@starreel/mcp 0.1.116 → 0.1.118

package/SKILL.md

@@ -80,8 +80,8 @@ announces a condensed version as MCP `instructions` at connect time.
 | A novel / outline / synopsis (not yet a screenplay) | `set_script` → `rewrite_script` (creative rewrite builds hooks and beats) → `review_script` | — |
 | A finished screenplay (scene headers + dialogue lines) | `set_script` → `rewrite_script` (auto two-pass fidelity: dialogue byte-locked) → `review_script`; force with `rewrite_pipeline: "two_pass"`, `fidelity_enforce: 1` | pasting it into `edit_rewritten_script` (400) |
 | Wants to rewrite on another AI / says our rewrite "changed too much" | `get_script_format_spec` → external rewrite → `check_script_format` → `adopt_external_script` (exit A) or `set_script` + `rewrite_script` (exit B) | skipping the check; `edit_rewritten_script` |
-| A finished **shot list** (per-shot seconds / shot size / camera move) | `get_storyboard_table_spec` (the contract: shot-size / camera-move vocabulary, body layout, text-card syntax, a prompt for the external AI) → `check_storyboard_table` (same parser as the import; clear `errors`, read every `warning` — a missing duration, an unrecognised shot size or an empty body all get imported as-is) → `import_storyboard_table` (defaults to `auto_complete`: one background batch fills the professional fields and expands every shot's platform-built base description into full image / video prompts — metered text, tell the customer first; `auto_complete: false` imports only) → `get_autofill_status` until `done` → `review_storyboards` | `rewrite_script` + `generate_storyboards` — strips every production parameter (measured: 8 shots / 36 s became 20 shots / 109 s); importing without the check; reviewing before the completion batch finishes (the token expires when shots change) |
-| **Structured data** — their own tool / spreadsheet export, or an external AI producing JSON (cast + scenes + shots in one go) | `get_bulk_import_spec` (contract + template + worked example + enums, same source as the validator) → `check_bulk_import` (same zod schema; unresolved character / scene references, dead shots and stage directions inside dialogue are surfaced) → `bulk_import_storyboards` (`mode: "replace"` only with the customer's explicit OK; defaults to `auto_complete`: professional fields for every shot, full image / video prompts only for shots that had no `image_prompt` — prompts you supply yourself are kept verbatim — metered text, tell the customer first) → `get_autofill_status` until `done` → `review_storyboards` | converting the JSON to text for `import_storyboard_table`; hand-building shots with `update_shot`; reviewing before the completion batch finishes |
+| A finished **shot list** (per-shot seconds / shot size / camera move) | `get_storyboard_table_spec` (the contract: shot-size / camera-move vocabulary, body layout, text-card syntax, a prompt for the external AI) → `check_storyboard_table` (same parser as the import; clear `errors`, read every `warning` — a missing duration, an unrecognised shot size or an empty body all get imported as-is) → `import_storyboard_table` (defaults to `auto_complete`: one background batch fills the professional fields and expands every shot's platform-built base description into full image / video prompts — metered text, tell the customer first; `auto_complete: false` imports only **and generates no composition contract** — read `contract_coverage` in the receipt and follow the remedy its `note` gives (normally `autofill_storyboards`) before generating frames) → `get_autofill_status` until `done` → `review_storyboards` | `rewrite_script` + `generate_storyboards` — strips every production parameter (measured: 8 shots / 36 s became 20 shots / 109 s); importing without the check; reviewing before the completion batch finishes (the token expires when shots change) |
