Initial TrueGrowth source import

This commit is contained in:
2026-07-07 09:36:36 +08:00
commit 3b6781d695
2283 changed files with 691996 additions and 0 deletions

View File

@@ -0,0 +1,16 @@
# Change: Update AI clip center into a commercial workbench
## Why
The current AI clip page exposes a useful local task entry, but it does not yet match the final product plan for a commercial short-video production center. Operators need a clear path from source media to business scenario, AI-selected clips, subtitle packaging, digital-human polish, finished asset storage, and publish-ready outputs.
## What Changes
- Expand the AI clip center from a small preset picker into 12 user-facing clip capabilities.
- Add industry and business scene templates for IP content, live commerce, courses, sales/customer success, and local business.
- Add digital-human talking-head polish packs so avatar outputs can be refined into commercial short videos.
- Keep the flow simple: capability, source, config, editor, preview.
- Submit the selected capability, scene, polish pack, business parameters, model route, local model status, and output options into the existing local runtime task payload.
## Impact
- Affected specs: ai-clip-workbench
- Affected code: `apps/web/src/workbench/WorkbenchShell.tsx`, `apps/web/src/workbench/workbench.scss`
- Runtime integration: existing `/local-api/funclip/tasks` local task endpoint

View File

@@ -0,0 +1,28 @@
## ADDED Requirements
### Requirement: Commercial Clip Capabilities
The system SHALL expose 12 customer-facing AI clip capabilities without showing underlying open-source component names to ordinary users.
#### Scenario: User chooses a clip capability
- **WHEN** the user opens AI 剪辑中心
- **THEN** the page presents AI精彩拆条, 主题与关键词, 问答提取, 文本剪辑, 说话人剪辑, 知识点与章节, 情绪与互动高光, 多段重组, 批量智能拆条, 转写与字幕, 音频剪辑, and 自定义AI剪辑 as selectable capabilities
### Requirement: Business Scene Templates
The system SHALL let users apply business-oriented scene templates before submitting a clip task.
#### Scenario: User applies a live commerce template
- **WHEN** the user selects a business scene such as 电商直播
- **THEN** the task payload includes the scene id, title, prompt, expected output, and relevant business fields
### Requirement: Digital-Human Polish Packs
The system SHALL provide commercial polish packs for digital-human and talking-head outputs.
#### Scenario: User chooses a polish pack
- **WHEN** the user selects 干净口播, 商业带货, 创始人IP, or 课程精修
- **THEN** the task payload includes the selected polish prompt and render packaging expectations
### Requirement: Landing-Ready Task Payload
The system SHALL submit structured runtime payloads that can be executed by local clip, transcription, and rendering services.
#### Scenario: User submits a configured clip job
- **WHEN** the user submits from the config step
- **THEN** the local task contains source media, capability definition, prompt, scene template, polish pack, business params, model route, local model status, and output options

View File

@@ -0,0 +1,8 @@
## 1. Implementation
- [x] 1.1 Expand clip capability definitions to the final 12 user-facing entries.
- [x] 1.2 Add business scene template selection.
- [x] 1.3 Add digital-human polish pack selection.
- [x] 1.4 Include business parameters and publish-ready output flags in the task payload.
- [x] 1.5 Redesign the ability/config surfaces with compact professional desktop UI.
- [x] 1.6 Run TypeScript verification.
- [x] 1.7 Run browser visual verification when the dev server is available.