@remixmate/cli 0.13.9 → 0.13.10

package/dist/handlers/gen-digital-human.js

@@ -1,7 +1,7 @@
-var U=Object.defineProperty;var w=(t,e)=>U(t,"name",{value:e,configurable:!0});import{mmPost as h,pollUntil as y,resolveHttpContext as b,SkillError as i}from"../http.js";import{getCapabilities as N}from"../capabilities.js";import{emitProgress as v}from"../progress.js";import{isTrue as g,toNumber as I}from"./shared.js";const E=["video","image"],k=["hifly","heygen"],S="hifly";function c(t){return t==null?"":String(t).trim()}w(c,"toId");async function M(t,e){const o=g(e.mine),n=!o&&g(e.all),a=!o&&!n,r={current:1,pageSize:100};a?r.featured=!0:n&&(r.source=c(e.source)||S),o&&(r.mine=!0),e.gender&&(r.gender=String(e.gender)),e.name&&(r.name=String(e.name));const u=await h(t,"/digital-human/avatar/page",r,{timeoutMs:3e4}),l=u.list??[];if(g(e.json_output)){process.stdout.write(JSON.stringify({avatars:l})+`
-`);return}if(l.length===0){process.stdout.write(a?`⚠️  No curated public avatars match. Try --all for the full inventory, or --mine for your own.
+var b=Object.defineProperty;var u=(t,e)=>b(t,"name",{value:e,configurable:!0});import{mmPost as p,pollUntil as S,resolveHttpContext as M,SkillError as i}from"../http.js";import{getCapabilities as N}from"../capabilities.js";import{emitProgress as v}from"../progress.js";import{isTrue as g,toNumber as I}from"./shared.js";const E=["video","image"],k=["hifly","heygen"],y="hifly";function c(t){return t==null?"":String(t).trim()}u(c,"toId");async function O(t,e){const r=g(e.mine),a=!r&&g(e.all),n=!r&&!a,o={current:1,pageSize:100};n?o.featured=!0:a&&(o.source=c(e.source)||y),r&&(o.mine=!0),e.gender&&(o.gender=String(e.gender)),e.name&&(o.name=String(e.name));const f=await p(t,"/digital-human/avatar/page",o,{timeoutMs:3e4}),l=f.list??[];if(g(e.json_output)){process.stdout.write(JSON.stringify({avatars:l})+`
+`);return}if(l.length===0){process.stdout.write(n?`⚠️  No curated public avatars match. Try --all for the full inventory, or --mine for your own.
 `:`⚠️  No avatars available
-`);return}if(a){process.stdout.write(`
+`);return}if(n){process.stdout.write(`
 ${"ID".padEnd(24)} ${"Name".padEnd(16)} ${"Gender".padEnd(8)} ${"Ratio".padEnd(6)} ${"Scenes".padEnd(22)} Voice
 `),process.stdout.write(`${"-".repeat(24)} ${"-".repeat(16)} ${"-".repeat(8)} ${"-".repeat(6)} ${"-".repeat(22)} ${"-".repeat(12)}
 `);for(const d of l)process.stdout.write(`${c(d.id).padEnd(24)} ${String(d.name??"").padEnd(16)} ${String(d.gender??"").padEnd(8)} ${String(d.aspectRatio??"").padEnd(6)} ${(d.tags??[]).join(",").padEnd(22)} ${d.defaultVoiceId||"-"}
@@ -9,31 +9,31 @@ ${"ID".padEnd(24)} ${"Name".padEnd(16)} ${"Gender".padEnd(8)} ${"Ratio".padEnd(6
 ${"ID".padEnd(24)} ${"Name".padEnd(16)} ${"Source".padEnd(14)} ${"Gender".padEnd(8)} ${"Ratio".padEnd(6)} Clone
 `),process.stdout.write(`${"-".repeat(24)} ${"-".repeat(16)} ${"-".repeat(14)} ${"-".repeat(8)} ${"-".repeat(6)} ${"-".repeat(8)}