+| **Structured data** — their own tool / spreadsheet export, or an external AI producing JSON (cast + scenes + shots in one go) | `get_bulk_import_spec` (contract + template + worked example + enums, same source as the validator) → `check_bulk_import` (same zod schema; unresolved character / scene references, dead shots and stage directions inside dialogue are surfaced) → `bulk_import_storyboards` (`mode: "replace"` only with the customer's explicit OK; defaults to `auto_complete`: professional fields for every shot, full image / video prompts only for shots that had no `image_prompt` — prompts you supply yourself are kept verbatim — metered text, tell the customer first; `auto_complete: false` imports only **and generates no composition contract** (visible head count, person / body-part / prop subject and shot size are then only inferred from text) — read `contract_coverage` in the receipt and follow the remedy its `note` gives (normally `autofill_storyboards`) before generating frames) → `get_autofill_status` until `done` → `review_storyboards` | converting the JSON to text for `import_storyboard_table`; hand-building shots with `update_shot`; reviewing before the completion batch finishes |
 | Their own portraits / scene / prop / shot images | `upload_image` · `set_character_portrait` · `upload_scene_image` · `upload_prop_sheet` · `upload_shot_frame` — **check the image before uploading** (see *Customer-supplied images* below); after a portrait swap, `generate_character_sheet` for that character (the old sheet is invalidated and **not** regenerated automatically) | rendering a "fix" elsewhere and uploading it — use `generate_shot_frame`; uploading a group photo, a captioned/watermarked image, or a multi-view sheet as a portrait; uploading a scene photo with people in it |
 | Their own **footage** for a shot (screen recording, product b-roll, an existing clip) | `upload_shot_footage` → that shot is no longer AI-generated (frames/video skipped, final cut uses the clip as-is, duration written back from the clip); **it overwrites whatever AI video that shot already had**, and afterwards every AI video call on it (including `edit_video_shot`) is refused with **409** until you `clear_shot_footage` or pass `replace_user_footage: true` | pasting an externally AI-generated clip to "fix" a shot — it carries none of this film's identity / style anchors; use `regenerate_shot_video` instead |
 | A clip that should **drive the motion** of an AI shot (a blocking/previz pass, a dance or action reference) | `edit_video_shot` with `reference_video_urls` — the shot keeps its AI video and borrows the clip's movement. To make the clip itself the source and restyle it in place, `upload_shot_footage` it first, then `edit_video_shot` with `replace_user_footage: true` | `upload_shot_footage` alone — that registers the clip as the finished shot, so nothing gets restyled and every later AI call on it returns 409 |
@@ -264,7 +264,15 @@ content that will be rejected.
    "transcript"`); no voice or no usable transcript is refused. Call it with
    `preview: true` first: you get `preview_url` and the anchor text without
    anything being written, then call again without `preview` (same
-   `start_s` / `dur_s`) to lock it.
+   `start_s` / `dur_s`) to lock it. Where the platform has sample hygiene
+   enabled the response carries `hygiene`: pauses holding only ambient sound
+   are cut out and the noise floor is measured before / after. `preview_url`
+   is then the cleaned sample and `raw_preview_url` the original — have the
+   customer compare both. `verdict: "fail"` means ambience or music sits under
+   the voice itself (cutting pauses cannot fix it; cloning and video vendors
+   would learn it as timbre), so the lock is refused: pick a cleaner shot or
+   window, and pass `allow_issues: true` only after the customer has listened
+   and insists.
 
    **No sample? Design one.** `design_voice` turns a one-sentence description
