Run optical character recognition on an image (png, jpg, webp, bmp) and return the recognized text with a confidence score. Supports 100+ languages via the language parameter (ISO 639-2 codes like 'eng', 'deu', 'fra', 'spa').
ConnectorNo auth
675,397 tools. Updated 2026-10-19 05:13
"Optical Character Recognition (OCR) Technology" matching MCP tools:
- Use this when the user asks which companies use a particular technology (for example Intercom, Zendesk, HubSpot, Shopify, Stripe), or which technologies are most used. With a slug: companies detected using that technology, 20 per page. company_count is every company detected using it; listed_count is how many the list can page through, so stop at page ceil(listed_count / 20). Without a slug: the technology index, most-used first, with company counts (default 100, up to 1,000). Slugs are lower-case with hyphens, e.g. intercom, hubspot, google-analytics, shopify. Works without an API key; no contact reveals.ConnectorNo auth
- Search Helium's meme database by text (OCR + caption). Returns matching memes ranked by relevance. Each result includes: - id, caption, ocr (text extracted from the image) - image: full URL to the meme image - source: origin platform (e.g. 'reddit') - num_likes: likes/upvotes on the original post - date, is_video, rank Args: query: Search keywords (required). Matched against OCR text and captions. limit: Max results (1-100, default 20). days_back: Only include memes from the last N days. 0 means no date filter (default).ConnectorNo auth
- Use this ONLY when the text exists as PIXELS and nothing else — a photo, a screenshot, a scan. It runs optical character recognition on an image and GUESSES the characters, so it is a best-effort transcription that misreads under blur, skew or low contrast. Returns JSON { text, language, confidence? }. If the file already stores real characters, this is the wrong tool and will be less accurate: parse_document reads them exactly. The deciding question is what the bytes contain, never the file extension — a .png of a letter needs this tool, a .txt never does. Layout is not preserved — no tables, columns or coordinates, just a flat string. Paid compute; 20 calls per hour per IP. There is no upload channel over MCP: pass `fileUrl`, a URL in razi.pro's own storage; third-party URLs are rejected. Obtain one by uploading the image over the REST API first (POST /api/v1/tools/execute with the file attached).ConnectorNo auth
- Use this ONLY when the text exists as PIXELS and nothing else — a photo, a screenshot, a scan. It runs optical character recognition on an image and GUESSES the characters, so it is a best-effort transcription that misreads under blur, skew or low contrast. Returns JSON { text, language, confidence? }. If the file already stores real characters, this is the wrong tool and will be less accurate: parse_document reads them exactly. The deciding question is what the bytes contain, never the file extension — a .png of a letter needs this tool, a .txt never does. Layout is not preserved — no tables, columns or coordinates, just a flat string. Paid compute; 20 calls per hour per IP. There is no upload channel over MCP: pass `fileUrl`, a URL in razi.pro's own storage; third-party URLs are rejected. Obtain one by uploading the image over the REST API first (POST /api/v1/tools/execute with the file attached).ConnectorNo auth
- Turn files into text on the server: a PDF (its text layer, or OCR for scanned pages), or photos or scans of pages (one image per page, in reading order). Also reads plain-text files. Use it before import_script for a screenplay that arrives as a PDF or as page images: then call import_script with the textId this returns. Never retype a long script with edit_script. For notes and other documents, read the returned text. Pages with a text layer are free. Each page read by OCR costs about half a credit; when more than 20 pages need OCR, the call returns confirmation_required with the estimate: ask the user, then call again with confirm: true. At most 300 pages per file are read. Returns the textId, per file the pages read by text layer and by OCR and any failedPages, the OCR spend of this call, and the text in pages of 60000 characters (pass nextOffset as offset for the next page). Calling again with the same files costs nothing for pages already read and retries failed pages.ConnectorOAuth
Matching MCP Servers
- AlicenseBqualityBmaintenanceCompress OCR-heavy PDFs into dense packed images so agents can work with long visual documents.342 PyPI1MIT
- AlicenseNot gradedqualityFmaintenanceMCP server for intelligent circuit diagram recognition, enabling detection of components, text, topology, and netlist from circuit images.5MIT
Matching MCP Connectors
Optical education, prescription transposition, horizontal decentration and approximate prism. PT/EN.