 `);for(const d of l)process.stdout.write(`${c(d.id).padEnd(24)} ${String(d.name??"").padEnd(16)} ${String(d.source??"").padEnd(14)} ${String(d.gender??"").padEnd(8)} ${String(d.aspectRatio??"").padEnd(6)} ${d.cloneStatus||"-"}
-`)}const s=u.pager?.total??l.length;process.stdout.write(`
+`)}const s=f.pager?.total??l.length;process.stdout.write(`
 ${l.length} avatar(s) shown${s>l.length?` of ${s}; narrow with --name`:""}
-`),process.stdout.write(a?`💡 Use --avatar-id <ID> to pick an avatar; pass its Voice as --voice-id for a matching voice
+`),process.stdout.write(n?`💡 Use --avatar-id <ID> to pick an avatar; pass its Voice as --voice-id for a matching voice
 `:`💡 Use --avatar-id <ID> to pick an avatar for generation
-`)}w(M,"listAvatars");async function O(t,e){for(const o of[{mine:!0},{featured:!0}]){const a=((await h(t,"/digital-human/avatar/page",{current:1,pageSize:100,...o},{timeoutMs:3e4})).list??[]).find(r=>c(r.id)===e);if(a)return a}return{id:e,source:S}}w(O,"fetchAvatarInfo");function R(t,e){const o=(t.source??"").trim()||S;if(!e.includes(o))throw new i(`❌ avatar uses provider "${o}", which is not available (supported: ${e.join("/")}). Pick another avatar from --list-avatars.`);return o}w(R,"resolveSource");async function T(t,e){const o=c(e.clone_url);if(!/^https?:\/\//i.test(o))throw new i("❌ --clone needs --clone-url <public http(s) URL of a video or portrait image> (local files are not uploaded)");const n=c(e.name);if(!n)throw new i("❌ --clone needs --name <avatar name>");const a=c(e.clone_type)||"video";if(!E.includes(a))throw new i(`❌ --clone-type must be one of ${E.join(" / ")}, received: ${a}`);const r=c(e.clone_provider)||S;if(!k.includes(r))throw new i(`❌ --clone-provider must be one of ${k.join(" / ")}, received: ${r}`);if(r==="heygen"&&a!=="image")throw new i("❌ HeyGen avatars are created from a photo only; add --clone-type image");const u={name:n,sourceType:a,fileUrl:o};r!==S&&(u.provider=r),e.thumbnail_url&&(u.thumbnailUrl=c(e.thumbnail_url)),e.gender&&(u.gender=c(e.gender)),v({phase:"gen-digital-human:clone-request",sourceType:a,provider:r});const l=await h(t,"/digital-human/avatar/clone",u,{timeoutMs:6e4}),s=c(l.id);if(!s)throw new i(`❌ could not parse the new avatar id; API returned: ${JSON.stringify(l)}`);v({phase:"gen-digital-human:clone-async",avatarId:s});const d=g(e.json_output);if(e.wait!==void 0&&!g(e.wait)){if(d){process.stdout.write(JSON.stringify({avatarId:s,cloneStatus:l.cloneStatus??"cloning"})+`