    (age, gender, timbre, pace, mood — say what you want, not what you don't)

package/dist/tools/guide-data.js

@@ -57,7 +57,7 @@ export const ENTRY_POINTS = [
             'import_storyboard_table(不传 content 则读本集原始内容;本集已有分镜要 confirm_replace:true)',
         ],
         avoid: '绝不走 set_script→rewrite_script→generate_storyboards:改写会把秒数/景别/运镜/STYLE/字卡当非剧情内容剥掉(生产实测 8 镜 36 秒被拆成 20 镜 109 秒)。',
-        note: '导入默认带 auto_complete(后台 AI 填专业字段 + 把每镜基础描述扩写成出图/视频提示词,文本步后付、调用前告知客户;auto_complete:false 只导入)——回执 started 后用 `get_autofill_status` 轮询到 done 再 `review_storyboards`;`[字卡 9s] 行一 | 行二` 会建成卡镜(成片层直接渲,不出图不出视频)。',
+        note: '导入默认带 auto_complete(后台 AI 填专业字段 + 把每镜基础描述扩写成出图/视频提示词,文本步后付、调用前告知客户;auto_complete:false 只导入,但**不会生成构图契约**,回执 contract_coverage 会报缺几镜和可行补法,按它的 note 处理)——回执 started 后用 `get_autofill_status` 轮询到 done 再 `review_storyboards`;`[字卡 9s] 行一 | 行二` 会建成卡镜(成片层直接渲,不出图不出视频)。',
         free: false,
         flow: 'get_storyboard_table_spec → 外部工具按范本整理 → check_storyboard_table(errors 清零) → import_storyboard_table(默认 auto_complete) → get_autofill_status 到 done → review_storyboards',
     },

package/dist/tools/produce.js

@@ -411,7 +411,7 @@ export function registerProduceTools(server, client) {
         '秒数/景别缺失的镜照常导入但会在 issues 里列出——系统不替客户猜,猜错一个秒数就是成片时长错。' +
         '★导入后**默认自动补全**(auto_complete,跟随剧目「自动补齐辅助资产」开关、默认开):同一后台批次给全镜 AI 填空专业字段' +
         '(镜头意图/节拍/潜台词/J-L cut/声线等,只填空)并把每镜由平台拼的基础描述扩写成完整出图/视频提示词——客户导完即可出图。' +
-        '文本步按 token 后付,**调用前告知客户**;只想导入不补传 auto_complete:false。回执 auto_complete.started=true 后' +
+        '文本步按 token 后付,**调用前告知客户**;只想导入不补传 auto_complete:false(**代价:不会生成构图契约**——可见人数、主体是人物/局部/道具、景别都只能靠文字推断,手部插入镜和画外配音镜容易被当完整人物出图和审核;回执 contract_coverage 列出缺几镜**并给出当前可行的补法,以它的 note 为准**——契约自动补全开着时是调 autofill_storyboards)。回执 auto_complete.started=true 后' +
         '用 get_autofill_status 轮询到 done(每镜数十秒)**再** review_storyboards——补全会改镜,先审查的 token 会失效。', {
         episode_id: z.number().int().positive(),
         content: z.string().optional().describe('分镜表全文;不传则用本集「原始内容」'),
@@ -458,7 +458,7 @@ export function registerProduceTools(server, client) {
         '★导入后:语速律会抬高装不下台词的镜(回执 speech_duration_raised / speech_duration_overflow)。' +
         '★导入后**默认自动补全**(auto_complete,跟随剧目「自动补齐辅助资产」开关、默认开):同一后台批次给全镜 AI 填空专业字段(只填空),' +
         '并只给**没填 image_prompt、由平台拼了基础描述**的镜扩写完整出图/视频提示词(客户自己写的绝不覆盖;回执 auto_complete.prompt_shots)。' +
-        '文本步按 token 后付,**调用前告知客户**;只想导入传 auto_complete:false。回执 started=true 后用 get_autofill_status 轮询到 done **再** review_storyboards。', {
+        '文本步按 token 后付,**调用前告知客户**;只想导入传 auto_complete:false(**代价:不会生成构图契约**——可见人数、主体是人物/局部/道具、景别都只能靠文字推断,手部插入镜和画外配音镜容易被当完整人物出图和审核;回执 contract_coverage 列出缺几镜**并给出当前可行的补法,以它的 note 为准**——契约自动补全开着时是调 autofill_storyboards)。回执 started=true 后用 get_autofill_status 轮询到 done **再** review_storyboards。', {
         episode_id: z.number().int().positive(),
         payload: z.record(z.string(), z.unknown()).describe('通过 check_bulk_import 的载荷对象:{ mode, episode_meta?, characters?, scenes?, storyboards }'),
         auto_complete: z.boolean().optional().describe('导入后自动补全(专业字段 + 平台拼基础描述的镜写出图/视频提示词;文本步后付)。不传=跟随剧目开关(默认开);false=只导入'),
@@ -684,7 +684,8 @@ export function registerProduceTools(server, client) {
     server.tool('review_all', '(推荐·免费)三层一次跑完的整体体检:剧本+分镜+镜头图片。开工前摸底、交付前复查用。' +
         '★它**不发 review_token** —— 进收费步前仍需对应层的 review_script / review_storyboards / review_frames 各跑一次' +
         '(闸认的是「针对当前产出物刚审过」,不是「审过一次」)。', { episode_id: z.number().int().positive() }, async ({ episode_id }) => jsonResult(await client.producePost(`/episodes/${episode_id}/review/all`)));
-    server.tool('autofill_storyboards', '(推荐)AI 一键给全集分镜补全空缺字段,默认**只填空缺、不覆盖已有**(overwrite=true 才覆盖)。提升分镜完整性,出图前做。后台异步,文本步后付。', { episode_id: z.number().int().positive(), overwrite: z.boolean().optional().describe('true=覆盖已有字段(默认 false 只填空缺)') }, async ({ episode_id, overwrite }) => jsonResult(await client.producePost(`/episodes/${episode_id}/storyboards/autofill`, overwrite ? { overwrite } : {})));
