@koda-sl/baker-cli 0.250.0-dev.e7a1a227e → 0.250.1-dev.e7a1a227e

package/README.md

@@ -3480,6 +3480,7 @@ Reach for it when you are relaying the **client's** own words. When you are writ
 
 ```bash
 baker studio animate "slow push in, model turns to camera and smiles" --image j57abc123def456ghi789
+baker studio animate "she looks to camera and says: \"Hola, soy Elena\"" --avatar elena --quality 720p --aspect-ratio 9:16
 baker studio animate "handheld drift right, steam rising from the cup" \
   --image ./out/hero.png --duration 6 --quality 1080p
 baker studio animate "product rotates once on a turntable" --image j57abc…,j57def… --from references
@@ -5790,6 +5791,7 @@ This CLI is designed for AI agent consumption. Key patterns:
 
 - **0.217.0**: new family `baker avatars list|get|create|update|delete`, plus `--avatar <handle>` on `baker studio generate` and `baker studio animate` — company-scoped reusable presenters, typed from the shared `@baker/api` avatars contract. `--avatar` is the way to cast one: it grounds the render on the identity sheet, reuses the subject description verbatim, and records the avatar on the batch. Passing the sheet through `--reference` instead does none of those and leaves the batch unattributed, so the avatar can never show what it has appeared in; the route refuses a handle that is not `ready`, because generating a stranger is worse than an error. `get` returns the sheet URL and that description; every command's `hints` carry the next move (still building → re-check with `get`; changed `--subject` without `--regenerate-sheet` → the sheet is stale). `create` refuses `--source-image` without `--likeness-confirmed` and returns an `error.fix` telling the agent to ask the user for that permission, since only they can give it; an unknown handle returns an `error.fix` pointing at `baker avatars list`. Profile flags on `update` merge over the stored profile instead of replacing it.
 
+- **0.250.1**: `--avatar <handle>` on `baker studio generate` and `baker studio animate` never worked. Both were declared in `registerSchema` and read at the call site, but neither was in the command's own citty `args` — and citty does not reject an undeclared flag, it keeps it and sets it to `true`. So `--avatar elena` reached the route as `avatar: true` with `elena` dropped as a stray positional, and the body schema (`z.string()`) refused it. Documented since 0.217.0, broken since 0.217.0: casting an avatar into a render was not possible from the CLI at all, which is why a clip of a named presenter came back with neither the identity sheet nor the voice. Both flags are now declared. A new test in `command-reference-integrity` compares every registered schema against the args of the command that parses it, so the two cannot drift again — it found a third case on the way, `actions status`, whose schema advertised a `--refs` flag for what the command actually takes as positionals.
 ## Publishing
 
 ### Auto-publish (CI)

package/dist/cli.js

@@ -7411,7 +7411,11 @@ registerSchema({
   command: "actions.status",
   description: "Resolve one or more Work Action refs by real action ID or temp_* ref in a single batch call. When BAKER_CHAT_ID is set, a temp_* ref still staged in THIS chat resolves to status 'draft' (not 'not_found') \u2014 staged ops only become published actions on chat publish.",
   args: {
-    refs: { type: "string", description: "One or more action refs: real action IDs or temp_* refs", required: true }
+    ref: {
+      type: "positional",
+      description: "One or more action refs: real action IDs or temp_* refs (space-separated)",
+      required: true
+    }
   }
 });
 var statusCommand = defineCommand11({
@@ -29926,7 +29930,7 @@ function buildVideoTodo(report, overlayCount, floatingCount, opts, blueprint) {
       ]
     },
     transitions: "Scene-to-scene cuts the deconstruct flagged as fade/whip/zoom/dissolve/swipe are reproduced as an ffmpeg xfade at the boundary (everything else stays a hard cut). The overlap is consumed from extra generated footage, so the picture stays exactly on the audio timeline. To change a transition, edit the scene's `transition_out.type` in prompt.json and re-scaffold, or hand-edit the `spine` node's ffmpeg args. For a richer HERO cut (whip-pan, glitch, light-leak, gravitational-lens\u2026), the overlay layer can run a Hyperframes shader/CSS transition instead \u2014 see references/hyperframes/blueprints-and-transitions.md (pick 2\u20133 transition types total; the motion IS the handoff).",