+`)}u(O,"listAvatars");async function R(t,e){for(const r of[{mine:!0},{featured:!0}]){const n=((await p(t,"/digital-human/avatar/page",{current:1,pageSize:100,...r},{timeoutMs:3e4})).list??[]).find(o=>c(o.id)===e);if(n)return n}return{id:e,source:y}}u(R,"fetchAvatarInfo");function L(t,e){const r=(t.source??"").trim()||y;if(!e.includes(r))throw new i(`❌ avatar uses provider "${r}", which is not available (supported: ${e.join("/")}). Pick another avatar from --list-avatars.`);return r}u(L,"resolveSource");async function T(t,e){const r=c(e.clone_url);if(!/^https?:\/\//i.test(r))throw new i("❌ --clone needs --clone-url <public http(s) URL of a video or portrait image> (local files are not uploaded)");const a=c(e.name);if(!a)throw new i("❌ --clone needs --name <avatar name>");const n=c(e.clone_type)||"video";if(!E.includes(n))throw new i(`❌ --clone-type must be one of ${E.join(" / ")}, received: ${n}`);const o=c(e.clone_provider)||y;if(!k.includes(o))throw new i(`❌ --clone-provider must be one of ${k.join(" / ")}, received: ${o}`);if(o==="heygen"&&n!=="image")throw new i("❌ HeyGen avatars are created from a photo only; add --clone-type image");const f={name:a,sourceType:n,fileUrl:r};o!==y&&(f.provider=o),e.thumbnail_url&&(f.thumbnailUrl=c(e.thumbnail_url)),e.gender&&(f.gender=c(e.gender)),v({phase:"gen-digital-human:clone-request",sourceType:n,provider:o});const l=await p(t,"/digital-human/avatar/clone",f,{timeoutMs:6e4}),s=c(l.id);if(!s)throw new i(`❌ could not parse the new avatar id; API returned: ${JSON.stringify(l)}`);v({phase:"gen-digital-human:clone-async",avatarId:s});const d=g(e.json_output);if(e.wait!==void 0&&!g(e.wait)){if(d){process.stdout.write(JSON.stringify({avatarId:s,cloneStatus:l.cloneStatus??"cloning"})+`
 `);return}process.stdout.write(`
 🧬 Clone submitted, avatar id: ${s}
 `),process.stdout.write(`💡 Check progress with --clone-status --avatar-id ${s}
-`);return}let m;try{m=await y(async()=>{const f=await h(t,"/digital-human/avatar/clone-status",{id:s},{timeoutMs:3e4}),p=f.cloneStatus??"unknown";if(v({phase:"gen-digital-human:clone-poll",avatarId:s,status:p}),p==="done")return f;if(p==="failed")throw new i(`❌ avatar clone failed: ${f.cloneError||"unknown error"}`);return null},{intervalMs:1e4,timeoutMs:20*6e4})}catch(f){throw f instanceof i&&/polling timed out/.test(f.message)?new i(`❌ avatar ${s} is still cloning; rerun --clone-status --avatar-id ${s} later`):f}if(d){process.stdout.write(JSON.stringify({avatarId:s,cloneStatus:m.cloneStatus})+`
+`);return}let m;try{m=await S(async()=>{const w=await p(t,"/digital-human/avatar/clone-status",{id:s},{timeoutMs:3e4}),$=w.cloneStatus??"unknown";if(v({phase:"gen-digital-human:clone-poll",avatarId:s,status:$}),$==="done")return w;if($==="failed")throw new i(`❌ avatar clone failed: ${w.cloneError||"unknown error"}`);return null},{intervalMs:1e4,timeoutMs:20*6e4})}catch(w){throw w instanceof i&&/polling timed out/.test(w.message)?new i(`❌ avatar ${s} is still cloning; rerun --clone-status --avatar-id ${s} later`):w}if(d){process.stdout.write(JSON.stringify({avatarId:s,cloneStatus:m.cloneStatus})+`
 `);return}process.stdout.write(`
 ✅ Avatar cloned, id: ${s}
 `),process.stdout.write(`💡 Generate with --avatar-id ${s} --text "..."
