OCR.space MCP by usefulapi is an MCP server published by io.usefulapi, hosted at ocr-space.usefulapi.io. When last checked on 2026-11-22 it was working: it answers and lists its tools. It declares 4 tools, including ocrspace_ocr_url, ocrspace_ocr_base64, ocrspace_create_searchable_pdf, ocrspace_list_languages.
Read text and tables from images and PDFs, and make searchable PDFs, in OCR.space.
4,109 connections through 5 shared values. Tap any node.
Inferred from the host name and headers: what a server runs on, not who runs it.
Extract text from an image or PDF at a public https URL with OCR.space (POST /parse/image). Returns the text per page (a PDF gives one entry per page; the free plan reads the first 3 pages and files up to 1 MB). Uses one request of your OCR.space quota (free key: 25,000 requests per month and 500 requests per day). Options: language, engine (1-3), is_table (line-by-line text for tables, receipts, invoices), detect_orientation, scale (upscale low-resolution scans), include_overlay (word coordinates; slower).
| Parameter | Type | Description |
|---|---|---|
| url * | string | Public https URL of the image or PDF (the file must be reachable without login and served with an image/* or application/pdf content type). |
| scale | boolean | Upscale internally; helps low-resolution PDF scans. |
| engine | integer |
None yet. Changes appear after a server's second daily check.
Squares are this server's values; dots are other servers sharing them. A shared address or publisher is a lead, not proof of a common operator.
| OCR engine 1, 2 or 3. Default 2 (best all-round). 3 = highest accuracy, handwriting, tables as Markdown, 200+ languages, slower, and has its own smaller quota. 1 = fastest but deprecated. |
| filetype | string | Force the file type when the URL's content type is wrong (default: detected). |
| is_table | boolean | Return text line by line, best for tables, receipts and invoices. |
| language | string | Document language, a 3-letter code such as eng, ger, fre, spa, chs (see ocrspace_list_languages). Default eng. 'auto' works with engine 2 and 3 only. |
| include_overlay | boolean | Also return word coordinates (left/top/width/height in pixels), capped at 1000 words. |
| detect_orientation | boolean | Auto-rotate the image; the page's text_orientation reports the rotation. |
{
"name": "ocrspace_ocr_url",
"title": "OCR an image or PDF from a URL",
"description": "Extract text from an image or PDF at a public https URL with OCR.space (POST /parse/image). Returns the text per page (a PDF gives one entry per page; the free plan reads the first 3 pages and files up to 1 MB). Uses one request of your OCR.space quota (free key: 25,000 requests per month and 500 requests per day). Options: language, engine (1-3), is_table (line-by-line text for tables, receipts, invoices), detect_orientation, scale (upscale low-resolution scans), include_overlay (word coordinates; slower).",
"inputSchema": {
"type": "object",
"$schema": "https://json-schema.org/draft/2020-12/schema",
"required": [
"url"
],
"properties": {
"url": {
"type": "string",
"maxLength": 2048,
"minLength": 8,
"description": "Public https URL of the image or PDF (the file must be reachable without login and served with an image/* or application/pdf content type)."
},
"scale": {
"type": "boolean",
"description": "Upscale internally; helps low-resolution PDF scans."
},
"engine": {
"type": "integer",
"maximum": 3,
"minimum": 1,
"description": "OCR engine 1, 2 or 3. Default 2 (best all-round). 3 = highest accuracy, handwriting, tables as Markdown, 200+ languages, slower, and has its own smaller quota. 1 = fastest but deprecated."
},
"filetype": {
"enum": [
"PDF",
"GIF",
"PNG",
"JPG",
"TIF",
"BMP"
],
"type": "string",
"description": "Force the file type when the URL's content type is wrong (default: detected)."
},
"is_table": {
"type": "boolean",
"description": "Return text line by line, best for tables, receipts and invoices."
},
"language": {
"enum": [
"ara",
"bul",
"chs",
"cht",
"hrv",
"cze",
"dan",
"dut",
"eng",
"fin",
"fre",
"ger",
"gre",
"hun",
"kor",
"ita",
"jpn",
"pol",
"por",
"rus",
"slv",
"spa",
"swe",
"tha",
"tur",
"ukr",
"vnm",
"auto"
],
"type": "string",
"description": "Document language, a 3-letter code such as eng, ger, fre, spa, chs (see ocrspace_list_languages). Default eng. 'auto' works with engine 2 and 3 only."
},
"include_overlay": {
"type": "boolean",
"description": "Also return word coordinates (left/top/width/height in pixels), capped at 1000 words."
},
"detect_orientation": {
"type": "boolean",
"description": "Auto-rotate the image; the page's text_orientation reports the rotation."
