Extract text from images and PDFs, including handwriting, tables, invoices, and receipts. Return plain text, Markdown tables, or structured JSON fields, with options to select a recognition provider and specific PDF pages.
中文简介正在准备中。当前显示英语简介。
OCR text recognition expert. ALWAYS invoke this skill when you need to extract text from images or PDFs — including invoices, receipts, ID cards, bank cards, business licenses, tables, handwritten documents, or any visual text content.
安装前可以查看交给 AI 代理的指令内容。
Two tools for optical character recognition, supporting both VLM-OCR (vision language models) and traditional OCR APIs (Baidu, Google, Azure, PaddleOCR).
Extract text from images or PDF files using OCR.
tai tool ocr_recognize --source /path/to/image.png
tai tool ocr_recognize --source https://example.com/document.jpg
tai tool ocr_recognize --source /path/to/table.png --type table --output_format markdown
tai tool ocr_recognize --source /path/to/invoice.pdf --type invoice --output_format json
tai tool ocr_recognize --source /path/to/doc.png --provider baidu
tai tool ocr_recognize --source /path/to/doc.png --provider llm:qwen-ocr --prompt "只提取表格中的金额列"
tai tool ocr_recognize --source /path/to/report.pdf --pages "1-5" --output_format markdown
| Parameter | Type | Required | Description |
|---|---|---|---|
| source | string | yes | Image or PDF file path/URL to recognize |
| provider | string | no | LLM connector ID (llm:xxx) or OCR settings key (baidu/paddleocr/google/azure). Auto-selects if omitted |
| type | string | no | Recognition type (default: general). See type table below |
| output_format | string | no | text (default), json (with coordinates/fields), or markdown (structured) |
| mode | string | no | accurate (default, best quality) or standard (faster) |
| language | string | no | Language hint (ISO 639-1, e.g. en, zh, ja). Auto-detected if omitted |
| prompt | string | no | Custom instruction for VLM-OCR only, appended to system prompt. Ignored by traditional OCR |
| pages | string | no | PDF page range, e.g. 1-5 or 1,3,7. All pages if omitted |
| extra | JSON | no | Provider-specific parameters as a JSON object |
| Type | Description | Best output_format |
|---|---|---|
general | General text (default) | text |
table | Table extraction | markdown |
handwriting | Handwritten text | text |
document | Document layout parsing | markdown |
invoice | Invoice (VAT) | json |
receipt | Receipt / ticket | json |
id_card | ID card | json |
bank_card | Bank card | json |
license | Business license | json |
vehicle_license | Vehicle license | json |
passport | Passport | json |
license_plate | License plate | json |
If the chosen provider does not support the requested type, it automatically degrades to general and annotates the response metadata with degraded_from. VLM-OCR supports all types via prompt adaptation.
List available OCR providers and their supported recognition types.
tai tool ocr_providers
Returns a list of providers including VLM-OCR models (from LLM connectors with ocr capability) and traditional API providers (from OCR settings). Each entry includes id, name, type (vlm or traditional), and supported_types.
| Provider | Notes | |
|---|---|---|
| Baidu | yes | pdf_file parameter |
| Azure | yes | Document Intelligence native support |
| PaddleOCR | yes | pdf + fileType=0 |
| no | Sync API does not support PDF | |
| VLM (llm:) | no | Vision models accept images only |
For providers that do not support PDF, use Baidu, Azure, or PaddleOCR instead.
Multi-page PDFs are automatically split page-by-page. Instead of printing all text, the tool returns a JSON summary with file paths for each page result:
{
"source": "report.pdf",
"total_pages": 10,
"pages": 3,
"results": [
{"page": 1, "file": ".tool-tmp/ocr-a1b2c3d4/page-1.txt", "preview": "Invoice No: INV-001..."},
{"page": 2, "file": ".tool-tmp/ocr-a1b2c3d4/page-2.txt", "preview": "Invoice No: INV-002..."},
{"page": 3, "file": ".tool-tmp/ocr-a1b2c3d4/page-3.txt", "preview": "Invoice No: INV-003..."}
]
}
To read full content of a specific page, use cat:
cat .tool-tmp/ocr-a1b2c3d4/page-2.txt
Single-page PDFs and images return inline text as usual (no file indirection).
output_format=text (default) when you just need the text content — simplest for LLM processingoutput_format=json for structured types (invoice, id_card, etc.) to get key-value fieldsoutput_format=markdown for documents and tables to preserve layoutprompt parameter only works with VLM-OCR providers; traditional OCR ignores itjson output to get fields with key-value pairsocr_providers first to check which providers are available and what types they supportcat <file> to read specific pages还没有评价。欢迎分享你的使用感受。
无需登录即可发表。欢迎分享使用感受、提问，或说明在哪些环境下无法运行。
还没有评论。来写第一条吧。
Manage agents on a host by listing their capabilities, reading source code, and downloading or deploying agents in the smith namespace. Check available language model connections and their roles without exposing API keys.
中文简介正在准备中。当前显示英语简介。
Turn recorded speech into text from supported audio files such as mp3, m4a, and wav. Detect the language automatically or specify it, and list available speech recognition providers and models before choosing one.
中文简介正在准备中。当前显示英语简介。
Check kanban boards, which organize tasks into columns, and retrieve task lists filtered by status, assistant, or board. Use the returned data to see task counts and review work in progress.
中文简介正在准备中。当前显示英语简介。
Start and manage long-running background services such as development servers. Check their status and output logs, verify listening ports, retrieve local URLs, and restart services while keeping the same service ID.
中文简介正在准备中。当前显示英语简介。
Classify messages, rate content, or estimate probabilities using defined question types. Receive machine-readable answers, with confidence scores for classification and ratings, and apply the same questions to multiple inputs in a batch.
中文简介正在准备中。当前显示英语简介。
Look up callable processes in Yao, inspect their arguments and return types, and check whether a process name is valid. Use the documentation to prepare calls and find suggested alternatives for misspelled names.
中文简介正在准备中。当前显示英语简介。