MCP servers that ship as code — a repository or a package you install and run locally (usually over stdio). They have no hosted endpoint, so we cannot probe or grade them: instead each one shows its repository, install command, license, stars and when it was last updated.
Collected from GitHub repositories and npm, PyPI, Cargo, Go, Maven, NuGet and RubyGems packages tagged as MCP servers, the official MCP Registry, Docker's MCP catalog, ModelScope, Baidu MCP World, Cline and other public directories; one entry per repository. A server that also offers a hosted endpoint is listed under Remote instead. These entries are not counted in our reachability, verification or readiness statistics.
-
Open-source MCP vision server that gives text-only LLMs and AI agents image understanding, OCR, visual analysis, UI inspection, and multimodal capabilities.
pypi MIT ★ 1 -
Privacy-first Multimodal Data Refinery, Repo Context Firewall (Prompt Injection & Trojan Source Defense), Offline Vietnamese OCR, Multilingual E5 ONNX + SQLite FTS5 RAG & 9-Tool MCP Server.
MIT ★ 7 -
TypeScript MCP server for image understanding, OCR and multimodal API integration
MIT -
Visual bridge for text-only models and visual enhancement toolkit for multimodal AI agents — OCR, zoom, PDF/Office rendering, web vision and interaction.
★ 20 -
dsh-qwen-mmgithub.com/RRRosmontis/dsh-qwen-mmupdated this year
Qwen-MM-Plugins integration bundle for DeepSeek Harness (dsh) — multimodal MCP tools (vision, OCR, ASR, search, video, Blender, FreeCAD) + image attachment bridge. 让 DeepSeek Harness 原生支持多模态。
MIT ★ 1 -
vision-mcp-servergithub.com/gerardodellan/vision-mcp-serverupdated this year
Governed multimodal Vision MCP server — describe/OCR/ask/detect over any OpenRouter vision model; pluggable backend, offline-testable
MIT -
Token-lean web microfetch for LLM agents: any URL → clean markdown via CLI, MCP server, and Claude Code plugin. Real browser-cookie auth, passkeys, anti-bot reach, on-by-default prompt-injection defense, plus on-device multimodal ASR/OCR. A single Rust binary — not a browser.
cargo MIT ★ 18 -
Local multimodal inference service
MIT ★ 5 -
MCP server for image generation and editing, speech recognition, text recognition, speech synthesis, voice design, video generation and multimodal media analysis.
pypi MIT -
Agent-first OCR CLI and MCP server for images, PDFs, URLs, structured data, and resumable extraction with Gemini, Kimi, Muse, OpenRouter, and compatible APIs.
npm MIT ★ 6 -
@scp3500/openvlupdated this year
让 AI 看懂图片 - 支持命令行和 MCP 服务器
npm MIT -
本地优先的多模态视觉 MCP —— 为无视觉 LLM(DeepSeek、GLM 等)提供可自定义端点的看图 / OCR / 文档解析能力。
MIT -
MCP stdio server for image recognition via an existing vision model
npm MIT -
ai-multimodal-fraud-price-enginegithub.com/tncyavci/ai-multimodal-fraud-price-engineupdated this year
Multi-modal marketplace price valuation & scam detection engine (0-100 Trust Score) using Vision OCR, feature regression, and FastMCP
-
vision-mcp-nodeupdated this year
MCP server: 直连远端多模态模型 API 识别图片/OCR(OpenAI 兼容,如阿里云百炼 qwen3-vl-plus / qwen3.7-plus),不经过本地代理
npm MIT -
visionsearch-mcpgithub.com/tiagohgms/visionsearch-mcpupdated this year
Multimodal Web Intelligence MCP — local-first image, PDF, and document research via LM Studio
npm MIT -
MCP server that compresses OCR-heavy PDFs into dense packed images so AI agents can handle long visual documents
pypi MIT ★ 1 -
Open-source multimodal MCP plugin for text-only AI agents: recognize and generate images, video, and audio through Qwen/DashScope. Supports Codex, Claude Code, and DeepSeek Harness ecosystem.
MIT ★ 4 -
look-at-deepseekupdated this year
Lightweight multi-channel MCP server for vision understanding and OCR. Integrates multi-provider failover, transient retry, and clarity-first image compression. stdio + Streamable HTTP transports.
npm MIT -
本地视觉理解 MCP 服务,为 AI agent 提供识图能力:通过 vision / ocr / list_models / providers 四个 MCP 工具描述截图、界面、图表与照片。
MIT ★ 5 -
Open-source Rust multimodal vector database & RAG platform: one unified space for text, image, audio, video & docs - hybrid search, reranking, built-in MCP server.
AGPL-3.0 -
Codex-native-long-video-memorygithub.com/NitcanKen/Codex-native-long-video-memoryupdated this year
Local-first Codex plugin that indexes long videos with OCR/ASR and answers with timestamped, SHA-verified visual evidence.
Apache-2.0 ★ 4 -
qwen-vision-mcpgithub.com/sahilchouksey/qwen-vision-mcpupdated this year
MCP server for image analysis using Qwen 3.5 via Ollama. Adds vision capabilities to models without native image support.
npm MIT -
Search your files by meaning and fact-check the web, fully offline.
