## Highlights - **`setMulmoErrorFormatter` injection point**: Host apps can now register a formatter that turns the schema validation error into a readable summary before `initializeContextFromFiles` logs it. Default behavior is unchanged when no formatter is registered or the formatter returns `null` (PR #1373) 📦 **npm**: - [`mulmocast@2.6.13`](https://www.npmjs.com/package/mulmocast/v/2.6.13) - [`@mulmocast/types@2.6.13`](https://www.npmjs.com/package/@mulmocast/types/v/2.6.13) --- ## What's Changed * feat: add MulmoErrorFormatter injection point for schema validation errors by @isamu in https://github.com/receptron/mulmocast-cli/pull/1373 * mulmocast@2.6.13 by @isamu in https://github.com/receptron/mulmocast-cli/pull/1374 **Full Changelog**: https://github.com/receptron/mulmocast-cli/compare/2.6.12...2.6.13

## Highlights - **Replicate image models**: Added Flux 2 Pro/Dev, Flux 1.1 Pro/Pro Ultra, Flux Pro/Dev/Schnell, Ideogram v3 (turbo/balanced/quality), Recraft v3, Stable Diffusion 3.5 Large, and Luma Photon. Generalized agent output handling to accept FileOutput, Array<FileOutput>, string URL, or `{ url: string }` shapes (PR #1362) - **Replicate movie models**: Added `alibaba/happyhorse-1.0`, `minimax/hailuo-2.3`, `minimax/hailuo-2.3-fast`, `pixverse/pixverse-v5`. Added `start_image_required` flag for i2v-only models (PR #1364, #1365) - **Vertex AI usability**: Clarified auth error message for Google GenAI (suggests both `GEMINI_API_KEY` and `vertexai_project` paths). Warn when `gemini-3-pro-image-preview` is used on Vertex AI without `location: "global"` (PR #1361) - **Docs**: Added Gemini API vs Vertex AI comparison and links to `vertexai_en.md` / `vertexai_ja.md` in `docs/README.md` (PR #1360) **Samples**: - [test_flux2.json](https://github.com/receptron/mulmocast-cli/blob/main/scripts/test/test_flux2.json) - [test_replicate_images.json](https://github.com/receptron/mulmocast-cli/blob/main/scripts/test/test_replicate_images.json) - [test_replicate_new_movies.json](https://github.com/receptron/mulmocast-cli/blob/main/scripts/test/test_replicate_new_movies.json) 📦 **npm**: - [`mulmocast@2.6.10`](https://www.npmjs.com/package/mulmocast/v/2.6.10) - [`@mulmocast/types@2.6.10`](https://www.npmjs.com/package/@mulmocast/types/v/2.6.10) --- ## What's Changed * docs: add Gemini API vs Vertex AI section to docs/README.md by @isamu in https://github.com/receptron/mulmocast-cli/pull/1360 * fix(genai): clarify auth error and warn on gemini-3-pro-image-preview region by @isamu in https://github.com/receptron/mulmocast-cli/pull/1361 * Replicate movie: i2v gate, tests, and sample fixes (#1364 followups) by @ystknsh in https://github.com/receptron/mulmocast-cli/pull/1365 * feat(replicate): support Flux/Ideogram/Recraft/SD3.5/Photon image models by @isamu in https://github.com/receptron/mulmocast-cli/pull/1362 * feat(replicate): add hailuo-2.3, happyhorse-1.0, pixverse-v5 movie models by @isamu in https://github.com/receptron/mulmocast-cli/pull/1364 **Full Changelog**: https://github.com/receptron/mulmocast-cli/compare/2.6.9...2.6.10

## Highlights - **imagePrompt reference images**: `referenceImageName` and `referenceImage` fields for image-to-image generation in `imageParams.images` — reference other imageRefs keys or specify direct source (path/url/base64) 📦 **npm**: [`mulmocast@2.6.5`](https://www.npmjs.com/package/mulmocast/v/2.6.5) --- ## What's Changed * Update package20260321 by @isamu in https://github.com/receptron/mulmocast-cli/pull/1316 * feat: add reference image support for imagePrompt in imageParams.images by @isamu in https://github.com/receptron/mulmocast-cli/pull/1315 **Full Changelog**: https://github.com/receptron/mulmocast-cli/compare/2.6.4...2.6.5

## Highlights - **Animation Runtime Extraction**: Extracted animation runtime JS from HTML template into separate, testable files (`animation_runtime.js`, `data_attribute_registration.js`, `auto_render.js`) - **Data-Attribute Declarative Animations**: New `[data-animation]` HTML attributes for declaring animations without writing script code - **Cover Pan/Zoom**: New `coverPan()` and `coverZoom()` animation helpers for Ken Burns-style photo effects - **Per-Image canvasSize**: `imagePrompt` reference images now support optional `canvasSize` for per-image aspect ratio override - **Browser JS Quality**: ESLint, Prettier, and CI syntax checks for `assets/html/js/` files; var→const/let modernization 📦 **npm**: [`mulmocast@2.5.0`](https://www.npmjs.com/package/mulmocast/v/2.5.0) --- ## What's Changed * update by @isamu in https://github.com/receptron/mulmocast-cli/pull/1273 * Add coverPan/coverZoom helpers with normalized pan semantics by @ystknsh in https://github.com/receptron/mulmocast-cli/pull/1276 * feat: add comprehensive animated patterns CI test by @isamu in https://github.com/receptron/mulmocast-cli/pull/1279 * chore(deps): bump undici from 7.22.0 to 7.24.1 by @dependabot[bot] in https://github.com/receptron/mulmocast-cli/pull/1282 * feat: data-attribute declarative animations for html_tailwind by @isamu in https://github.com/receptron/mulmocast-cli/pull/1284 * refactor: extract animation runtime into separate JS files by @isamu in https://github.com/receptron/mulmocast-cli/pull/1285 * feat: add optional canvasSize to imagePrompt reference images by @isamu in https://github.com/receptron/mulmocast-cli/pull/1286 * update by @isamu in https://github.com/receptron/mulmocast-cli/pull/1287 **Full Changelog**: https://github.com/receptron/mulmocast-cli/compare/2.4.9...2.5.0