OCR for images and Korean ID documents
- Filter pages within a world by one or more tags. The page's documentType is concatenated into the same keyword set imagine searches, so a tag query of 'character' also matches every page with documentType=character. Use mode='and' to require every tag; mode='or' (default) returns pages matching any tag. Common patterns: all NPCs (tag='npc'), all locations (tag='location'), every character-type page (tag='character').ConnectorOAuth
- Find which files in a repository use a given technology, library or framework — with the line number and the import keyword that proves it. Use after code_analyze reports a technology in tech_stack and you need to know which files it is actually in, rather than searching for import statements by hand. Returns file, line, evidence keyword and language. COVERAGE LIMIT: technology detection reads only the first 150 lines of each file, so a file that imports the technology further down, or uses it without a top-of-file import, does not appear. Absence from these results is not proof the file does not use the technology. The count matches the files_using value code_analyze reports for the same technology. Shares the analysis cache with code_analyze — no duplicate work if analyze already ran.ConnectorNo auth
- [PAID — 0.04 USDC on Base via x402] Deterministic PDF -> per-page plain text + metadata, no OCR, no table extraction, no third-party PDF/OCR vendor — text pulled directly from the PDF's own embedded text objects via pdfjs-dist (Mozilla's own PDF.js, Apache-2.0) running in this process. Pass exactly one of `url` (public http(s) link, SSRF-hardened fetch) or `pdf_base64` (raw base64 bytes). Max 10 MiB, max 40 pages processed per call. Scanned/image-only PDFs return empty page text with a warning, never a fabricated OCR guess. $0.04 USDC on Base, paid via x402 by the calling agent's own wallet. Call with no payment_header first to receive the payment requirements; your own wallet pays, never this server's.ConnectorNo auth
- Search Meta Careers open-job result records with optional team/technology, location, role, remote, and keyword filters; filters combine with AND, and no filters returns the open-requisition catalog. `q` matches team, technology, location, or req codes—not job titles/descriptions. Use meta_jobs_job for one posting or meta_jobs_list for list discovery.ConnectorNo auth
- Ranked search over Tangent's catalogue of design and design-engineering resources (sites, products, component libraries, tools, articles). Use it when you know roughly what you want; narrow with vertical, category, technology, type, pricing or open_source, and page with cursor. For a whole brief, use find_inspiration. Example: {"query": "command menu", "technology": "react", "limit": 5}ConnectorNo auth
- Search Meta Careers open-job result records with optional team/technology, location, role, remote, and keyword filters; filters combine with AND, and no filters returns the open-requisition catalog. `q` matches team, technology, location, or req codes—not job titles/descriptions. Use meta_jobs_job for one posting or meta_jobs_list for list discovery.ConnectorNo auth
- Watermark a document per recipient and trace a leaked copy. Mode 'mark' takes text and a list of recipient names (up to 255) and returns one copy per recipient carrying that recipient's number as an 18-character invisible marker (U+2060 frame, 8-bit number plus 8-bit check written in U+200B and U+200C) after the first word and every twentieth word, plus a tracer key that stores the list; set spacing to also encode the pattern in no-break spaces. Mode 'trace' takes a leaked copy plus the same recipients or the key and returns the recipient, the copy number, how many markers were found and agreed, and a confidence level. Marks survive copy and paste but not retyping, screenshots, OCR or a hidden-character cleaner. Source: https://hopi.co.uk/leak-tracer/ConnectorNo auth
- A YouTube video's captions as plain text and timestamped segments, with the video's title, channel, duration, views, publish date, chapters and the list of caption languages. Accepts watch, youtu.be, shorts, embed and live links or the 11-character id; pick a language with lang (default English, else the first track). Auto-generated captions contain recognition errors. A video without captions is a 404 and not charged. Price: $0.01 in USDC per successful call, paid with x402 (Base, Solana or Arc); failed calls are not charged.ConnectorNo auth
- Use this when the user asks where AudD sends their stream recognition results. Returns the account's callback URL, or null when AudD reports that none is set (a temporary AudD error can look the same, so confirm with the user before replacing it). Read-only.ConnectorOAuth
- List the renter’s private inbound forwarding aliases on forward.mailbox.bot. These are the unique intake email addresses an operator, assistant, provider, or external agent can forward scans, PDFs, photos, provider notices, notes, and other context-aware documents to so mailbox.bot can build OCR-backed inbound context. Forwarding/emailing attachments here initiates OCR/extraction; this tool discovers the address and does not upload files directly into OCR. The alias is member-scoped, so live and sandbox agent keys for the same member resolve to the same intake address.ConnectorNo auth
- Free: return the page count and exact cost_usd to read a document URL with read_document, without running OCR. Use before reading large documents.ConnectorNo auth
- Evaluate visual rhythm from a rendered page's geometry. Call with no arguments to get a DevTools snippet to paste into your page — it prints {elements, viewport} JSON. Call again with that JSON to get alignment, gap-rhythm, and optical-balance scores. This is the complement to audit_page for things only visible once rendered.ConnectorNo auth
- Extract the spoken-word transcript from a video URL. Works even when the video has no captions (server-side speech recognition). Supports TikTok, X/Twitter, TED, Twitch VODs, podcast hosts and 1000+ other sites. Max video length: 30 minutes.ConnectorNo auth
- Designers and design engineers in the catalogue, featured first, then by name, with optional technology and role filters and cursor paging. Use search_people to find someone by name or focus. Example: {"role": "design engineer", "technology": "nextjs", "limit": 20}ConnectorNo auth
- Return fixture-backed product-page handoff details for one lens, including DynamoDB-sourced optical specs and gated datasheet policy. Product-page/catalog optical fields are not a substitute for sensor-specific FoV; call calculate_field_of_view for the lens/sensor pair. FoV rule: never estimate sensor-specific FoV from catalog fields; use calculate_field_of_view or match_lens_to_sensor. Use read_shopify_products for live product URL, price, availability, variant IDs, and metafields.ConnectorNo auth