+    server.tool('autofill_storyboards', '(推荐)AI 一键给全集分镜补全空缺字段,默认**只填空缺、不覆盖已有**(overwrite=true 才覆盖)。提升分镜完整性,出图前做。后台异步,文本步后付。' +
+        '★平台开启契约自动补全时,也会尝试给**没有构图契约**的镜补上契约(是否开启以导入回执 contract_coverage 的 note 为准;可见人数/主体范围/景别,出图与审核都按它判;默认 overwrite:false 时已有契约不动,**overwrite:true 会连已有契约一起重写**)——导入时关了 auto_complete、回执 contract_coverage 的 note 指向本工具的,出图前调一次。本集已有补全在跑时本次**不排队**(防重复点击重复扣费),新导入的镜要等 get_autofill_status 到 done 后再调。', { episode_id: z.number().int().positive(), overwrite: z.boolean().optional().describe('true=覆盖已有字段(默认 false 只填空缺)') }, async ({ episode_id, overwrite }) => jsonResult(await client.producePost(`/episodes/${episode_id}/storyboards/autofill`, overwrite ? { overwrite } : {})));
     server.tool('enhance_shot_prompts', '(可选增强·★非必须·有副作用)AI 批量增强全集 image_prompt。★会**改写已有 prompt**(含手调的);且**务必 generate_frames 之前**做——' +
         '出图后再改 prompt 会让图陈旧、要重生浪费钱。非跑通一部片的必需,客户点名再做。后台异步,文本步后付。', { episode_id: z.number().int().positive() }, async ({ episode_id }) => jsonResult(await client.producePost(`/episodes/${episode_id}/storyboards/enhance-prompts`)));
     server.tool('complete_ending_motifs', '(可选增强·★非必须·条件性)给结尾段补全缺失的视觉母题(嫁接视觉回响到结尾镜)。' +
@@ -1843,12 +1844,17 @@ export function registerProduceTools(server, client) {
         '\n★台词栏为空也能用:以实际音频为准——平台用已标定的人声探测确认所选区间里有人声,并把转写文字当锚文本;' +
         '探测不到人声、或转写不出可用文字就拒(锚选错会污染此后所有镜的嗓音,宁拒不猜)。响应里 text_source 标明文本来自台词还是转写。' +
         '\n★先 preview:true 试听:只截取并返回 preview_url 与锚文本,**不写入角色**;听着对再去掉 preview、用同样的 start_s/dur_s 正式提交。' +
+        '\n★去环境音与验收(平台按剧开启时生效,响应里有 hygiene 字段):自动剔除两句之间的纯环境音停顿,并测样本底噪(noise_floor_before/after,阈值 threshold)。' +
+        'preview 时 preview_url 是**清理后**的样本、raw_preview_url 是原样本——**请让客户前后对比试听**。' +
+        'verdict=fail 表示人声下面垫着明显的环境声或音乐(剔停顿救不了,克隆和视频厂商会把它当音色学走),正式提交会被拒:' +
+        '换一镜念得更干净的、或调 start_s/dur_s 避开背景声;只有客户听过后明确坚持,才带 allow_issues:true 写入。' +
         '\n★有客户授权的外部音源时改用 clone_voice + set_character_voice;两者二选一,后设的覆盖先设的。', {
         character_id: z.number().int().positive(),
         storyboard_id: z.number().int().positive().describe('本剧里已出视频、该角色在里面说话的镜(台词栏为空时以实际音频为准)'),
         start_s: z.number().min(0).optional().describe('可选:从该镜第几秒开始截'),
         dur_s: z.number().positive().optional().describe('可选:截多长(秒)'),
         preview: z.boolean().optional().describe('true=只截取试听、返回 preview_url 与锚文本,不写入角色。建议先试听再正式提交'),
+        allow_issues: z.boolean().optional().describe('声线样本验收未达标(hygiene.verdict=fail)时,客户听过试听后明确坚持使用才传 true;否则不要传'),
     }, async ({ character_id, ...rest }) => jsonResult(await client.producePost(`/characters/${character_id}/voice-anchor-from-shot`, rest)));
     // ── 素材交接包：客户自己拼片的通道 ──────────────────────────────────────
     // 与 compose_episode 的分工：compose_episode = 「平台替你拼」；本组工具 = 「素材给你，

package/openapi.json

@@ -2,8 +2,8 @@
   "openapi": "3.1.0",
   "info": {
     "title": "StarReel Production API",
-    "version": "0.1.116",
-    "description": "Turn a script into a finished, downloadable short-drama episode over REST.\n\nPipeline: script → AI rewrite → cast/scenes/props extraction → storyboards → portraits & sheets + scene plates → keyframes → video shots → TTS → final cut (.mp4).\n\n**Billing is prepaid and agent-safe**: big-ticket stages are quote-then-generate (`quote_*` returns a `quote_id`). For video, quote == actual charge. For images the quote is a *range* — models billed per reference image are priced before those are assembled, so `estimated_points` is the upper bound (size the balance by it and a run never dies half-way) and `typical_points` is the usual cost; fixed-price models return the two equal. Insufficient balance returns 402 — nothing half-runs and the account never goes negative.\n\nAuth: exchange your API key at `POST /v1/agent/token` for a 15-minute bearer token.\n\nGenerated from the @starreel/mcp v0.1.116 tool surface (operationIds match MCP tool names 1:1)."