## Highlights - **Video-Audio Drift Fix**: Fixed cumulative audio-video desync caused by FFmpeg's `trim=duration` rounding up to frame boundaries (~0.03s per beat). Now uses frame-exact trimming with cumulative frame tracking (Bresenham-style) for precise synchronization. 📦 **npm**: [`mulmocast@2.4.3`](https://www.npmjs.com/package/mulmocast/v/2.4.3) --- ## What's Changed * feat: add html_tailwind animation guidance to story skill by @isamu in https://github.com/receptron/mulmocast-cli/pull/1251 * fix: prevent video-audio drift from FFmpeg frame quantization by @isamu in https://github.com/receptron/mulmocast-cli/pull/1253 **Full Changelog**: https://github.com/receptron/mulmocast-cli/compare/2.4.2...2.4.3

## Highlights - **Slide Branding**: スライドマスターのようにロゴ・背景画像をグローバル設定し、全スライドに自動適用する機能を追加。per-beatで無効化(`branding: null`)やオーバーライドも可能 - **bgOpacity**: スライド背景色の透明度を制御し、背景画像を透過表示する`bgOpacity`オプションを追加 📦 **npm**: [`mulmocast@2.2.6`](https://www.npmjs.com/package/mulmocast/v/2.2.6) 📦 **npm**: [`@mulmocast/types@2.3.0`](https://www.npmjs.com/package/@mulmocast/types/v/2.3.0) --- ## What's Changed * feat: add branding support to slide plugin by @isamu in https://github.com/receptron/mulmocast-cli/pull/1214 **Full Changelog**: https://github.com/receptron/mulmocast-cli/compare/2.2.5...2.2.6

## Highlights - **Card layout vertical spacing fix**: Removed `justify-center` from comparison/columns card layouts to flow content naturally from top, with footer pinned to bottom via `mt-auto` - **Card height optimization**: Changed `items-stretch` to `items-start` so cards size to their content instead of stretching to fill available space - **Slide layout improvements**: Improved vertical centering and spacing across slide layouts - **Story skill**: Added `/story` skill for structured multi-phase MulmoScript creation - **Slide reference**: Added reference field to slide media for beat-level citations - **Slide chart & mermaid**: Added chart and mermaid content block types to slide DSL - **TTS speed option**: Added TTS speed configuration support 📦 **npm**: [`mulmocast@2.2.2`](https://www.npmjs.com/package/mulmocast/v/2.2.2) --- ## What's Changed * fix: correct CLI command typo in slide skill (image → images) by @ystknsh in https://github.com/receptron/mulmocast-cli/pull/1191 * feat: add /story skill for structured MulmoScript creation by @isamu in https://github.com/receptron/mulmocast-cli/pull/1193 * update by @isamu in https://github.com/receptron/mulmocast-cli/pull/1195 * feat: add reference field to slide media by @isamu in https://github.com/receptron/mulmocast-cli/pull/1197 * Add TTS speed parameter test script by @ystknsh in https://github.com/receptron/mulmocast-cli/pull/1199 * chore(deps): bump hono from 4.11.9 to 4.12.0 by @dependabot[bot] in https://github.com/receptron/mulmocast-cli/pull/1198 * update by @isamu in https://github.com/receptron/mulmocast-cli/pull/1200 * Improve slide layout centering and enhance /story skill by @isamu in https://github.com/receptron/mulmocast-cli/pull/1201 * Add sample MulmoScript presentations from /story skill by @isamu in https://github.com/receptron/mulmocast-cli/pull/1202 * fix: remove justify-center from card layouts to fix vertical spacing by @isamu in https://github.com/receptron/mulmocast-cli/pull/1203 * update by @isamu in https://github.com/receptron/mulmocast-cli/pull/1204 **Full Changelog**: https://github.com/receptron/mulmocast-cli/compare/2.2.1...2.2.2

## Highlights - **Slide imageRef Content Block**: New `imageRef` content block type that references images from `imageRefs` in MulmoScript, enabling reusable image assets across slide layouts - **Chart Content Block**: Embed Chart.js charts (bar, pie, line, etc.) directly inside slide content blocks with conditional CDN loading - **Mermaid Content Block**: Embed Mermaid diagrams (flowcharts, sequence diagrams, etc.) inside slide content blocks with automatic dark/light theme detection **Sample**: [`test_slide_chart_mermaid.json`](https://github.com/receptron/mulmocast-cli/blob/main/scripts/test/test_slide_chart_mermaid.json) | [`test_slide_image_ref.json`](https://github.com/receptron/mulmocast-cli/blob/main/scripts/test/test_slide_image_ref.json) 📦 **npm**: [`mulmocast@2.2.0`](https://www.npmjs.com/package/mulmocast/v/2.2.0) --- ## What's Changed * feat: add chart and mermaid content block types to slide DSL by @isamu in https://github.com/receptron/mulmocast-cli/pull/1189 * feat: slide imageRef content block for referencing imageParams.images by @isamu in https://github.com/receptron/mulmocast-cli/pull/1188 **Full Changelog**: https://github.com/receptron/mulmocast-cli/compare/2.1.40...2.2.0