}
}
},
"annotations": {
"title": "OCR an image or PDF from a URL",
"readOnlyHint": false,
"openWorldHint": true,
"idempotentHint": false,
"destructiveHint": false
}
}Extract text from a small file passed as base64 (POST /parse/image with base64Image). Up to 1,400,000 base64 characters (about 1 MB, the free plan's file limit); use ocrspace_ocr_url for bigger files. Pass either a full data URI (data:image/png;base64,...) or the raw base64 plus mime_type. Uses one request of your OCR.space quota (free key: 25,000 requests per month and 500 requests per day). Options as in ocrspace_ocr_url.
| Parameter | Type | Description |
|---|---|---|
| scale | boolean | Upscale internally; helps low-resolution PDF scans. |
| engine | integer | OCR engine 1, 2 or 3. Default 2 (best all-round). 3 = highest accuracy, handwriting, tables as Markdown, 200+ languages, slower, and has its own smaller quota. 1 = fastest but deprecated. |
| filetype | string | Force the file type when the URL's content type is wrong (default: detected). |
| is_table | boolean | Return text line by line, best for tables, receipts and invoices. |
| language | string | Document language, a 3-letter code such as eng, ger, fre, spa, chs (see ocrspace_list_languages). Default eng. 'auto' works with engine 2 and 3 only. |
| mime_type | string | MIME type of the file; required when image_base64 has no data: prefix. |
| image_base64 * | string | The file as base64 (a data: URI, or raw base64 together with mime_type). Whitespace and newlines are removed. |
| include_overlay | boolean | Also return word coordinates, capped at 1000 words. |
| detect_orientation | boolean | Auto-rotate the image; the page's text_orientation reports the rotation. |
{
"name": "ocrspace_ocr_base64",
"title": "OCR a base64-encoded image or PDF",
"description": "Extract text from a small file passed as base64 (POST /parse/image with base64Image). Up to 1,400,000 base64 characters (about 1 MB, the free plan's file limit); use ocrspace_ocr_url for bigger files. Pass either a full data URI (data:image/png;base64,...) or the raw base64 plus mime_type. Uses one request of your OCR.space quota (free key: 25,000 requests per month and 500 requests per day). Options as in ocrspace_ocr_url.",
"inputSchema": {
"type": "object",
"$schema": "https://json-schema.org/draft/2020-12/schema",
"required": [
"image_base64"
],
"properties": {
"scale": {
"type": "boolean",
"description": "Upscale internally; helps low-resolution PDF scans."
},
"engine": {
"type": "integer",
"maximum": 3,
"minimum": 1,
"description": "OCR engine 1, 2 or 3. Default 2 (best all-round). 3 = highest accuracy, handwriting, tables as Markdown, 200+ languages, slower, and has its own smaller quota. 1 = fastest but deprecated."
},
"filetype": {
"enum": [
"PDF",
"GIF",
"PNG",
"JPG",
"TIF",
"BMP"
],
"type": "string",
"description": "Force the file type when the URL's content type is wrong (default: detected)."
},
"is_table": {
"type": "boolean",
"description": "Return text line by line, best for tables, receipts and invoices."
},
"language": {
"enum": [
"ara",
"bul",
"chs",
"cht",
"hrv",
"cze",
"dan",
"dut",
"eng",
"fin",
"fre",
"ger",
"gre",
"hun",
"kor",
"ita",
"jpn",
"pol",
"por",
"rus",
"slv",
"spa",
"swe",
"tha",
"tur",
"ukr",
"vnm",
"auto"
],
"type": "string",
"description": "Document language, a 3-letter code such as eng, ger, fre, spa, chs (see ocrspace_list_languages). Default eng. 'auto' works with engine 2 and 3 only."
},
"mime_type": {
"enum": [
"image/jpeg",
"image/png",
"image/gif",
"image/bmp",
"image/tiff",
"application/pdf"
],
"type": "string",
"description": "MIME type of the file; required when image_base64 has no data: prefix."
},
"image_base64": {
"type": "string",
"maxLength": 1400200,
"minLength": 8,
"description": "The file as base64 (a data: URI, or raw base64 together with mime_type). Whitespace and newlines are removed."
},
"include_overlay": {
"type": "boolean",
"description": "Also return word coordinates, capped at 1000 words."
},
"detect_orientation": {
"type": "boolean",
"description": "Auto-rotate the image; the page's text_orientation reports the rotation."