+    "version": "0.1.118",
+    "description": "Turn a script into a finished, downloadable short-drama episode over REST.\n\nPipeline: script → AI rewrite → cast/scenes/props extraction → storyboards → portraits & sheets + scene plates → keyframes → video shots → TTS → final cut (.mp4).\n\n**Billing is prepaid and agent-safe**: big-ticket stages are quote-then-generate (`quote_*` returns a `quote_id`). For video, quote == actual charge. For images the quote is a *range* — models billed per reference image are priced before those are assembled, so `estimated_points` is the upper bound (size the balance by it and a run never dies half-way) and `typical_points` is the usual cost; fixed-price models return the two equal. Insufficient balance returns 402 — nothing half-runs and the account never goes negative.\n\nAuth: exchange your API key at `POST /v1/agent/token` for a 15-minute bearer token.\n\nGenerated from the @starreel/mcp v0.1.118 tool surface (operationIds match MCP tool names 1:1)."
   },
   "servers": [
     {
@@ -567,7 +567,7 @@
       "post": {
         "operationId": "set_voice_anchor_from_shot",
         "summary": "从本剧某一镜选定角色的声线(声音基准)",
-        "description": "从本剧某一镜选定角色的声线(声音基准)。免费。不需要客户提供外部音源——就用片子里这个角色已经念出来的声音。\n★★【原声剧(use_clip_audio=true,默认)里这就是「锁声线」】选定后,之后生成的视频会把这段声音交给视频厂商当参考音频,同一角色前后镜的嗓音跟它走;不锁的话每镜由厂商各自采样,前后镜声线会漂。\n★选哪一镜:该角色**独自说话、台词较长、没有背景音乐和别人插话**的镜;start_s/dur_s 可只截其中一段(不传则取默认窗口,时长会按各引擎上限自动钳制)。\n★只对**之后**生成的视频生效。锁之前已经出好的镜会出现在 get_pipeline_status 的 native_voice_anchor.stale_shots 里,用 regenerate_shot_video 重出才会统一。所以最省的做法是:每个说话角色先出一镜、选定声线,再批量出其余镜。\n★台词栏为空也能用:以实际音频为准——平台用已标定的人声探测确认所选区间里有人声,并把转写文字当锚文本;探测不到人声、或转写不出可用文字就拒(锚选错会污染此后所有镜的嗓音,宁拒不猜)。响应里 text_source 标明文本来自台词还是转写。\n★先 preview:true 试听:只截取并返回 preview_url 与锚文本,**不写入角色**;听着对再去掉 preview、用同样的 start_s/dur_s 正式提交。\n★有客户授权的外部音源时改用 clone_voice + set_character_voice;两者二选一,后设的覆盖先设的。",
+        "description": "从本剧某一镜选定角色的声线(声音基准)。免费。不需要客户提供外部音源——就用片子里这个角色已经念出来的声音。\n★★【原声剧(use_clip_audio=true,默认)里这就是「锁声线」】选定后,之后生成的视频会把这段声音交给视频厂商当参考音频,同一角色前后镜的嗓音跟它走;不锁的话每镜由厂商各自采样,前后镜声线会漂。\n★选哪一镜:该角色**独自说话、台词较长、没有背景音乐和别人插话**的镜;start_s/dur_s 可只截其中一段(不传则取默认窗口,时长会按各引擎上限自动钳制)。\n★只对**之后**生成的视频生效。锁之前已经出好的镜会出现在 get_pipeline_status 的 native_voice_anchor.stale_shots 里,用 regenerate_shot_video 重出才会统一。所以最省的做法是:每个说话角色先出一镜、选定声线,再批量出其余镜。\n★台词栏为空也能用:以实际音频为准——平台用已标定的人声探测确认所选区间里有人声,并把转写文字当锚文本;探测不到人声、或转写不出可用文字就拒(锚选错会污染此后所有镜的嗓音,宁拒不猜)。响应里 text_source 标明文本来自台词还是转写。\n★先 preview:true 试听:只截取并返回 preview_url 与锚文本,**不写入角色**;听着对再去掉 preview、用同样的 start_s/dur_s 正式提交。