@remixmate/cli 0.13.6 → 0.13.8

package/dist/manifest.json

@@ -1,7 +1,7 @@
 {
   "schemaVersion": 1,
-  "version": "0.13.6",
-  "generatedAt": "2026-10-30T07:14:13.248Z",
+  "version": "0.13.8",
+  "generatedAt": "2026-10-31T00:45:42.523Z",
   "skills": [
     {
       "id": "doc-parse",
@@ -614,7 +614,7 @@
               "on",
               "off"
             ],
-            "description": "Narration on / off, overriding the template's needsNarration (auto = the template decides). Templates where narration is optional (image sequences, galton-board) default to none — **must** pass `on` when the replicate card says `旁白: <名称>（narrated）`, otherwise the narration you write is dropped without an error and the video comes back silent. `off` drops narration from a narrated template."
+            "description": "Narration on / off, overriding the template's needsNarration (auto = the template decides). Templates declaring narration optional default to none — **must** pass `on` when the replicate card says `旁白: <名称>（narrated）`, otherwise the narration you write is dropped without an error and the video comes back silent. `off` drops narration from a narrated template."
           },
           "background_asset": {
             "type": "string",

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@remixmate/cli",
-  "version": "0.13.6",
+  "version": "0.13.8",
   "description": "AI media generation skills for Claude Code / Codex — 12 skills covering image, video, voice, digital human, web screenshot, web recording, script, template registry, rendering, Jianying export, and video deconstruction.",
   "type": "module",
   "main": "dist/index.js",

package/skills/gen-script/SKILL.md

@@ -491,7 +491,7 @@ python3 <SkillDir>/scripts/gen_script.py \
 | `--payload KEY=VALUE` | Template-specific card choice, repeatable. `KEY` is a dotted path in `customPayloadSchema` (or `capabilities.frameLevelKeys`), e.g. `<field>=<value>`, `<object>.<field>=<value>`. Value is converted to the declared type (true/false, numbers) and checked against enum / range; undeclared keys are rejected. Written to every scene with sibling keys kept, applied before fit-caption duration estimation. | template default |
 | `--ending-payload KEY=VALUE` | Same as `--payload`, but written only to the ending scene, e.g. `<object>.<field>=<closing line>`. On countdown templates it also appends an ending scene after #1 (it carries the #1 index; fill its narration and the item field `capabilities.countdown.itemProp` names). Errors when the script has no ending scene. | none |
 | `--avatar-id` | Digital-human avatar id (`dh_xxx`) for templates declaring `capabilities.digitalHuman`. Adds asset `asset-avatar` (`type: avatar`, `source: gen-digital-human`, `payload.avatarId`) and sets `visuals.avatar.assetRef` on every scene whose slot reads it — one continuous presenter driven by the narration. Placement: template default, or `--payload digitalHuman.<field>=…` (a photo-only presenter is `--payload digitalHuman.posterUrl=<url>` without this flag). | none |
-| `--motion` | Camera motion for every scene (`animationHints.motion`): `static` / `kenburns-in` / `kenburns-out` / `pan-left` / `pan-right` / `float`. `auto` removes the motion from every scene so the template picks (image-slide rotates push / pan / pull / pan); nothing is locked then. | skeleton alternates in / out |
+| `--motion` | Camera motion for every scene (`animationHints.motion`): `static` / `kenburns-in` / `kenburns-out` / `pan-left` / `pan-right` / `float`. `auto` removes the motion from every scene so the template picks its own motion; nothing is locked then. | skeleton alternates in / out |
 | `--font-id` | Font library key (`font.uniq_id`), written to `global.font.fontId`; ab-render resolves and installs the real font at render time. Pass it whenever the user picked a font — the replicate card shows it as `名字（id）`, e.g. `字体: <名字>（<uniqId>）`; pass the id in the parentheses. Never pass a CSS family or the display name: an unknown key falls back to the template font with no error. | template font |
 | `--font-name` | Display name of the picked font → `global.font.fontName` (display only, never used for rendering). Pass it with `--font-id`. | — |
 | `--list-fonts` | Print the live font library (`/font/page`: id, name, lang, category, license) and exit; no script is generated and `--topic` is not needed. The ids it prints are the only valid `--font-id` values — use it when the user asks which fonts exist or names a font without its id. | — |
@@ -510,6 +510,8 @@ python3 <SkillDir>/scripts/gen_script.py \
 