-`)}w(T,"cloneAvatar");async function L(t,e,o){const n=await h(t,"/digital-human/avatar/clone-status",{id:e},{timeoutMs:3e4}),a=n.cloneStatus||"unknown";if(o){process.stdout.write(JSON.stringify({avatarId:e,cloneStatus:a,error:n.cloneError||null})+`
+`)}u(T,"cloneAvatar");async function j(t,e,r){const a=await p(t,"/digital-human/avatar/clone-status",{id:e},{timeoutMs:3e4}),n=a.cloneStatus||"unknown";if(r){process.stdout.write(JSON.stringify({avatarId:e,cloneStatus:n,error:a.cloneError||null})+`
 `);return}process.stdout.write(`
-clone status: ${a}
-`),a==="failed"&&process.stdout.write(`   failure reason: ${n.cloneError||"unknown error"}
-`)}w(L,"checkCloneStatus");async function j(t,e,o){const n=await h(t,"/digital-human/video/status",{generationId:e},{timeoutMs:3e4}),a=n.status??"unknown";if(o){process.stdout.write(JSON.stringify({status:a,url:n.videoUrl??null,generationId:e})+`
+clone status: ${n}
+`),n==="failed"&&process.stdout.write(`   failure reason: ${a.cloneError||"unknown error"}
+`)}u(j,"checkCloneStatus");async function C(t,e,r){const a=await p(t,"/digital-human/video/status",{generationId:e},{timeoutMs:3e4}),n=a.status??"unknown";if(r){process.stdout.write(JSON.stringify({status:n,url:a.videoUrl??null,generationId:e})+`
 `);return}if(process.stdout.write(`
-task status: ${a}
-`),n.videoUrl&&process.stdout.write(`   video URL: ${n.videoUrl}
-`),a==="failed"){const r=n.errorMessage??n.error??n.message??"unknown error";process.stdout.write(`   failure reason: ${r}
-`)}}w(j,"checkStatus");const C=30*6e4;async function D(t,e){return await y(async()=>{const n=await h(t,"/digital-human/video/status",{generationId:e},{timeoutMs:3e4}),a=n.status??"unknown";if(v({phase:"gen-digital-human:poll",generationId:e,status:a}),a==="completed")return n.videoUrl??"";if(a==="failed"){const r=n.errorMessage??n.error??n.message??"unknown error";throw new i(`❌ video generation failed: ${r}`)}return null},{intervalMs:5e3,timeoutMs:C}).catch(n=>{throw n instanceof i&&/polling timed out/.test(n.message)?new i(`${n.message}; still generating on the server — resume with --check-status --generation-id ${e}`):n})}w(D,"pollVideoStatus");async function q(t,e){const o=await b(e.skillName,{apiBaseUrl:t.api_base_url,privateToken:t.priv_token,preflightToken:e.auth?.token}),n=g(t.json_output);if(g(t.list_avatars))return M(o,t);if(g(t.clone))return T(o,t);if(g(t.clone_status)){const $=c(t.avatar_id);if(!$)throw new i("❌ clone-status mode requires --avatar-id");return L(o,$,n)}if(g(t.check_status)){const $=I(t.generation_id,"generation_id");if($===void 0)throw new i("❌ check-status mode requires --generation-id");return j(o,$,n)}const a=c(t.avatar_id);if(!a)throw new i("❌ provide --avatar-id to pick the avatar (use --list-avatars to see available avatars)");const r=(t.audio_url??"").trim(),u=t.text??"";if(r&&!/^https?:\/\//i.test(r))throw new i(`❌ --audio-url needs a public http(s) URL (local files are not uploaded), received: ${r}`);if(!r&&!u.trim())throw new i("❌ in TTS mode provide --text, or use --audio-url for audio-driven mode");const l=(await N(t.api_base_url)).digitalHuman;v({phase:"gen-digital-human:resolve-avatar",avatarId:a});const s=await O(o,a);if(s.cloneStatus==="cloning")throw new i(`❌ avatar ${a} is still cloning; check with --clone-status --avatar-id ${a}`);if(s.cloneStatus==="failed")throw new i(`❌ avatar ${a} failed to clone (${s.cloneError||"unknown error"}); clone it again`);const d=R(s,l.sources),m={avatarId:a,source:d};r?m.audioUrl=r:(m.text=u,m.voiceId=c(t.voice_id)||c(s.defaultVoiceId)||l.defaultVoiceId,t.voice_name&&(m.voiceName=String(t.voice_name))),v({phase:"gen-digital-human:request",avatarId:a,source:d,mode:r?"audio":"tts"});const f=await h(o,"/digital-human/video/generate",m,{timeoutMs:6e4}),p=f.generationId;if(!p)throw new i(`❌ could not parse generationId; API returned: ${JSON.stringify(f)}`);v({phase:"gen-digital-human:async",generationId:p});const _=await D(o,p);if(!