> ## Documentation Index
> Fetch the complete documentation index at: https://docs.talkcut.com/llms.txt
> Use this file to discover all available pages before exploring further.

# CLAUDE

# Documentation project instructions

## About this project

* This is the TalkCut product documentation site built on [Mintlify](https://mintlify.com)
* Pages are MDX files with YAML frontmatter
* Configuration lives in `docs.json`
* Run `npx mintlify dev` to preview locally
* Run `npx mintlify broken-links` to check links

## Quick start

```bash theme={null}
npm install          # Install Mintlify CLI
npx mintlify dev     # Preview at localhost:3000
npx mintlify broken-links  # Verify all links
```

## Structure

```
.
├── docs.json              # Navigation, theme, and site config
├── introduction.mdx       # English pages (default language)
├── getting-started.mdx
├── core-concepts/         # Canvas, AI agent, vision engine, workspace memory
├── creating/              # Image/video generation, analysis
├── workspace/             # Projects, assets, skills, team, settings, pricing
├── zh/                    # Chinese translations (mirror English structure)
└── ja/                    # Japanese translations (mirror English structure)
```

All content exists in 3 languages (en, zh, ja). When adding or editing a page, update all 3 versions and add the page to all 3 language sections in `docs.json`.

## Adding a new page

1. Create the English `.mdx` file with YAML frontmatter (`title`, `description`, `icon`)
2. Add the page path to the appropriate group in `docs.json` under the `en` language
3. Create the `zh/` and `ja/` translated versions
4. Add those paths to the `zh` and `ja` language sections in `docs.json`
5. Run `npx mintlify broken-links` to verify all cross-references

## Terminology

| Term             | Usage                                                                                                    |
| ---------------- | -------------------------------------------------------------------------------------------------------- |
| Canvas           | The infinite spatial workspace. Always lowercase except at sentence start.                               |
| AI agent         | The chat-based creative partner. Not "bot", "assistant", or "chatbot".                                   |
| Vision engine    | The image/video analysis system. Not "analyzer" or "scanner".                                            |
| Workspace memory | The persistent context system. Not "AI memory" or "brain".                                               |
| Credits          | The unit of currency for AI generation. Not "tokens" or "points".                                        |
| Workspace        | A shared team environment. Not "organization" or "account".                                              |
| Project          | A single creative work containing a canvas and chat history.                                             |
| Generation       | The act of AI creating an image, video, or audio. Not "creation" or "output" as a noun for the process.  |
| Reference video  | Video generated with reference images for visual consistency. Not "image-to-video" or "multi-shot".      |
| Asset library    | The workspace-level media hub for uploaded files. Not "media library" or "file manager".                 |
| Auto Classify    | AI-powered asset tagging through visual similarity clustering. Not "auto-tag" or "smart tags".           |
| Skill            | A reusable prompt template invoked via /slash command. Not "macro", "shortcut", or "preset".             |
| Tag              | A label for organizing assets. Not "folder", "category", or "group" (tag groups are a separate concept). |

## Translation conventions

| English          | Chinese (zh) | Japanese (ja) |
| ---------------- | ------------ | ------------- |
| AI agent         | AI 助手        | AIエージェント      |
| Canvas           | 画布           | キャンバス         |
| Vision engine    | 视觉引擎         | ビジョンエンジン      |
| Workspace memory | 工作区记忆        | ワークスペースメモリー   |
| Credits          | 积分           | クレジット         |
| Asset library    | 素材库          | アセットライブラリ     |
| Skill            | 技能           | スキル           |
| Tag              | 标签           | タグ            |
| Reference video  | 参考视频         | リファレンス動画      |
| Auto Classify    | 自动分类         | 自動分類          |
| Auto Name        | 自动命名         | 自動ネーミング       |

Internal links in translations use language prefix: `/zh/creating/...`, `/ja/creating/...`

## Style preferences

* Use active voice and second person ("you")
* Keep sentences concise — one idea per sentence
* Use sentence case for headings
* Bold for UI elements: Click **Settings**
* Code formatting for file names, commands, paths, and code references
* No marketing language: avoid "powerful", "seamless", "robust", "cutting-edge"
* No filler phrases: avoid "it's important to note", "in order to"
* No emojis in content
* Present credit costs as ranges per operation type, not per model
* Use user-friendly descriptions for AI capabilities, not internal model names

## Content boundaries

* **Document only**: End-user features visible in the product UI
* **Do not document**: Internal architecture, API endpoints, database schemas, model IDs, code patterns
* **Do not name**: Specific AI model providers or model identifiers (e.g., don't mention "kling", "veo", "gemini" — describe capabilities instead)
* **Coming soon**: NLE export (AAF/XML for DaVinci/Premiere/FCP) — mention only on pricing page with "coming soon" label
* **No developer docs**: This site is for end users/creators, not developers or API consumers

## Mintlify components

Use these MDX components where appropriate:

* `<Steps>` / `<Step title="...">` — Sequential workflows
* `<Tabs>` / `<Tab title="...">` — Tabbed content (e.g., annual vs monthly pricing)
* `<Tip>` — Helpful advice
* `<Note>` — Important context
* `<Warning>` — Caution about credit costs or destructive actions
* `<Accordion title="...">` — Collapsible FAQ items
* `<Card title="..." icon="..." href="...">` — Navigation cards
* `<Columns cols={N}>` — Multi-column layout for cards

## Syncing with the product

* Product codebase: `~/repo/github.com/talkcut/talkcut`
* Sync docs by reading product codebase for user-facing features — never expose implementation details
* Check `lib/ai/tools/` for current AI capabilities (tool names, parameters)
* Check `lib/credits/constants.ts` for current credit costs
* Check `components/assets/` for asset management UI behavior