AGPL-3.0 -
Perceptron MCP server for high-accuracy visual perception powered by fast, efficient vision-language models
npm Apache-2.0 -
dsh-luma-visiongithub.com/fuzzysoul/dsh-luma-visionupdated this year
Vision bridge for DeepSeek Harness (dsh): image understanding for text-only models via luma-mcp, free Qwen3-VL-Flash by default. 视觉理解插件:让纯文本模型获得看图能力(截图/报错/OCR/UI 分析)。
npm MIT -
vision-toolkitgithub.com/leiming2333/vision-toolkitupdated this year
多模态视觉 MCP + 文生图 Skill 工具包 (OpenAI GPT-4o · 通义千问 Qwen-VL · Google Gemini · Anthropic Claude)
npm MIT -
mcp-vision-bridgegithub.com/KuaaMU/mcp-vision-bridgeupdated this year
MCP server that gives text-only LLM coding agents vision — analyze images via any multimodal model (mimo, Claude, Gemini, OpenAI-compatible). Works with Claude Code, Codex, Kimi, opencode, PI.
MIT ★ 13 -
A local-first video intelligence layer for AI agents with CLI, MCP, and standalone Windows executable
MIT -
Vision sidecar MCP server — gives text-only LLMs the ability to see images. Supports OpenAI AND Anthropic API formats.
MIT -
opencode-visiongithub.com/nickrivers1983/opencode-visionupdated this year
Vision-empowered MCP server for OpenCode text-only models. PaddleOCR (SOTA deep learning) + Google Gemini fallback for model-agnostic image analysis.
pypi MIT -
Local MCP server to analyze video, audio & images — extract frames, transcribe speech, OCR on-screen text. Give Claude, Cursor & VS Code video context. No API keys, no cloud.
npm Apache-2.0 ★ 2 -
MCP server that lets AI agents analyze, describe and read text in images through the free-first three.ws vision API.
NOASSERTION ★ 10 -
Give text-only LLM agents eyes: an MCP server + CLI that describes images through a vision-language model (VLM) API.
npm MIT -
vision-primitives-mcpgithub.com/zouyuanqing/vision-primitives-mcpupdated this year
Vision-primitives MCP server for text-only LLMs: describe, locate (coordinates), OCR with bbox, annotate, crop, zoom, automated anomaly scanning, computer use — powered by Xiaomi MiMo V2.5. Single-file Python, Pillow-only dependency.
MIT ★ 9 -
Media generation, editing, and extraction as MCP tool surfaces — image, video, audio, OCR, TTS/STT. Chainable into workflows with per-step prompt engineering and quality gates. A slice of a 250+ model production roster.
go MIT -
Let Claude (or any LLM) actually watch a video — scene-aware, deduplicated frames + transcript, from a URL or local file. Runs locally, MIT.
pypi MIT ★ 2,319 -
mcp-vision-bridgeupdated this year
Give text-only LLM coding agents (DeepSeek V4 Flash/Pro, Qwen, Kimi) vision via any multimodal model. One analyze_image tool — paste or drag in one or many images, analyzed in a single call, scoped to the current session. Works with Claude Code, opencode,
npm MIT -
Self-hosted VLM image-recognition MCP server (stdio). Paste an image in Claude Code / Codex / OpenCode → get recognition. OpenAI-compatible backend (vLLM / Ollama / GLM-4V / Qwen-VL).
MIT ★ 1 -
dsh-vision-mcpgithub.com/moton16/dsh-vision-mcpupdated this year
Give your text-only LLM eyes: an MCP server that reads images via external OpenAI-compatible vision models (OpenAI, Qwen-VL, GLM, Gemini, Grok, Ollama...) with automatic provider fallback on rate-limit/errors.
npm MIT -
Offline semantic search for everything on your disk - documents, code, PDFs, scanned images (OCR), video by the second. 100% local AI: hybrid RAG with keyword + vector + rerank, no API key, no cloud. Desktop app + 26 MCP tools for Claude Code, plus a multi-engine web researcher that quotes sources verbatim. 本地离线多模态语义检索
AGPL-3.0 ★ 4 -
dsh-qwen-mm-pluginsgithub.com/ShuiHan268/dsh-qwen-mm-pluginsupdated this year
DeepSeek Harness 的 Qwen-MM-Plugins 集成插件:12 个多模态 MCP 工具(视觉/OCR/定位/ASR/音视频)、Web 设置页(粘贴 Qwen API Key 即用)、内置技能与一键安装器
Apache-2.0 ★ 4 -
GLOBAL-Old_Firm_RAG_Querygithub.com/eqramul01/GLOBAL-Old_Firm_RAG_Queryupdated this year
v2: Hybrid (BM25+vector) legal RAG over a 145-matter law firm archive with OCR'd scans readable in-chat — Gemini Embedding 2, LanceDB, Cloud Run, MCP, queryable from Claude.ai with cited source links.
MIT -
RAG-Sandbox is a comprehensive, multi-tenant Agentic RAG (Retrieval-Augmented Generation) platform that redefines how users interact with their data. It moves beyond simple text retrieval to offer a fully multimodal, intelligent experience—transforming static documents into conversable agents, podcasts, and immersive stories.
MIT