\n★去环境音与验收(平台按剧开启时生效,响应里有 hygiene 字段):自动剔除两句之间的纯环境音停顿,并测样本底噪(noise_floor_before/after,阈值 threshold)。preview 时 preview_url 是**清理后**的样本、raw_preview_url 是原样本——**请让客户前后对比试听**。verdict=fail 表示人声下面垫着明显的环境声或音乐(剔停顿救不了,克隆和视频厂商会把它当音色学走),正式提交会被拒:换一镜念得更干净的、或调 start_s/dur_s 避开背景声;只有客户听过后明确坚持,才带 allow_issues:true 写入。\n★有客户授权的外部音源时改用 clone_voice + set_character_voice;两者二选一,后设的覆盖先设的。",
         "tags": [
           "characters"
         ],
@@ -604,6 +604,9 @@
                   },
                   "preview": {
                     "type": "boolean"
+                  },
+                  "allow_issues": {
+                    "type": "boolean"
                   }
                 },
                 "required": [
@@ -7029,7 +7032,7 @@
       "post": {
         "operationId": "autofill_storyboards",
         "summary": "(推荐)AI 一键给全集分镜补全空缺字段,默认**只填空缺、不覆盖已有**(overwrite=true 才覆盖)",
-        "description": "(推荐)AI 一键给全集分镜补全空缺字段,默认**只填空缺、不覆盖已有**(overwrite=true 才覆盖)。提升分镜完整性,出图前做。后台异步,文本步后付。",
+        "description": "(推荐)AI 一键给全集分镜补全空缺字段,默认**只填空缺、不覆盖已有**(overwrite=true 才覆盖)。提升分镜完整性,出图前做。后台异步,文本步后付。★平台开启契约自动补全时,也会尝试给**没有构图契约**的镜补上契约(是否开启以导入回执 contract_coverage 的 note 为准;可见人数/主体范围/景别,出图与审核都按它判;默认 overwrite:false 时已有契约不动,**overwrite:true 会连已有契约一起重写**)——导入时关了 auto_complete、回执 contract_coverage 的 note 指向本工具的,出图前调一次。本集已有补全在跑时本次**不排队**(防重复点击重复扣费),新导入的镜要等 get_autofill_status 到 done 后再调。",
         "tags": [
           "episodes"
         ],
@@ -7091,7 +7094,7 @@
       "post": {
         "operationId": "bulk_import_storyboards",
         "summary": "把一份结构化 JSON 一次性建成本集的分镜(按 storyboard_number 去重更新)",
-        "description": "把一份结构化 JSON 一次性建成本集的分镜(按 storyboard_number 去重更新)。**不建角色/场景/道具**——它们由 extract_assets 从剧本提取,导入只按名字绑定已有的(回执 characters_unmatched / char_refs_unresolved 列出没绑上的;先 extract 再导)。免费·确定性·不调 AI。★三步走:① get_bulk_import_spec 取契约与成品示例;② check_bulk_import 自检到 errors 清零;③ 再调本工具。★mode=merge(默认)保留未提到的镜;mode=replace 替换本集**全部**分镜(旧镜与已生成图/视频归档可恢复、不会自动重挂)——replace 必须先得到客户明确同意。★image_prompt / video_prompt 想自己写就直接写在每镜里,**逐字照用不被覆盖**;没填的镜平台拼基础描述(回执 base_prompts_built 是拼了几镜),导入后可用 get_shot_prompts 读、update_shot 改。(frame_visual_contract 是内部帧契约,导入时忽略。)★导入后:语速律会抬高装不下台词的镜(回执 speech_duration_raised / speech_duration_overflow)。★导入后**默认自动补全**(auto_complete,跟随剧目「自动补齐辅助资产」开关、默认开):同一后台批次给全镜 AI 填空专业字段(只填空),并只给**没填 image_prompt、由平台拼了基础描述**的镜扩写完整出图/视频提示词(客户自己写的绝不覆盖;回执 auto_complete.prompt_shots)。文本步按 token 后付,**调用前告知客户**;只想导入传 auto_complete:false。回执 started=true 后用 get_autofill_status 轮询到 done **再** review_storyboards。",