-    overlay_capabilities: 'The overlay layer (video-overlay-composition/index.html) is a REAL Hyperframes composition, not a plain text layer \u2014 our hyperframe_render node runs the genuine `npx hyperframes` renderer. So it can do far more than fade/slide/pop: (1) PULL a ready-made block \u2014 `npx hyperframes catalog --type block` then `npx hyperframes add <id>` into the composition dir, and nest it with a <div data-composition-src="compositions/<id>.html" data-start data-track-index data-width data-height> clip (~97 blocks: lower-thirds, social-proof cards, stat counters, charts, code, logo stings); (2) animate with the named GSAP motion-rule vocabulary (kinetic-beat-slam, svg-path-draw, counting-dynamic-scale, multi-phase-camera\u2026); (3) frame a talking head as a video-call/PIP/split. The whole engine + how to build new effects is documented in references/hyperframes/ (start at README.md \u2192 catalog.md, motion-rules.md). Reach for this whenever the reference\'s graphics are richer than plain text.',
+    overlay_capabilities: 'The overlay layer (video-overlay-composition/index.html) is a REAL Hyperframes composition, not a plain text layer \u2014 our hyperframe_render node runs the genuine `npx hyperframes` renderer. So it can do far more than fade/slide/pop: (1) PULL a ready-made block \u2014 `npx hyperframes catalog --type block` then `npx hyperframes add <id>` into the composition dir, and nest it with a <div data-composition-src="compositions/<id>.html" data-start data-track-index data-width data-height> clip (~154 blocks: lower-thirds, social-proof cards, stat counters, charts, code, logo stings); (2) animate with the named GSAP motion-rule vocabulary (kinetic-beat-slam, svg-path-draw, counting-dynamic-scale, multi-phase-camera\u2026); (3) frame a talking head as a video-call/PIP/split. The whole engine + how to build new effects is documented in references/hyperframes/ (start at README.md \u2192 catalog.md, motion-rules.md). Reach for this whenever the reference\'s graphics are richer than plain text.',
     caption_system: "Sound-off feed = burned-in captions carry the message. The deconstruct produced a word-level transcript \u2014 drive a real caption track off it (karaoke highlight \u2192 themed \u2192 kinetic), not hand-typed text. Floor: `npx hyperframes add caption-highlight` (TikTok karaoke); escalate by content register (caption-kinetic-slam for hype, caption-editorial-emphasis for premium, caption-neon-glow for AI/crypto). Group on meaning, sync within 80ms. Full ladder + transcript contract: references/hyperframes/captions-and-audio.md.",
     talking_head_overlay: "For a scene where a presenter shows something (screen-share, app demo, product, a 'video call' look), frame the talking-head clip in the overlay composition: presenter full-bleed \u2192 shrinks to a rounded corner PIP pill while a content card takes the stage (or split/stack/overlay-glass-card). The pattern, the #video-wrap\u2192corner-pill tween, the glass call card, and the corner-coord presets are in references/hyperframes/talking-head-and-overlays.md. A ready exemplar ships as the `video-call-composition` move \u2014 copy it next to your canvas, drop brand fonts in the dir, point a `hyperframe_render` at it with the presenter clip as its `background`.",
     text_overlays: {
@@ -48249,10 +48253,15 @@ function costHintsFor(body) {
 var animateCommand = defineCommand189({
   meta: {
     name: "animate",
-    description: "Render a clip. With an image the look is already fixed, so the prompt describes MOVEMENT \u2014 what the camera does, what the subject does, in what order. With --from text there is no image and the prompt is the whole shot.\n\nA rendered clip is NOT usable anywhere until you keep it: `baker studio keep <id> --slot N` is what puts it in the video library. Takes nobody keeps are never ingested, which is what makes a rejected batch cheap.\n\nFor anything longer than 15 seconds, or to build on footage that already exists, use --model bytedance/seedance-2.5: it renders 4-30s and is the only model that reads an existing clip or an existing soundtrack.\n\nExamples:\n  baker studio animate 'slow push in, model turns to camera and smiles' --image j57abc123def456ghi789\n  baker studio animate 'handheld drift right, steam rising from the cup' --image './out/hero.png' --duration 6 --quality 1080p\n  baker studio animate 'product rotates once on a turntable' --image j57abc\u2026,j57def\u2026 --from references\n  baker studio animate 'she keeps walking, camera stays with her, then she stops and looks up' --image j57abc\u2026 --from references --from-clip j57batch\u2026:0 --model bytedance/seedance-2.5 --duration 20\n  baker studio animate 'hold on the product, then a slow push-in' --from references --from-video j57vid\u2026 --model bytedance/seedance-2.5\n  baker studio animate 'slow drone pull-back over a solar farm at golden hour, no people' --from text --model bytedance/seedance-2.5 --duration 12"
