OCR MCP
Converts screenshots of web pages into structured Markdown using GLM-OCR (Zhipu), with tools for OCR on images, webpages, and health checks.
README
OCR MCP
MCP server che converte screenshot di pagine web in Markdown strutturato usando GLM-OCR (Zhipu). Si registra nel mcp-gateway: i tool compaiono in Claude come mcp__mcp-gateway__ocr__*.
Tool
| Tool | Cosa fa |
|---|---|
ocr_image(image, instructions="") |
OCR di un'immagine già catturata. image = base64, data URI, o URL http(s) a un'immagine. |
ocr_webpage(url, full_page=True, viewport_width=0, wait_until="networkidle", instructions="") |
Cattura uno screenshot full-page dell'URL con Chromium headless, poi OCR. |
ocr_health() |
Stato e configurazione (host GLM-OCR, API style, modello, key, Chromium, tiling). Nessuna chiamata a pagamento. |
I tool restituiscono il Markdown come testo. Il server NON scrive file .md: gira su un LXC remoto senza accesso al disco locale. È la sessione Claude a salvare il .md dove serve (es. "salva il risultato in ./pagina.md").
Motore OCR — .env
Il server è agnostico all'endpoint: si passa da cloud a self-hosted cambiando solo il .env.
| Variabile | Cloud Zhipu | vLLM self-hosted |
|---|---|---|
GLM_OCR_BASE_URL |
https://api.z.ai/api/paas/v4 |
http://<host-vllm>:8000 |
GLM_OCR_API_STYLE |
layout_parsing |
chat |
GLM_OCR_MODEL |
glm-ocr |
zai-org/GLM-OCR |
GLM_OCR_API_KEY |
chiave API Zhipu | token del server vLLM |
chat→POST {BASE}/v1/chat/completions(OpenAI-compatibile).layout_parsing→POST {BASE}/layout_parsing(endpoint dedicato Zhipu).
Screenshot molto alti: tagliati in tile sovrapposte (MAX_TILE_HEIGHT / TILE_OVERLAP / MAX_TILES) e il Markdown viene ricucito automaticamente.
Deploy (CT102)
./deploy/install.sh root@192.168.1.107
rsync → /opt/ocr-mcp, crea venv, playwright install --with-deps chromium, installa e avvia il servizio systemd ocr-mcp. Poi su CT102 compila /opt/ocr-mcp/.env e systemctl restart ocr-mcp.
Registrazione nel gateway
Su CT102, aggiungere alla lista upstreams: in /etc/mcp-gateway/upstreams.yaml:
- name: ocr
url: http://127.0.0.1:8011/sse
tags:
- core
- ocr
poi systemctl restart mcp-gateway.
Porta 8011: la 8009 è occupata da
gmaile la 8008 da un altro servizio. Se va cambiata, aggiornareconfig.py,.env,deploy/ocr-mcp.servicee la voceupstreams.yaml. Il tagcorerende i tool visibili nelle sessioni che caricano il set core del gateway.
Verifica
curl http://127.0.0.1:8011/→ 200 con stato.- In Claude Code
/mcp→mcp__mcp-gateway__ocr__ocr_image | ocr_webpage | ocr_health. - Chiamare
ocr_health, poiocr_webpagesu una pagina lunga reale.
Stack
Python 3.12 · FastMCP (mcp==1.27.0) · Uvicorn · httpx · Playwright/Chromium · Pillow. Transport SSE su /sse, servizio systemd, log di audit dei job in SQLite (/var/lib/ocr-mcp/ocr_mcp.db).
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.