菜单
Use when converting confirmed Dasheng drafts into channel-ready WeChat article, talking-head video, and podcast production packages.
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
Use when Dasheng Publish needs to route Xiaohongshu publishing or content access through API-first CLI, MCP, or browser fallback bridges.
Use when Dasheng Publish needs to prepare, upload, or package Bilibili video submissions from completed transwrite video channel packs.
Use when running Stage 3 Draft from approved Brief topics and producing Reasoning Sheets, standard drafts, and draft quality gates.
Use when the workflow enters Postmortem after Publish and published content needs to be evaluated with performance data, Publish Guard status, accuracy checks, and knowledge-base writeback suggestions.
Use when Dasheng draft or transwrite needs HTML Anything templates, editable article HTML, social cards, posters, or visual HTML references.
Use when Dasheng transwrite needs no-human or human-material talking-head videos rendered through the local html-video project.
| name | dasheng-stage-transwrite |
| description | Use when converting confirmed Dasheng drafts into channel-ready WeChat article, talking-head video, and podcast production packages. |
这是 Draft 之后、Publish 之前的正式主链阶段。
正式阶段顺序：
intake -> brief -> draft -> transwrite -> publish -> postmortem
Draft 负责事实、数据、图表、配图和自包含 HTML。Transwrite 只负责把已确认稿转成不同渠道的表达形态，不重新发明事实链。
本阶段采用轻内核：Python 只生成任务包、提示词、请求体、最终产物槽位、QC 契约和 manifest；真正生产由 Agent 调用对应技能完成，并回写 lane manifest。
draft_manifest.jsonfinal_structure_snapshot.jsontranswrite_decision.json缺少 final_structure_snapshot.json 或 transwrite_decision.json 时禁止执行。
目标：
baoyu-cover-image / baoyu-imagine 生成封面wechat_article.final.md 与 wechat_article.final.htmlwechat_article_qc_report.json，确认图表、表格、图片、事实锚点没有丢失configs/publish/wechat_layout_rules.json：H2 使用阿拉伯数字大标题并左对齐，不用居中块状标题；表格内文字约 12px，单元格紧凑，避免手机端换行挤压；正文不得全篇蓝色或全篇加粗。继承技能：
dasheng-style-profilerwechat-style-profilerbaoyu-markdown-to-htmlbaoyu-cover-imagescripts/wechat_layout_variants.py（生成候选排版预览，并可做最终 HTML 版式后处理）文字洁癖：
目标：
默认渲染器：
dasheng-html-video-bridge/Volumes/PSSD/html-videoframe-liquid-bg-hero，可按题材切换 frame-data-chart-nyt、frame-electric-studio、frame-light-leak-cinema典型模式：
human video/audio -> transcription -> timeline -> HTML visual layer -> html-video/FFmpeg composedraft script -> MiniMax CLI TTS -> animation timeline -> html-video render推荐模块：
dasheng-html-video-bridgedasheng-html-anything-bridgedasheng-video-talking-headdasheng-video-explainer-htmlmotion-framesremotion-best-practices（复杂透明合成兜底）WhisperX / stable-tsmmx（生产配音、配乐、图片生成、口播音频）FFmpeg目标：
podcast_qc_report.json默认 MiniMax CLI：
mmx auth status --no-color
mmx speech synthesize --text-file <podcast_script.txt> --out <podcast.wav> --model speech-2.8-hd --voice "Chinese (Mandarin)_Radio_Host"
未配置 CLI/auth/API key 时，manifest 必须标记 blocked_missing_audio_provider，不得误报“已生成音频”。
python3 scripts/build_stage4_transwrite.py \
--draft-manifest 产物/05_初稿生成/<run_id>/draft_manifest.json \
--transwrite-decision 产物/06_转写生产/<run_id>/transwrite_decision.json
统一入口：
python3 scripts/run_mainline_stage.py transwrite --run-id <run_id>
{
"run_id": "<run_id>",
"gate": "Transwrite Gate",
"status": "approved",
"topics": [
{
"topic_id": "topic-demo",
"lanes": ["wechat_article", "talking_head_video", "podcast"],
"wechat_article": {
"dna_profile": "project_or_user_default",
"humanize": true,
"cover_generation": {"enabled": true}
},
"talking_head_video": {
"base_video": "/path/to/user-talking-head.mp4",
"visual_layer": {
"mode": "html_overlay",
"background": "transparent"
},
"audio": {"mode": "human_audio"},
"alignment": {
"mode": "active_to_existing_audio",
"engine": "whisperx"
},
"render": {
"engine": "html-video",
"template_id": "frame-liquid-bg-hero",
"aspect_ratios": ["9:16", "16:9"]
}
},
"podcast": {
"provider": "minimax",
"mode": "solo"
}
}
]
}
如果没有真人口播素材，把 base_video 移除，并把 audio.mode 改为 synthetic_audio；系统会走被动对齐路径。
04_转写计划.mdtranswrite_manifest.jsonwechat_article/wechat_article_manifest.jsonwechat_article/agent_rewrite_prompt.mdwechat_article/cover_prompt.mdtalking_head_video/talking_head_video_manifest.jsontalking_head_video/video_storyboard.jsontalking_head_video/talking_head_script.mdtalking_head_video/html_overlay.htmltalking_head_video/render_plan.jsontalking_head_video/html_video_project_vars.jsontalking_head_video/html_video_project_plan.jsontalking_head_video/html_video_commands.shpodcast/podcast_manifest.jsonpodcast/podcast_script.mdpodcast/provider_request.json脚本初次生成的 lane 通常只到：
ready_for_agent_execution：等待 Agent 做文字转写、DNA、人味化、封面、QC。ready_for_skill_execution：等待具体技能/API/渲染器执行，例如视频、播客。blocked_missing_human_media / blocked_missing_audio_provider：缺少输入素材或外部服务。只有以下状态可进入 Publish 执行包：
packageablecompleted兼容旧产物时，ready_base_package 可被 publish 当作文字包可用；新产物不要再主动使用这个状态。
transwrite_manifest.json 是进入 Publish 的唯一正式输入。final_artifacts、qc.status 和 lane status，否则 Publish 只能等待。skills/ 目录；默认写入 产物/06_转写生产/<run_id>/...，实验缓存写入 tmp/。references/html-video-workflow.md。references/html-anything-workflow.md。