_)throw new i(`❌ video generated but no URL was returned; rerun --check-status --generation-id ${p} later`);if(n){process.stdout.write(JSON.stringify({url:_,generationId:p})+`
+task status: ${n}
+`),a.videoUrl&&process.stdout.write(`   video URL: ${a.videoUrl}
+`),n==="failed"){const o=a.errorMessage??a.error??a.message??"unknown error";process.stdout.write(`   failure reason: ${o}
+`)}}u(C,"checkStatus");const D=30*6e4;async function J(t,e){return await S(async()=>{const a=await p(t,"/digital-human/video/status",{generationId:e},{timeoutMs:3e4}),n=a.status??"unknown";if(v({phase:"gen-digital-human:poll",generationId:e,status:n}),n==="completed")return a.videoUrl??"";if(n==="failed"){const o=a.errorMessage??a.error??a.message??"unknown error";throw new i(`❌ video generation failed: ${o}`)}return null},{intervalMs:5e3,timeoutMs:D}).catch(a=>{throw a instanceof i&&/polling timed out/.test(a.message)?new i(`${a.message}; still generating on the server — resume with --check-status --generation-id ${e}`):a})}u(J,"pollVideoStatus");function P(t){const e=String(t.video_url??"").trim(),r=String(t.audio_url??"").trim(),a=String(t.lipsync_mode??"speed");for(const[n,o]of[["--video-url",e],["--audio-url",r]])if(!/^https?:\/\//i.test(o))throw new i(`❌ lip-sync needs a public http(s) URL for ${n}, received: ${o}`);if(a!=="speed"&&a!=="precision")throw new i(`❌ --lipsync-mode must be speed or precision, received: ${a}`);return{videoUrl:e,audioUrl:r,mode:a}}u(P,"buildLipsyncRequest");async function H(t,e){const r=await M(e.skillName,{apiBaseUrl:t.api_base_url,privateToken:t.priv_token,preflightToken:e.auth?.token}),a=g(t.json_output);if(g(t.list_avatars))return O(r,t);if(g(t.clone))return T(r,t);if(g(t.clone_status)){const h=c(t.avatar_id);if(!h)throw new i("❌ clone-status mode requires --avatar-id");return j(r,h,a)}if(g(t.check_status)){const h=I(t.generation_id,"generation_id");if(h===void 0)throw new i("❌ check-status mode requires --generation-id");return C(r,h,a)}if(t.video_url){const h=P(t);v({phase:"gen-digital-human:request",mode:"lipsync",lipsyncMode:h.mode});const _=await p(r,"/digital-human/lipsync/generate",h,{timeoutMs:6e4});if(!_.generationId)throw new i(`❌ could not parse generationId; API returned: ${JSON.stringify(_)}`);return U(r,_.generationId,a)}const n=c(t.avatar_id);if(!n)throw new i("❌ provide --avatar-id to pick the avatar (use --list-avatars to see available avatars)");const o=(t.audio_url??"").trim(),f=t.text??"";if(o&&!/^https?:\/\//i.test(o))throw new i(`❌ --audio-url needs a public http(s) URL (local files are not uploaded), received: ${o}`);if(!o&&!f.trim())throw new i("❌ in TTS mode provide --text, or use --audio-url for audio-driven mode");const l=(await N(t.api_base_url)).digitalHuman;v({phase:"gen-digital-human:resolve-avatar",avatarId:n});const s=await R(r,n);if(s.cloneStatus==="cloning")throw new i(`❌ avatar ${n} is still cloning; check with --clone-status --avatar-id ${n}`);if(s.cloneStatus==="failed")throw new i(`❌ avatar ${n} failed to clone (${s.cloneError||"unknown error"}); clone it again`);const d=L(s,l.sources),m={avatarId:n,source:d};o?m.audioUrl=o:(m.text=f,m.voiceId=c(t.voice_id)||c(s.defaultVoiceId)||l.defaultVoiceId,t.voice_name&&(m.voiceName=String(t.voice_name))),v({phase:"gen-digital-human:request",avatarId:n,source:d,mode:o?"audio":"tts"});const w=await p(r,"/digital-human/video/generate",m,{timeoutMs:6e4}),$=w.generationId;if(!