+        "description": "把一份结构化 JSON 一次性建成本集的分镜(按 storyboard_number 去重更新)。**不建角色/场景/道具**——它们由 extract_assets 从剧本提取,导入只按名字绑定已有的(回执 characters_unmatched / char_refs_unresolved 列出没绑上的;先 extract 再导)。免费·确定性·不调 AI。★三步走:① get_bulk_import_spec 取契约与成品示例;② check_bulk_import 自检到 errors 清零;③ 再调本工具。★mode=merge(默认)保留未提到的镜;mode=replace 替换本集**全部**分镜(旧镜与已生成图/视频归档可恢复、不会自动重挂)——replace 必须先得到客户明确同意。★image_prompt / video_prompt 想自己写就直接写在每镜里,**逐字照用不被覆盖**;没填的镜平台拼基础描述(回执 base_prompts_built 是拼了几镜),导入后可用 get_shot_prompts 读、update_shot 改。(frame_visual_contract 是内部帧契约,导入时忽略。)★导入后:语速律会抬高装不下台词的镜(回执 speech_duration_raised / speech_duration_overflow)。★导入后**默认自动补全**(auto_complete,跟随剧目「自动补齐辅助资产」开关、默认开):同一后台批次给全镜 AI 填空专业字段(只填空),并只给**没填 image_prompt、由平台拼了基础描述**的镜扩写完整出图/视频提示词(客户自己写的绝不覆盖;回执 auto_complete.prompt_shots)。文本步按 token 后付,**调用前告知客户**;只想导入传 auto_complete:false(**代价:不会生成构图契约**——可见人数、主体是人物/局部/道具、景别都只能靠文字推断,手部插入镜和画外配音镜容易被当完整人物出图和审核;回执 contract_coverage 列出缺几镜**并给出当前可行的补法,以它的 note 为准**——契约自动补全开着时是调 autofill_storyboards)。回执 started=true 后用 get_autofill_status 轮询到 done **再** review_storyboards。",
         "tags": [
           "episodes"
         ],
@@ -7330,7 +7333,7 @@
       "post": {
         "operationId": "import_storyboard_table",
         "summary": "【已有分镜表时用这个】把客户**做完的分镜表**直接建成分镜，跳过改写与拆镜——中间没有任何 agent",
-        "description": "【已有分镜表时用这个】把客户**做完的分镜表**直接建成分镜，跳过改写与拆镜——中间没有任何 agent。免费。★三步走，别直接导：① get_storyboard_table_spec 取范本（景别/运镜白名单、正文落点、字卡语法、外部 AI 提示词）；② check_storyboard_table 自检到 errors 清零、逐条看 warnings（缺秒数/景别没认出/没画面叙述都会原样建进去）；③ 再调本工具。★什么时候用:客户交来的不是剧本而是成品分镜表(逐镜写了秒数/景别/运镜,常见于给电视台/品牌方的样片)。走 set_script→rewrite_script→generate_storyboards 那条路会把这些制作参数当非剧情内容剥掉(生产实测:8 镜 36 秒的表跑完变成 20 镜 109 秒,STYLE 块与数据卡全丢),所以这类客户必须走本工具。★声明格式(每镜一行,字段用 | 或全角 ｜ 分隔):`[镜 001 | 4s | 微距 | 固定]`,也认客户已有的加粗写法 `**镜 001 ｜ 4″ ｜ 微距 ｜ 摄影机固定不动**`(不必让客户重打一遍)。顺序 = 镜号|秒数|景别|运镜|标记;标记 `不切` 表示该镜绝不可再拆、`★` 表示关键镜;运镜的「横移」映射 tracking、「摇」映射 pan(两者在反光面上的倒影行为不同,别混)。`[字卡 9s] 行一 | 行二` 会建成卡镜(成片层直接渲黑底卡,不出图不出视频)。★道具:`[道具] 名A、名B` 行按**本剧道具库里的道具名**绑定 prop_ids,正文里完整出现的道具名也自动绑;回执 prop_names_unresolved 列出没绑上的名字(先建道具再导,或改名)。★不传 content 时读本集「原始内容」。没有任何逐镜声明会被 400 拒(那是剧本不是分镜表,请走正常拆镜)。★本集已有分镜时返回 409 并告知镜数,确认要替换再带 confirm_replace:true(旧镜转已删除状态、可恢复)。秒数/景别缺失的镜照常导入但会在 issues 里列出——系统不替客户猜,猜错一个秒数就是成片时长错。★导入后**默认自动补全**(auto_complete,跟随剧目「自动补齐辅助资产」开关、默认开):同一后台批次给全镜 AI 填空专业字段(镜头意图/节拍/潜台词/J-L cut/声线等,只填空)并把每镜由平台拼的基础描述扩写成完整出图/视频提示词——客户导完即可出图。文本步按 token 后付,**调用前告知客户**;只想导入不补传 auto_complete:false。回执 auto_complete.started=true 后用 get_autofill_status 轮询到 done(每镜数十秒)**再** review_storyboards——补全会改镜,先审查的 token 会失效。",