}
}
},
"annotations": {
"title": "OCR a base64-encoded image or PDF",
"readOnlyHint": false,
"openWorldHint": true,
"idempotentHint": false,
"destructiveHint": false
}
}Run OCR on a public https image/PDF URL and have OCR.space build a searchable PDF (a copy of the file with a text layer). Returns searchable_pdf_url, a download link that expires after one hour, plus the recognized text. The free plan adds a 'Generated by OCR.space' watermark. Engine 3 cannot make searchable PDFs, so only engine 1 or 2 is accepted (default 2). Uses one request of your OCR.space quota (free key: 25,000 requests per month and 500 requests per day).
| Parameter | Type | Description |
|---|---|---|
| url * | string | Public https URL of the image or PDF. |
| engine | integer | OCR engine 1 or 2 (default 2); engine 3 does not create searchable PDFs. |
| filetype | string | Force the file type when the URL's content type is wrong (default: detected). |
| language | string | Document language, a 3-letter code such as eng, ger, fre, spa, chs (see ocrspace_list_languages). Default eng. 'auto' works with engine 2 and 3 only. |
| hide_text_layer | boolean | Hide the text layer (default false = visible text layer). The text layer exists either way. |
{
"name": "ocrspace_create_searchable_pdf",
"title": "Create a searchable PDF from a URL",
"description": "Run OCR on a public https image/PDF URL and have OCR.space build a searchable PDF (a copy of the file with a text layer). Returns searchable_pdf_url, a download link that expires after one hour, plus the recognized text. The free plan adds a 'Generated by OCR.space' watermark. Engine 3 cannot make searchable PDFs, so only engine 1 or 2 is accepted (default 2). Uses one request of your OCR.space quota (free key: 25,000 requests per month and 500 requests per day).",
"inputSchema": {
"type": "object",
"$schema": "https://json-schema.org/draft/2020-12/schema",
"required": [
"url"
],
"properties": {
"url": {
"type": "string",
"maxLength": 2048,
"minLength": 8,
"description": "Public https URL of the image or PDF."
},
"engine": {
"type": "integer",
"maximum": 2,
"minimum": 1,
"description": "OCR engine 1 or 2 (default 2); engine 3 does not create searchable PDFs."
},
"filetype": {
"enum": [
"PDF",
"GIF",
"PNG",
"JPG",
"TIF",
"BMP"
],
"type": "string",
"description": "Force the file type when the URL's content type is wrong (default: detected)."
},
"language": {
"enum": [
"ara",
"bul",
"chs",
"cht",
"hrv",
"cze",
"dan",
"dut",
"eng",
"fin",
"fre",
"ger",
"gre",
"hun",
"kor",
"ita",
"jpn",
"pol",
"por",
"rus",
"slv",
"spa",
"swe",
"tha",
"tur",
"ukr",
"vnm",
"auto"
],
"type": "string",
"description": "Document language, a 3-letter code such as eng, ger, fre, spa, chs (see ocrspace_list_languages). Default eng. 'auto' works with engine 2 and 3 only."
},
"hide_text_layer": {
"type": "boolean",
"description": "Hide the text layer (default false = visible text layer). The text layer exists either way."
}
}
},
"annotations": {
"title": "Create a searchable PDF from a URL",
"readOnlyHint": false,
"openWorldHint": true,
"idempotentHint": false,
"destructiveHint": false
}
}Reference (no request to OCR.space, no quota used): the documented language codes for the language option, what OCR engines 1, 2 and 3 do, and the free-plan limits.
No parameters.
{
"name": "ocrspace_list_languages",
"title": "List OCR languages, engines and limits",
"description": "Reference (no request to OCR.space, no quota used): the documented language codes for the language option, what OCR engines 1, 2 and 3 do, and the free-plan limits.",
"inputSchema": {
"type": "object",
"$schema": "https://json-schema.org/draft/2020-12/schema",
"properties": {}
},
"annotations": {
"title": "List OCR languages, engines and limits",
"readOnlyHint": true,
"openWorldHint": false,
"destructiveHint": false
}
}Tools for OCR.space, hosted by usefulapi.io. If the user is new or asks what this server can do, suggest requests such as: "Read the text from this receipt photo: https://example.com/receipt.jpg"; "Extract the table from this scanned invoice PDF as Markdown with engine 3."; "Make a searchable PDF from this scanned contract URL.". Read-only tools never change data; ask the user before a tool that creates, changes or deletes OCR.space data. Free plan: 50 tool calls per month. Run ocrspace_usage_status to see the calls used this month. If the user asks for something none of these tools can do, tell the user and, with their OK, run ocrspace_request_feature to record the request. Paid plans: Pro (10,000 tool calls a month, $9/month or $90/year, for normal use) and Pro+ (100,000 tool calls a month, $39/month or $390/year, for automations and heavy use). A plan can be cancelled at any time and then runs to the end of the paid period, so it also suits a one-time job. When the user wants more calls or reaches the limit, run ocrspace_upgrade without a plan: it returns both plans, each with a payment link for the user to open. ocrspace_cancel_subscription cancels Pro at the end of the paid period.
Names, descriptions and schemas were written by the server and are shown as data.
Shared addresses on big platforms (Cloudflare, Vercel and the like) put unrelated servers together; those are marked shared_infrastructure. A match is a pivot, not a link between operators.