$)throw new i(`❌ could not parse generationId; API returned: ${JSON.stringify(w)}`);return U(r,$,a)}u(H,"genDigitalHuman");async function U(t,e,r){v({phase:"gen-digital-human:async",generationId:e});const a=await J(t,e);if(!a)throw new i(`❌ video generated but no URL was returned; rerun --check-status --generation-id ${e} later`);if(r){process.stdout.write(JSON.stringify({url:a,generationId:e})+`
 `);return}process.stdout.write(`
 🔗 Video URL (renders directly, no download needed):
-${_}
+${a}
 `),process.stdout.write(`
 🎉 Digital-human video generation done!
-`)}w(q,"genDigitalHuman");export{q as genDigitalHuman};
+`)}u(U,"finishGeneration");export{P as buildLipsyncRequest,H as genDigitalHuman};

package/dist/manifest.json

@@ -1,7 +1,7 @@
 {
   "schemaVersion": 1,
-  "version": "0.13.9",
-  "generatedAt": "2026-11-03T15:23:12.126Z",
+  "version": "0.13.10",
+  "generatedAt": "2026-11-05T14:34:43.632Z",
   "skills": [
     {
       "id": "doc-parse",
@@ -269,6 +269,18 @@
             "type": "string",
             "description": "Narration text (TTS mode, ≤ 10000 chars). Synthesized with voice_id, then drives the avatar; the speech must last at least 5 s. Ignored when audio_url is given."
           },
+          "video_url": {
+            "type": "string",
+            "description": "Lip-sync mode: public http(s) URL of the video to re-voice. Needs audio_url. Replaces avatar_id."
+          },
+          "lipsync_mode": {
+            "type": "string",
+            "enum": [
+              "speed",
+              "precision"
+            ],
+            "description": "Lip-sync quality, default speed. precision costs about 4x as much."
+          },
           "audio_url": {
             "type": "string",
             "description": "Audio-driven mode: public http(s) URL of mp3 / m4a / wav, 5 s – 30 min, ≤ 100 MB. Takes precedence over text / voice_id."
@@ -301,7 +313,9 @@
           "avatar_id",
           "text",
           "voice_id",
-          "audio_url"
+          "audio_url",
+          "video_url",
+          "lipsync_mode"
         ],
         "advanced": [],
         "hidden": [
@@ -625,6 +639,43 @@
             "type": "string",
             "description": "Media fill mode for templates declaring capabilities.mediaBinding.options.fit; one of that option's enum in the template's customPayloadSchema (its descriptions explain each value). Beats the preset's own fit. **Must** pass when the user picked it: `显示方式: <名称>（<id>）` → pass `<id>`; `auto` keeps the template / preset default."
           },
+          "source": {
+            "type": "string",
+            "description": "Presenter source, for templates that declare capabilities.sources: digital-human, lip-sync or real-footage. Omit to use the template's default."
+          },
+          "template_style": {
+            "type": "string",
+            "description": "Visual style for templates that ship several, e.g. talking-head-studio: douyin-bold, news-anchor, live-ecommerce."
