OCR.space MCP Server
MCP server for OCR.space API to extract text from images and PDFs, with tools for file, URL, automatic oversized handling, and language support.
README
OCR.space MCP Server
Python MCP server for OCR.space - extract text from images and PDFs using OCR.
Features
-
Six comprehensive tools:
ocr_file- Extract text from local images/PDFsocr_url- Extract text from images/PDFs at URLsocr_auto- Smart OCR that auto-handles oversized filessplit_pdf- Split large PDFs into API-compatible chunkslist_languages- Show supported OCR languagescheck_tier_status- Verify API configuration
-
Automatic oversized file handling:
- PDFs too large? Automatically split by pages, OCR each chunk, join results
- Images too large? Automatically compress while preserving quality
-
Dual-tier support:
- Free tier: 1 MB limit, US servers
- PRO tier: 5 MB limit, EU endpoint for GDPR compliance
-
Full API parameter support:
- 26+ languages including auto-detection
- Two OCR engines (optimized for different use cases)
- Table optimization mode
- Auto-rotation detection
- Searchable PDF generation
- Save results as TXT or JSON
Installation
# With uv (recommended)
uv sync
# With pip
pip install -e .
Configuration
API Keys
Get your free API key at ocr.space/ocrapi/freekey.
For PRO features (larger files, EU endpoint), sign up at ocr.space/ocrapi.
Environment Variables
Copy .envrc.example to .envrc and configure:
# Required: Free tier API key
export OCR_SPACE_API_KEY="your-free-api-key"
# Optional: PRO tier for GDPR-compliant EU processing
export OCR_SPACE_PRO_API_KEY="your-pro-api-key"
export OCR_SPACE_PRO_ENDPOINT="https://eu.api.ocr.space/parse/image"
Then enable with direnv allow.
Claude Code / MCP Config
Add to .mcp.json:
{
"mcpServers": {
"ocr-space": {
"command": "uv",
"args": ["run", "--project", "/path/to/ocr-space-mcp", "ocr-space-mcp"],
"env": {
"OCR_SPACE_API_KEY": "${OCR_SPACE_API_KEY}",
"OCR_SPACE_PRO_API_KEY": "${OCR_SPACE_PRO_API_KEY}",
"OCR_SPACE_PRO_ENDPOINT": "${OCR_SPACE_PRO_ENDPOINT}"
}
}
}
}
OpenCode Config
Add to opencode.json:
{
"mcp": {
"ocr-space": {
"command": "uv",
"args": ["run", "--project", "/path/to/ocr-space-mcp", "ocr-space-mcp"],
"env": {
"OCR_SPACE_API_KEY": "${OCR_SPACE_API_KEY}",
"OCR_SPACE_PRO_API_KEY": "${OCR_SPACE_PRO_API_KEY}",
"OCR_SPACE_PRO_ENDPOINT": "${OCR_SPACE_PRO_ENDPOINT}"
}
}
}
}
Usage
Check Configuration
Check my OCR.space tier status
OCR a Local File
Extract text from /path/to/document.pdf
OCR this image using the PRO tier: /path/to/scan.png
Parameters:
| Parameter | Description | Default |
|---|---|---|
file_path |
Path to image/PDF (required) | - |
tier |
free or pro |
free |
language |
OCR language code | eng |
ocr_engine |
1 (fast) or 2 (accurate) | 1 |
output_path |
Save result to file | - |
output_format |
txt or json |
txt |
detect_orientation |
Auto-rotate image | false |
scale |
Upscale low-res images | false |
is_table |
Optimize for tables | false |
is_create_searchable_pdf |
Generate searchable PDF | false |
OCR from URL
Extract text from https://example.com/document.png
Smart OCR for Large Files (ocr_auto)
The ocr_auto tool automatically handles files that exceed the API size limits:
OCR this large document: /path/to/big-scan.pdf
How it works:
- PDFs: Splits into chunks by page, OCRs each chunk, joins results
- Images: Compresses (quality + resize) until under limit, then OCRs
Parameters: Same as ocr_file, but no size limit errors!
Split PDF Only
If you just want to split a large PDF without OCR:
Split /path/to/large.pdf for free tier processing
Creates multiple smaller PDF files in a temp directory (or specify output_dir).
List Languages
What OCR languages are supported?
Supported languages: Arabic, Bulgarian, Chinese (Simplified/Traditional), Croatian, Czech, Danish, Dutch, English, Finnish, French, German, Greek, Hungarian, Italian, Japanese, Korean, Polish, Portuguese, Russian, Slovenian, Spanish, Swedish, Thai, Turkish, Ukrainian, Vietnamese.
Use language=auto with Engine 2 for automatic language detection.
Tier Comparison
| Feature | Free | PRO |
|---|---|---|
| File size limit | 1 MB | 5 MB |
| Server location | US only | EU available |
| Rate limit | 500/day | Unlimited |
| GDPR compliance | No | Yes (EU endpoint) |
| Price | Free | $30/month |
OCR Engines
| Engine | Best For |
|---|---|
| Engine 1 | Faster processing, Asian languages, large images |
| Engine 2 | Auto-language detection, special characters, rotated text |
Development
# Install with dev dependencies
uv sync --extra dev
# Run linter
uv run ruff check .
# Format code
uv run ruff format .
# Run tests
uv run pytest -v
# Run tests with coverage
uv run pytest --cov=ocr_space_mcp --cov-report=term-missing
License
MIT - see LICENSE
Links
推荐服务器
Baidu Map
百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。
VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
Exa MCP Server
模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。