# translate-image-skills

Agent skills for [TranslateImage](https://translateimage.io) — AI-powered image translation, OCR, and text removal.

## Skills

### translate-image

Translate text in images, extract text via OCR, and remove text using AI inpainting. Supports manga, comics, street signs, menus, documents, and any image with foreign-language text.

**4 tools included:**

| Tool | Description |
|---|---|
| **Translate Image** | Translate text in images across 130+ languages while preserving the original visual layout |
| **Extract Text (OCR)** | Extract all text from images with bounding boxes, language detection, and confidence scores |
| **Remove Text** | Detect and remove text overlays using AI inpainting |
| **Image to Text** | Gemini-powered text extraction with optional simultaneous multi-language translation |

## Installation

### Claude Code

```bash
npx skills add translateimage/translate-image-skills
```

Or add the skill manually — copy [`skills/translate-image/SKILL.md`](skills/translate-image/SKILL.md) into your project's `.claude/skills/` directory.

### Other Agents (Codex, Cursor, Gemini CLI)

Copy `skills/translate-image/SKILL.md` into your agent's skills directory. The skill follows the open [Agent Skills](https://skills.sh) specification (SKILL.md with YAML frontmatter).

## Setup

1. Get an API key at [translateimage.io/dashboard](https://translateimage.io/dashboard)
2. Set the environment variable:

```bash
export TRANSLATEIMAGE_API_KEY=your-api-key
```

## Requirements

- `curl`
- `python3`
- `TRANSLATEIMAGE_API_KEY` environment variable

## Supported Languages

130+ languages including English, Japanese, Chinese, Korean, Spanish, French, German, Arabic, Hindi, and more.

## Fonts

Choose a font that matches your content:

| Font | Best for |
|---|---|
| `NotoSans` | General use (default) |
| `WildWords` | Manga, comics |
| `BadComic` | Manga, comics |
| `MaShanZheng` | Chinese content |
| `RIDIBatang` | Korean content |
| `Bangers` | Bold, impact text |
| `Edo` | Japanese-style |
| `KomikaJam` | Comics |

## Translation Models

| Model | Notes |
|---|---|
| `gemini-2.5-flash` | Default — fast and high quality |
| `gpt-5.1` | OpenAI |
| `grok-4-fast` | xAI |
| `deepseek` | DeepSeek |
| `kimi-k2` | Moonshot |

## Project Structure

```
├── AGENTS.md              → symlink to skills/translate-image/SKILL.md
├── CLAUDE.md              → symlink to AGENTS.md
├── LICENSE                → MIT
├── .claude-plugin/
│   └── marketplace.json   → Claude Code plugin manifest
└── skills/
    └── translate-image/
        ├── SKILL.md       → Skill definition (Agent Skills spec)
        └── metadata.json  → Skill metadata
```

## License

[MIT](LICENSE) — TranslateImage