# OCR-AI

> Hebrew and English OCR for PDFs and images: text, key fields and tables, page by page.

- name: com.ocr-ai/ocr-ai
- version: 1.0.0
- connection class: R1 (autonomous)
- trust: 81/100 flags: no-repository
- quality: 85/100 (new)
- owner: not claimed
- descriptor: https://api.protogrid.dev/v1/servers/com.ocr-ai%2Focr-ai
- tools: https://api.protogrid.dev/v1/servers/com.ocr-ai%2Focr-ai/tools

**An agent can connect on its own if it already holds `OCR_AI_TOKEN`.** Remote endpoint behind a static API key or header. The registry never sees the value; the agent fills the placeholder.

## Connection (mcpServers)

```json
{
  "mcpServers": {
    "ocr-ai": {
      "type": "http",
      "url": "https://ocr-ai.com/mcp",
      "headers": {
        "Authorization": "Bearer ${OCR_AI_TOKEN}"
      }
    }
  }
}
```

Placeholders to fill: `${OCR_AI_TOKEN}`.

## Trust

- hygiene: 80
- liveness: 96
- freshness: 100
- provenance: 55
- drivers: +dns-namespace +website +reachable +uptime-91% +updated-3d-ago -no-repository
- Derived from observable signals (official registry feed and our own credential-free probes); no code audit performed.

## Quality 85/100 (new)

### protocol: 75 (weight 25)

- warn `protocol.modern`: newest supported version is 2025-11-25; 2026-07-28 not supported, older versions are deprecated until 2027-07-28
- n/a `protocol.stateless`: only applies to 2026-07-28 servers
- pass `protocol.transport`: streamable HTTP
- n/a `protocol.list_ttl`: only applies to 2026-07-28 servers

### authorization: n/a (weight 15)

- n/a `auth.prm`, `auth.as_metadata`, `auth.cimd`: no remote uses OAuth
- n/a `auth.secret_in_url`: no templated URL

### tool hygiene: 93 (weight 30)

- pass `tools.descriptions`: every tool has a description
- pass `tools.description_length`: descriptions are concise
- pass `tools.schemas`: every tool has an object input schema
- pass `tools.annotations`: 3 of 3 tools declare readOnlyHint or destructiveHint
- warn `tools.directory_hints`: 1 of 3 tools declare all four hints; missing: buy_pages (destructiveHint, idempotentHint); check_balance (destructiveHint, idempotentHint)
- pass `tools.token_cost`: about 559 tokens to load every tool
- pass `tools.api_dump`: tools are not a one-to-one API dump

### stability: n/a (weight 15)

- n/a `stability.changes`, `stability.rug_pull`: tool history recorded for 2 of the 7 days needed

### dependencies: n/a (weight 15)

- n/a `deps.known_vulns`, `deps.mcp_sdk_version`, `deps.resolvable`: no npm or PyPI package

- tools: 3, about 559 tokens to load them all
- tool set hash: a6419f8770457dfecca9c4a4fc414b9f
- badge: [![protogrid quality](https://protogrid.dev/badge/com.ocr-ai/ocr-ai.svg)](https://protogrid.dev/servers/com.ocr-ai/ocr-ai)
- Checks run on what our credential-free, read-only probes observe; tools are never called and no code audit is performed.

### Tool changes

_No tool definition changes recorded._

## Remotes

- https://ocr-ai.com/mcp (streamable-http, auth api_key ${OCR_AI_TOKEN}, reachable true, uptime30d 0.9166667)

## Packages

_none_

## Tools (3)

- `buy_pages`: Return a Stripe Checkout link for a page pack. Give the link to your user: they pay in their browser, and the pages are added to this key's account as soon as the payment completes. This tool never charges anything by itself.
- `check_balance`: How many pages the API key's account can still read, and a link where the user can add pages.
- `read_document`: Read a PDF or an image (JPEG, PNG, WebP) and return its text, its key fields (vendor, dates, totals, VAT, ids…) and its tables, page by page. Built for Hebrew and English documents, including mixed right-to-left lines. Costs one page from the key's account per page read; a page that fails is not charged. Up to 20 pages per call: read longer PDFs in ranges with "pages". A call of many pages can take more than a minute. Pages your client stops waiting for may still be charged, so if a call times out, read smaller ranges, for example 5 pages per call.