+          },
+          "presenter_video": {
+            "type": "string",
+            "description": "URL of the user's talking-head video. Required with source real-footage or lip-sync."
+          },
+          "lipsync_mode": {
+            "type": "string",
+            "enum": [
+              "speed",
+              "precision"
+            ],
+            "description": "Lip-sync quality for source lip-sync, default speed. precision costs about 4x as much."
+          },
+          "broll_asset": {
+            "type": "array",
+            "items": {
+              "type": "string"
+            },
+            "description": "URLs of images or videos the user picked as b-roll from their asset library (the replicate card's 空镜素材 asset lines), on any source. Not scene_images: scene_images treats every entry as an image."
+          },
+          "broll_mode": {
+            "type": "string",
+            "enum": [
+              "auto",
+              "ai-image",
+              "ai-video",
+              "none"
+            ],
+            "description": "Scenes without a broll_asset (the replicate card's 其余画面): auto (the visual style's default), ai-image, ai-video (at most 3 scenes, the rest stay images) or none."
+          },
           "narration": {
             "type": "string",
             "enum": [
@@ -770,6 +821,10 @@
           "background_preset",
           "sequence_preset",
           "sequence_fit",
+          "source",
+          "template_style",
+          "presenter_video",
+          "lipsync_mode",
           "narration",
           "background_asset",
           "payload",
@@ -787,6 +842,8 @@
           "subheadline",
           "carousel_items",
           "scene_images",
+          "broll_asset",
+          "broll_mode",
           "caption_lines",
           "stub_image_url",
           "stub_video_url",
@@ -1171,6 +1228,10 @@
             "minimum": 1,
             "description": "Render job id (positive integer, e.g. 8). This is the integer N parsed from prepare_video_assets's stdout line '📦 render job jobId: N' (also tolerated: legacy zh '📦 渲染任务 jobId: N'). Never pass 0, a placeholder string, or descriptive text."
           },
+          "restyle": {
+            "type": "string",
+            "description": "Render the same video again in another visual style (a template_style id). Reuses the script, assets and timing, so nothing is regenerated. Only for templates with visual styles such as talking-head-studio."
+          },
           "save_job": {
             "type": "boolean",
             "description": "Persist the render Manifest back to the database under the same jobId (default true). Auto-degrades to false when PRIV_TOKEN is missing. Pass false to opt out explicitly."
@@ -1205,6 +1266,7 @@
           "job_id"
         ],
         "advanced": [
+          "restyle",
           "save_job",
           "upload_title",
           "no_upload",

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@remixmate/cli",
-  "version": "0.13.9",
+  "version": "0.13.10",
   "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-digital-human/SKILL.md

@@ -116,6 +116,8 @@ remixmate gen-digital-human --clone-status --avatar-id dh_xxx
 | `--voice-id` | generate | Voice id (text-driven) | the avatar's own voice (curated suggestion / HeyGen twin's cloned voice), else the service default |
 | `--voice-name` | generate | Voice display name (recordkeeping only) | none |
 | `--audio-url` | generate | Audio URL (audio-driven) | none |
+| `--video-url` | generate | Lip-sync an existing video instead of an avatar; needs `--audio-url` | none |
+| `--lipsync-mode` | generate | `speed` or `precision` (about 4x the price) | `speed` |
 | `--check-status` / `--generation-id` | status | Poll a generation job | — |
 | `--clone` / `--clone-url` | clone | Clone mode and source URL (required) | — |
 | `--clone-type` | clone | `video` / `image` | `video` |

package/skills/gen-digital-human/skill.json

@@ -95,6 +95,15 @@
         "type": "string",
         "description": "Narration text (TTS mode, ≤ 10000 chars). Synthesized with voice_id, then drives the avatar; the speech must last at least 5 s. Ignored when audio_url is given."