 Values written by `--background-preset`, `--background-asset`, `--payload`, `--ending-payload`, `--avatar-id` and `--motion` are listed in `meta.lockedScenePaths` (scene-relative paths such as `customPayload.<object>.<field>`, `animationHints.motion`). When the agent later saves its own `dsl_json` in merge mode, ab-agent restores those paths from this skeleton after merging, so a model-written `customPayload` object cannot wipe a card choice. Do not hand-write these keys in `dsl_json`; pass the flag. Templates that author their own scenes (`meta.authoredScenes`) are not merged: there ab-agent writes each locked value that the skeleton holds on every scene (with one value) onto every submitted scene, and adds the skeleton assets those values reference (`asset-background`, `asset-avatar`) when the submitted `assets` lacks them.
 
+For replicate cards exposing transition effects or duration, read the template's `customPayloadSchema` and `llmHint` for the exact field paths, units, allowed values and any paired override switch. Pass each explicit selection through `--payload <declared-path>=<value>`, preserving the chosen preset and typography; following the sample omits the overrides. These payload keys are locked into every generated scene, so do not re-author them in a saved DSL.
+
 ## DSL generation principles
 
 Apply the following principles when producing the DSL:

package/skills/gen-script/scripts/gen_script.py

@@ -643,6 +643,8 @@ def _template_needs_image(template_config: dict | None) -> bool:
 
 
 
+
+
 
 
 
@@ -707,6 +709,7 @@ def _resolve_contract(template_config: dict | None) -> dict:
         "duration_strategy": duration_strategy,
         "duration_model": caps.get("durationModel"),
         "payload_defaults": payload_defaults,
+        "media_binding": caps.get("mediaBinding"),
         "primary_visual_type": primary_visual_type,
         "background_strategy": "shared" if caps.get("backgroundStrategy") == "shared" else "per-scene",
         "default_layout": caps.get("defaultLayout"),
@@ -1275,6 +1278,8 @@ def _media_option_choice(option_schema: dict, value: str | None, flag: str,
 
 
 
+
+
 
 _MEDIA_OPTION_FLAGS = {"preset": "--sequence-preset", "fit": "--sequence-fit"}
 _LEGACY_MEDIA_BINDINGS = {
@@ -2493,7 +2498,7 @@ def _attach_narration(scene: dict, narration_text: str, narration_asset_id: str
 def _visual_scene(
         ctx: _SkeletonContext, idx: int, scene_id: str, plan: dict, visual_asset_id: str,
 ) -> dict:
-    return {
+    scene = {
         "id": scene_id,
         "purpose": plan["purpose"],
         "duration": plan["duration"],
@@ -2507,6 +2512,24 @@ def _visual_scene(
     }
 
 
+
+    binding = ctx.contract.get("media_binding")
+    if isinstance(binding, dict) and "assetRefKey" in binding:
+        path = binding.get("itemsPath")
+        ref_key = binding.get("assetRefKey")
+        if not isinstance(path, str) or not all(path.split(".")) or not isinstance(ref_key, str) or not ref_key:
+            raise ValueError("mediaBinding needs non-empty itemsPath and assetRefKey")
+        item = deepcopy(binding.get("item", {}))
+        if not isinstance(item, dict):
+            raise ValueError("mediaBinding.item must be an object")
+        item[ref_key] = visual_asset_id
+        payload = _build_custom_payload(ctx.template_config)
+        _write_payload(payload, [(path.split("."), [item])])
+        scene["customPayload"] = payload
+        del scene["visuals"]
+    return scene
+
+
 def _payload_scene(ctx: _SkeletonContext, scene_id: str, plan: dict) -> dict:
 
 
@@ -2979,7 +3002,7 @@ Examples:
         "--motion", default=None,
         help="Camera motion for every scene (animationHints.motion): static / kenburns-in / kenburns-out / "
              "pan-left / pan-right / float. Replaces the skeleton's alternating default. `auto` removes the "
-             "motion from every scene so the template picks (image-slide rotates push / pan / pull / pan).",
+             "motion from every scene so the template picks its own motion.",
     )
     parser.add_argument("--allow-digital-human", action="store_true", help="Allow digital-human assets")
     parser.add_argument("--avatar-id", default=None, help="Digital-human avatar id (dh_xxx) for templates declaring capabilities.digitalHuman: adds one gen-digital-human asset referenced by every scene's visuals.avatar.assetRef (one continuous presenter driven by the narration). Placement stays the template default unless set with --payload digitalHuman.<field>=...")

