expansion research + rebuild grok.js
• fix(grok): default to --manual-paste for login (most reliable cross-platform)
• chore: regenerate test inventory after adding agent-video-intent tests
• feat(agent-video): fix Korean intent detection + add test fixtures
• fix(windows): remove device-code auto-fallback that invalidates user's code
• feat(video): skill documentation + agent param parsing from prompt
• fix(video-series): store topic from videoSeries in node metadata
• fix(i18n): add missing video.seriesTopic translation keys
• ci: add nix flake check job
• docs: add Contributors section — credit @aorying and all contributors
• fix(nix): correct progrok version in comment (0.1.0 → 0.1.1)
• Add Nix flake: reproducible node_modules, CLI via nix run, and a test check

Release v1.1.19

Previous: v1.1.18
Commits: 27

Changes

• chore: rebuild JS outputs (video.js, capabilities.js)
• fix(node-video): prevent drag when interacting with video controls
• chore: bump progrok 0.1.0 → 0.1.1 (device-code default + manual-paste)
• feat(node-video): classic UX parity — metadata, prompt, topic, recovery
• feat(node-video): store and display video metadata (duration, resolution)
• assets: use video gallery screenshot for Video card (more impressive than spinner)
• assets: replace hero screenshot with video gallery view (more impressive)
• fix(node-video): V2V extracts last frame from parent video as sourceImage
• assets: add real screenshots for setup wizard, video generation, Grok provider, and video gallery
• fix(node-video): use sourceFilename instead of URL path for parent I2V
• fix: tsc build errors in runVideoGenerate node update
• feat(agent): auto I2V — use last session image as video source
• docs: update Japanese and Chinese READMEs for v1.1.18 (Video/Grok)
• docs(site): update install lede — Codex → ChatGPT/Grok
• docs(site): update quickstart and overview pages for v1.1.18
• fix(agent): render video in agent workspace via AgentSafeImage
• feat: video integration parity across node/agent/CLI
• fix(node-mode): video not rendering + parent image not used as reference
• fix(node-mode): spinner not visible during image generation
• feat: trash fallback + video series chain + agent video tool
• fix(node): route video generation in generateNodeVariation + runNodeBatch
• fix(video): guard model param type for backward compat with old localStorage
• chore: remove internal plan file from repo
• chore: remove devlog-map from tracked files
• chore: remove devlog/ from git (already in .gitignore, keep local only)
• test: add grok-imagine-video-1.5-preview model validation + payload tests
• feat(video): grokv1.5 T2V via white canvas injection (I2V workaround)

Release v1.1.18

Previous: v1.1.17
Commits: 15

Changes

• fix(ui): show video-specific metadata (duration, resolution, aspect ratio)
• fix: model selector layout overlap + stabilize flaky port test
• fix(ui): copy image converts to PNG for clipboard compatibility
• test: align agent-mode test with classic-default change
• feat(video): restore grokv1.5 with auto-fallback to v1.0 on failure
• fix(video): remove grokv1.5-preview (T2V unsupported, causes 400 error)
• fix(windows): use shell:true for OAuth proxy spawn (same fix as grok)
• fix(ui): default to Classic mode for new users (was Agent)
• ci: add video --help + capabilities smoke tests on all platforms
• docs(site): add video generation to GitHub Pages (EN + KO)
• docs: add video generation to README, CLI.md, and API.md
• feat(cli): add 'ima2 video' command + capabilities video section
• feat(cli): add ima2 video command for video generation via SSE
• feat(ui): distinguish video model selection (grokv1.0 vs grokv1.5)
• feat: 4-way setup wizard (GPT/Grok/Both/API) + grok-imagine-video-1.5-preview

Release v1.1.17

Previous: v1.1.16
Commits: 20

Changes

• fix(oauth): stop restart loop when openai-oauth is unavailable
• ci: limit package-install smoke to ubuntu (prepack needs unix shell)
• ci: run package-install smoke + CLI smoke on all OS/node matrix
• fix(windows): use absolute path to progrok binary instead of PATH lookup
• fix: Windows grokProxyLauncher rootDir + test regex alignment
• fix(grok): sync grok.js with grok.ts (bin/ excluded from tsc build)
• fix: video phase persisted server-side + Windows grok login quoting fix
• fix(inflight): preserve local phase during reconcile with server
• feat(ui): show planning/streaming phase in video in-flight queue
• feat(ui): drag gallery thumbnails to prompt area as reference (no prompt injection)
• Revert "revert(planner): rollback to original prompts, keep visible text language rule"
• revert(planner): rollback to original prompts, keep visible text language rule
• feat(planner): add content policy + visible text rules from oauth prompts
• feat(planner): rewrite system prompts with Grok Imagine best practices
• fix(grok): auto-fallback to device-code flow when browser login fails
• fix(video): pass requestId to server to prevent duplicate in-flight entries
• fix(video-progress): show progress only in InFlightList, integrate animate with in-flight tracking
• fix(ui): hide 0% progress label until actual progress arrives
• fix(video): normalize progress from 0-100 to 0-1 for UI consumption
• feat(video): continue-here extracts last frame from video as reference

Release v1.1.16

Previous: v1.1.15
Commits: 8

Changes

• fix(css): video controls buttons flex-fill like image settings
• fix(css): center-align video controls option rows
• feat(ui): show video progress % in in-flight queue list
• fix: show video generation in inflight queue (spinner)
• fix(video-ui): rename label grok (v) → grokv, fix aspect ratio button overflow
• feat(video): video model selector + ref2v compose surface (Classic+Node), progress %
• feat(video): add reference-to-video backend (count-derived mode, reference_images payload)
• chore: regenerate runtime test inventory