       },
+      "video_url": {
+        "type": "string",
+        "description": "Lip-sync mode: public http(s) URL of the video to re-voice. Needs audio_url. Replaces avatar_id."
+      },
+      "lipsync_mode": {
+        "type": "string",
+        "enum": ["speed", "precision"],
+        "description": "Lip-sync quality, default speed. precision costs about 4x as much."
+      },
       "audio_url": {
         "type": "string",
         "description": "Audio-driven mode: public http(s) URL of mp3 / m4a / wav, 5 s – 30 min, ≤ 100 MB. Takes precedence over text / voice_id."
@@ -127,7 +136,9 @@
       "avatar_id",
       "text",
       "voice_id",
-      "audio_url"
+      "audio_url",
+      "video_url",
+      "lipsync_mode"
     ],
     "advanced": [],
     "hidden": [

package/skills/gen-script/SKILL.md

@@ -487,6 +487,12 @@ python3 <SkillDir>/scripts/gen_script.py \
 | `--sequence-preset` | Style preset for templates declaring `capabilities.mediaBinding.options.preset`; one of that option's enum in the template's `customPayloadSchema` (`none` keeps the template defaults). Written next to the media list; the preset sets transitions, motion, text layout and pacing, and the scene duration follows. **Must** pass when the user picked a style preset on the replicate card (pass the id in parentheses, e.g. `风格预设: <名称>（<id>）` → `<id>`); otherwise the choice is silently dropped. Rejected on templates without that option. |
 | `--sequence-fit` | Fill mode for templates declaring `capabilities.mediaBinding.options.fit`; one of that option's enum (`auto` keeps the template / preset default). Beats the preset's own fit. **Must** pass when the user picked it on the replicate card (`显示方式: <名称>（<id>）` → `<id>`). |
 | `--narration` | `auto` (the template's `capabilities.needsNarration` decides), `on` or `off`. Templates where narration is optional default to none; **must** pass `on` when the user picked narration on the replicate card (`旁白: <名称>（narrated）`), otherwise the narration written later is dropped without an error. `on` adds the narration slot, a `gen-voice` asset and subtitles; on `fit-images` templates that declare `customPayload.minDurationSec` it also writes the media's natural length there, so a short narration does not compress the media (a longer one stretches the scene). |
+| `--template-style` | Visual style for templates that ship several (`talking-head-studio`: `douyin-bold`, `news-anchor`, `live-ecommerce`, `minimal-talk`, `interview-serif`, `gold-quote`, `lecture`, `soft-glow`, `cinema-talk`). |
+| `--presenter-video` | URL of the user's talking-head video, with `--source real-footage` (used as is, no cutting) or `--source lip-sync` (re-voiced with the script). |
+| `--broll-asset` | Repeatable. URL of an image or video the user picked as b-roll from their asset library (replicate card field `空镜素材`). Works on every `--source`. Use the description on the asset line to anchor it to the sentence it matches with `atText`; if one fits nowhere, remove it and tell the user which. |
+| `--broll-mode` | Scenes with no `--broll-asset`: `auto` (the visual style's default), `ai-image`, `ai-video` (at most 3 scenes, the rest stay images) or `none`. Pass the replicate card field `其余画面`. |
+| `--lipsync-mode` | `speed` (default) or `precision` (about 4x the price), with `--source lip-sync`. |
+| `--source` | Presenter source for templates that declare `capabilities.sources`: `digital-human`, `lip-sync` or `real-footage`. Omit to use the template's default. Writes `global.presenter.source` and `global.configLock`; capabilities, `llmHint` and the payload schema then follow that source. |
 | `--background-asset` | URL of the user's background image or video (`.mp4/.mov/.webm/.m4v` → video). Added as existing asset `asset-background`; every scene's `customPayload.background` gets `preset: image` + `assetRef`, or `preset: video` + `videoAssetRef`. Conflicts with a different `--background-preset`; rejected when the template's background enum lacks that kind. | none |
 | `--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 |