package/skills/gen-script/skill.json

@@ -67,7 +67,7 @@
       "narration": {
         "type": "string",
         "enum": ["auto", "on", "off"],
-        "description": "Narration on / off, overriding the template's needsNarration (auto = the template decides). Templates where narration is optional (image sequences, galton-board) default to none — **must** pass `on` when the replicate card says `旁白: <名称>（narrated）`, otherwise the narration you write is dropped without an error and the video comes back silent. `off` drops narration from a narrated template."
+        "description": "Narration on / off, overriding the template's needsNarration (auto = the template decides). Templates declaring narration optional default to none — **must** pass `on` when the replicate card says `旁白: <名称>（narrated）`, otherwise the narration you write is dropped without an error and the video comes back silent. `off` drops narration from a narrated template."
       },
       "background_asset": {
         "type": "string",

package/skills/gen-script/version.json

@@ -2,6 +2,6 @@
   "skillName": "gen-script",
   "repoName": "agent-skill-media-maker",
   "skillId": "474",
-  "version": "V24",
+  "version": "V26",
   "skillDescription": "Video-script generation skill. Turns a user-supplied topic into a structured Video DSL (JSON) that describes the full video — scene structure, asset requirements, and narrative flow.\n\nUse this skill as soon as the user mentions any of these intents:\n- Write a video script, generate a video script, plan a video, write storyboards\n- Create a short video, plan video content, generate a Video DSL\n- Turn a topic into a video structure / video content plan\n\nEven when the user does not say \"generate the DSL\", use this skill whenever they want to turn a topic into a structured video plan.\n\n⚠️ Stop-and-confirm gate: after this skill returns a DSL, show the full script and wait for the user's explicit confirmation. Never call `prepare_video_assets` in the same turn."
 }

package/skills/render-video/scripts/_narration_alignment.py

@@ -11,6 +11,8 @@ import math
 import re
 import unicodedata
 
+from video_dsl.runtime.timeline_compiler import assign_sentence_indices, split_narration_sentences
+
 
 def _normalized(text: str) -> str:
     text = re.sub(r"<#\s*\d+(?:\.\d+)?\s*#>", "", text)
@@ -86,3 +88,41 @@ def align_narration_lines(
         end = start_ms + (end_ms - start_ms) * offset / total
         result.append({"text": line, "timeBegin": begin, "timeEnd": end})
     return result, "estimated"
+
+
+def restore_narration_sentences(
+    text: str, subtitles: list[dict], duration_ms: float, speech_span_ms=None,
+) -> tuple[list[dict], str]:
+
+
+
+
+
+
+    sentences = split_narration_sentences(text)
+    if not subtitles:
+        if len(sentences) > 1:
+            return align_narration_lines(sentences, [], duration_ms, speech_span_ms)
+        return [], "tts-unchanged"
+
+    result = []
+    method = "tts-unchanged"
+    for sub in subtitles:
+        parts = split_narration_sentences(sub.get("text") or "")
+        if len(parts) <= 1:
+            result.append(sub)
+            continue
+        split, timing = align_narration_lines(parts, [sub], duration_ms, speech_span_ms)
+        result.extend(split)
+        if method != "estimated":
+            method = "estimated" if timing == "estimated" else "tts-interpolated"
+
+    if len(sentences) > 1:
+        if _normalized(text) != _normalized("".join(s.get("text") or "" for s in subtitles)):
+            return result, "tts-text-mismatch"
+        indices = assign_sentence_indices([s.get("text") or "" for s in result])
+        if len(set(indices)) != len(sentences):
+
+
+            return align_narration_lines(sentences, subtitles, duration_ms, speech_span_ms)
+    return result, method

package/skills/render-video/scripts/render_video.py

@@ -880,6 +880,9 @@ def build_render_plan(dsl: dict, binding: dict) -> dict:
             "retryCount": 0,
             "maxRetries": 3,
         }
+        narration_text = (asset.get("payload") or {}).get("text")
+        if asset.get("type") == "audio" and isinstance(narration_text, str) and narration_text.strip():
+            assets_by_id[asset["assetId"]]["narrationText"] = narration_text
 
 
 