+        "description": "【已有分镜表时用这个】把客户**做完的分镜表**直接建成分镜，跳过改写与拆镜——中间没有任何 agent。免费。★三步走，别直接导：① get_storyboard_table_spec 取范本（景别/运镜白名单、正文落点、字卡语法、外部 AI 提示词）；② check_storyboard_table 自检到 errors 清零、逐条看 warnings（缺秒数/景别没认出/没画面叙述都会原样建进去）；③ 再调本工具。★什么时候用:客户交来的不是剧本而是成品分镜表(逐镜写了秒数/景别/运镜,常见于给电视台/品牌方的样片)。走 set_script→rewrite_script→generate_storyboards 那条路会把这些制作参数当非剧情内容剥掉(生产实测:8 镜 36 秒的表跑完变成 20 镜 109 秒,STYLE 块与数据卡全丢),所以这类客户必须走本工具。★声明格式(每镜一行,字段用 | 或全角 ｜ 分隔):`[镜 001 | 4s | 微距 | 固定]`,也认客户已有的加粗写法 `**镜 001 ｜ 4″ ｜ 微距 ｜ 摄影机固定不动**`(不必让客户重打一遍)。顺序 = 镜号|秒数|景别|运镜|标记;标记 `不切` 表示该镜绝不可再拆、`★` 表示关键镜;运镜的「横移」映射 tracking、「摇」映射 pan(两者在反光面上的倒影行为不同,别混)。`[字卡 9s] 行一 | 行二` 会建成卡镜(成片层直接渲黑底卡,不出图不出视频)。★道具:`[道具] 名A、名B` 行按**本剧道具库里的道具名**绑定 prop_ids,正文里完整出现的道具名也自动绑;回执 prop_names_unresolved 列出没绑上的名字(先建道具再导,或改名)。★不传 content 时读本集「原始内容」。没有任何逐镜声明会被 400 拒(那是剧本不是分镜表,请走正常拆镜)。★本集已有分镜时返回 409 并告知镜数,确认要替换再带 confirm_replace:true(旧镜转已删除状态、可恢复)。秒数/景别缺失的镜照常导入但会在 issues 里列出——系统不替客户猜,猜错一个秒数就是成片时长错。★导入后**默认自动补全**(auto_complete,跟随剧目「自动补齐辅助资产」开关、默认开):同一后台批次给全镜 AI 填空专业字段(镜头意图/节拍/潜台词/J-L cut/声线等,只填空)并把每镜由平台拼的基础描述扩写成完整出图/视频提示词——客户导完即可出图。文本步按 token 后付,**调用前告知客户**;只想导入不补传 auto_complete:false(**代价:不会生成构图契约**——可见人数、主体是人物/局部/道具、景别都只能靠文字推断,手部插入镜和画外配音镜容易被当完整人物出图和审核;回执 contract_coverage 列出缺几镜**并给出当前可行的补法,以它的 note 为准**——契约自动补全开着时是调 autofill_storyboards)。回执 auto_complete.started=true 后用 get_autofill_status 轮询到 done(每镜数十秒)**再** review_storyboards——补全会改镜,先审查的 token 会失效。",
         "tags": [
           "episodes"
         ],

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@starreel/mcp",
-  "version": "0.1.116",
+  "version": "0.1.118",
   "mcpName": "ai.starreel/starreel",
   "description": "StarReel MCP server — drive the AI short-drama production pipeline (script → storyboards → frames → video → final cut) from Claude, Cursor, or any MCP client",
   "license": "MIT",

package/server.json

@@ -2,13 +2,13 @@
   "$schema": "https://static.modelcontextprotocol.io/schemas/2025-09-29/server.schema.json",
   "name": "ai.starreel/starreel",
   "description": "Turn a script into a finished, downloadable short-drama episode — via MCP or REST.",
-  "version": "0.1.116",
+  "version": "0.1.118",
   "websiteUrl": "https://starreel.ai",
   "packages": [
     {
       "registryType": "npm",
       "identifier": "@starreel/mcp",
-      "version": "0.1.116",
+      "version": "0.1.118",
       "transport": {
         "type": "stdio"
       },