+    description: "Render a clip. With an image the look is already fixed, so the prompt describes MOVEMENT \u2014 what the camera does, what the subject does, in what order. With --from text there is no image and the prompt is the whole shot.\n\nA rendered clip is NOT usable anywhere until you keep it: `baker studio keep <id> --slot N` is what puts it in the video library. Takes nobody keeps are never ingested, which is what makes a rejected batch cheap.\n\nFor anything longer than 15 seconds, or to build on footage that already exists, use --model bytedance/seedance-2.5: it renders 4-30s and is the only model that reads an existing clip or an existing soundtrack.\n\nExamples:\n  baker studio animate 'slow push in, model turns to camera and smiles' --image j57abc123def456ghi789\n  baker studio animate 'she looks to camera and says: \"Hola, soy Elena\"' --avatar elena --quality 720p --aspect-ratio 9:16\n  baker studio animate 'handheld drift right, steam rising from the cup' --image './out/hero.png' --duration 6 --quality 1080p\n  baker studio animate 'product rotates once on a turntable' --image j57abc\u2026,j57def\u2026 --from references\n  baker studio animate 'she keeps walking, camera stays with her, then she stops and looks up' --image j57abc\u2026 --from references --from-clip j57batch\u2026:0 --model bytedance/seedance-2.5 --duration 20\n  baker studio animate 'hold on the product, then a slow push-in' --from references --from-video j57vid\u2026 --model bytedance/seedance-2.5\n  baker studio animate 'slow drone pull-back over a solar farm at golden hour, no people' --from text --model bytedance/seedance-2.5 --duration 12"
   },
   args: {
     prompt: { type: "positional", description: "What MOVES", required: false },
+    avatar: {
+      type: "string",
+      description: "Cast an avatar by short name \u2014 grounds the clip on their identity sheet AND gives it their voice",
+      required: false
+    },
     image: {
       type: "string",
       description: "Comma-separated library image ids, local file paths, and/or image URLs",
@@ -48507,6 +48516,11 @@ var generateCommand = defineCommand190({
     description: "Start here to make an image. Renders 1-8 takes of one brief, ingests each into the media library as it lands, and shows the batch in the dashboard Studio next to the ones the client ran.\n\nModel choice: google/gemini-3.1-flash-image-preview (default \u2014 fast, best at editing a reference and at extreme ratios), google/gemini-3-pro-image-preview (highest fidelity, slower), openai/gpt-image-2 (photoreal and the cleanest in-image text \u2014 no --image-size, no 4:5 / 5:4), recraft/recraft-v4.1-pro-vector (vector/flat marks with palette control).\n\n--reference is the biggest quality lever there is: a real logo, product shot, Pinterest pin or sandbox screenshot beats any amount of adjectives.\n\nExamples:\n  baker studio generate 'matte black bottle on wet marble, hard studio light, 35mm' --aspect-ratio 3:2 --count 3\n  baker studio generate 'this bottle on a sunlit kitchen counter' --reference './src/brand/product.png,https://\u2026/kitchen.jpg'\n  baker studio generate 'founder-style selfie, kitchen background, natural light' --skill ugc-selfie-hook\n  baker studio generate 'flat geometric mascot, brand palette' --model recraft/recraft-v4.1-pro-vector --rgb-colors '[[10,10,10],[255,80,0]]'"
   },
   args: {
+    avatar: {
+      type: "string",
+      description: "Cast an avatar by short name \u2014 grounds the image on their identity sheet",
+      required: false
+    },
     prompt: { type: "positional", description: "What to generate", required: false },
     model: { type: "string", description: "Model id (default google/gemini-3.1-flash-image-preview)", required: false },
     "aspect-ratio": { type: "string", description: "Aspect ratio (default 1:1)", required: false },