@@ -1397,6 +1400,8 @@ def _resolve_single_asset(
 
 
             src_payload = asset_with_payload.get("payload", {}) or {}
+            if source == "gen-voice" and isinstance(src_payload.get("text"), str):
+                asset["narrationText"] = src_payload["text"]
             if isinstance(src_payload.get("narrationItems"), list):
                 asset["narrationLineCount"] = len(src_payload["narrationItems"])
                 asset["narrationIntroLines"] = int(src_payload.get("narrationIntroLines") or 0)
@@ -1628,6 +1633,8 @@ def resolve_narration_run(
         if per_scene_subs[k]:
             asset["ttsSubtitles"] = per_scene_subs[k]
         payload = payloads.get(c["assetId"]) or {}
+        if isinstance(payload.get("text"), str):
+            asset["narrationText"] = payload["text"]
         if isinstance(payload.get("narrationItems"), list):
             asset["narrationLineCount"] = len(payload["narrationItems"])
             asset["narrationIntroLines"] = int(payload.get("narrationIntroLines") or 0)
@@ -1990,32 +1997,38 @@ def _template_default_tail_pad_frames(tpl: dict, fps: int) -> int | None:
 
 def _structured_narration_subtitles(entry: dict, asset: dict, render_plan: dict) -> list[dict]:
 
-    from _narration_alignment import align_narration_lines
+    from _narration_alignment import align_narration_lines, restore_narration_sentences
 
     subtitles = asset.get("ttsSubtitles") or []
     line_count = asset.get("narrationLineCount")
-    if not isinstance(line_count, int) or isinstance(line_count, bool) or line_count <= 0:
-        return subtitles
-
 
     old_subs = entry.get("subtitleSegments") or []
     at_sec = entry.get("subtitleAtSec") or []
     if len(at_sec) == len(old_subs) and any(isinstance(v, (int, float)) for v in at_sec):
         return []
 
-    lines = asset.get("narrationLines")
-    if not isinstance(lines, list) or len(lines) != line_count:
-        lines = [s.get("text") for s in old_subs] if len(old_subs) == line_count else []
-    if len(lines) != line_count or any(not isinstance(v, str) or not v.strip() for v in lines):
-        method = "authored-lines-unavailable"
-        message = "Authored narration lines unavailable; cannot restore per-line timing"
+    if not isinstance(line_count, int) or isinstance(line_count, bool) or line_count <= 0:
+        text = asset.get("narrationText") or ""
+        subtitles, method = restore_narration_sentences(
+            text, subtitles, asset["duration"], asset.get("speechSpanMs"),
+        )
+        if method == "tts-unchanged":
+            return subtitles
+        message = f"Restored narration sentence boundaries using {method} timing"
     else:
+        lines = asset.get("narrationLines")
+        if not isinstance(lines, list) or len(lines) != line_count:
+            lines = [s.get("text") for s in old_subs] if len(old_subs) == line_count else []
+        if len(lines) != line_count or any(not isinstance(v, str) or not v.strip() for v in lines):
+            method = "authored-lines-unavailable"
+            message = "Authored narration lines unavailable; cannot restore per-line timing"
+        else:
 
-        asset["narrationLines"] = list(lines)
-        subtitles, method = align_narration_lines(
-            lines, subtitles, asset["duration"], asset.get("speechSpanMs"),
-        )
-        message = f"Preserved {line_count} narration lines using {method} timing"
+            asset["narrationLines"] = list(lines)
+            subtitles, method = align_narration_lines(
+                lines, subtitles, asset["duration"], asset.get("speechSpanMs"),
+            )
+            message = f"Preserved {line_count} narration lines using {method} timing"
 
     logs = render_plan.setdefault("logs", [])
     if not any(l.get("phase") == "narration-align" and l.get("sceneId") == entry.get("sceneId")

package/skills/render-video/version.json

@@ -2,6 +2,6 @@
   "skillName": "render-video",
   "repoName": "agent-skill-media-maker",
   "skillId": "473",
-  "version": "V42",
+  "version": "V43",
   "skillDescription": "Final-render skill: loads a persisted RenderPlan by `job_id` and drives the Remotion engine to produce the final video.\n\nUse this skill as soon as the user mentions any of these intents (after assets are already prepared):\n- Render the video, composite the video, export the video\n- Turn the prepared assets into the final clip\n- Render with Remotion\n\nPrerequisite: assets must already be generated via `prepare_video_assets`. This skill never resolves or regenerates assets — pass it a `job_id` from a previous `prepare_video_assets` call.\n\n⚠️ Stop-and-confirm gate: never call this skill until the user has explicitly confirmed the assets prepared by `prepare_video_assets`. If those assets were prepared in the current turn and the user has not replied since, stop and ask instead of rendering."
 }

package/skills/template-registry/version.json

@@ -2,6 +2,6 @@
   "skillName": "template-registry",
   "repoName": "agent-skill-media-maker",
   "skillId": "475",
-  "version": "V30",
+  "version": "V32",
   "skillDescription": "Video-template registry skill. Stores every video-template definition and lists the available templates (templateId / name / aspect ratio / style tags).\n\nUse this skill as soon as the user mentions any of these intents:\n- View available templates / list every template\n\nNote: DSL→TemplateBinding is no longer a separate exposed step — once prepare_video_assets receives a template_id it builds the binding internally."
 }

package/skills/template-registry/video_dsl/runtime/dsl_validator.py

@@ -1273,6 +1273,44 @@ def _check_narration_language_strict(dsl: dict) -> list[ValidationError]:
 _TEXT_LAYER_SKELETON_PREFIXES = ("[骨架待填充]", "[skeleton placeholder]")
 
 
+def _check_presenter_visibility(dsl: dict) -> list[ValidationError]:
+
+
+
+
+    tpl = _get_template_cfg(_pick_template_id(dsl)) or {}
+    caps = tpl.get("capabilities")
+    caps = caps if isinstance(caps, dict) else {}
+    spec = caps.get("digitalHuman") or {}
+    if not isinstance(spec, dict) or spec.get("backgroundOccludes") is not True:
+        return []
+    presenter_scenes = []
+    for scene in dsl.get("scenes") or []:
+        if not isinstance(scene, dict):
+            continue
+        visuals = scene.get("visuals") or {}
+        if not isinstance(visuals, dict):
+            continue
+        if not isinstance(visuals.get("avatar"), dict):
+            continue
+        if (visuals.get("avatar") or {}).get("assetRef"):
+            presenter_scenes.append(scene)
+    if not presenter_scenes or not all(
+        isinstance(scene["visuals"].get("background"), dict)
+        and scene["visuals"]["background"].get("assetRef")
+        for scene in presenter_scenes
+    ):
+        return []
+    binding = caps.get("mediaBinding") or {}
+    path = binding.get("itemsPath") if isinstance(binding, dict) else None
+    hint = f" Use timed media items in customPayload.{path} instead." if path else " Leave presenter scenes without full-scene backgrounds."
+    return [ValidationError(
+        "scenes[].visuals.background.assetRef",
+        "Full-scene backgrounds cover every presenter scene; the presenter would never be visible."
+        + hint,
+    )]
+
+
 def _check_text_layer_rules(dsl: dict) -> list[ValidationError]:
 
 
@@ -1398,6 +1436,7 @@ def validate_integrity(dsl: dict) -> list[ValidationError]:
     errors += _check_step_narration_alignment(dsl)
     errors += _check_slide_cross_field(dsl)
     errors += _check_text_layer_rules(dsl)
+    errors += _check_presenter_visibility(dsl)
 
     errors += _check_duration_estimate(dsl)
     return errors
@@ -1435,6 +1474,7 @@ def validate_dsl(dsl: dict) -> list[ValidationError]:
     errors += _check_step_narration_alignment(dsl)
     errors += _check_slide_cross_field(dsl)
     errors += _check_text_layer_rules(dsl)
+    errors += _check_presenter_visibility(dsl)
 
     errors += _check_duration_estimate(dsl)
     return errors

package/skills/template-registry/video_dsl/runtime/timeline_compiler.py

@@ -213,15 +213,19 @@ def assign_sentence_indices(raw_texts: list[str]) -> list[int]:
     return indices
 
 
+def split_narration_sentences(text: str) -> list[str]:
+
+
+
+    return [part.strip(" \t\r") for part in _PRIMARY_BREAK.split(text.strip()) if part.strip()] if text else []
+
+
 def count_narration_sentences(text: str) -> int:
 
 
 
 
-    if not text or not text.strip():
-        return 0
-    chunks = [s for s in _PRIMARY_BREAK.split(text.strip()) if s.strip()]
-    return len(chunks)
+    return len(split_narration_sentences(text))
 
 
 def _plain_len(text: str) -> int: