{"$schema":"https://wellknown.network/schemas/agent-record-v1.json","schemaVersion":"1","id":"ag_8pt8cma2zq4r","handle":"openeye-mcp","url":"https://wellknown.network/agents/openeye-mcp","links":{"self":"https://wellknown.network/agents/openeye-mcp/record.json","html":"https://wellknown.network/agents/openeye-mcp","markdown":"https://wellknown.network/agents/openeye-mcp/record.md","api":"https://wellknown.network/api/v1/agents/openeye-mcp","status":"https://wellknown.network/api/v1/agents/openeye-mcp/status","claim":"https://wellknown.network/agents/openeye-mcp/claim","claimApi":"https://wellknown.network/api/v1/claims","claimDescriptor":"https://wellknown.network/agents/openeye-mcp/claim.json","badge":"https://wellknown.network/agents/openeye-mcp/badge.svg","openapi":"https://wellknown.network/openapi.json","history":"https://wellknown.network/api/v1/agents/openeye-mcp/history","tools":"https://wellknown.network/api/v1/agents/openeye-mcp/tools"},"ard":{"identifier":"urn:air::server:openeye-mcp","type":"application/mcp-server-card+json"},"kind":"mcp_server","declared":{"name":"openeye-mcp","summary":"为纯文本大模型提供视觉能力的 MCP Server：读图 · 识字 · 解表格 · 复刻 UI。Give text-only LLMs vision via MCP: image description, OCR, VQA, UI layout & tables.","description":"<div align=\"center\"> \n \n# OpenEye \n \n**为纯文本大模型提供视觉能力的 MCP Server —— 让不会看图的模型也能读图、识字、解表格、复刻 UI。** \n \n<a href=\"https://pypi.org/project/openeye-mcp/\"><img src=\"https://img.shields.io/pypi/v/openeye-mcp.svg\" alt=\"PyPI version\"></a> \n<img src=\"https://img.shields.io/pypi/pyversions/openeye-mcp.svg\" alt=\"Python 3.11+\"> \n<a href=\"LICENSE\"><img src=\"https://img.shields.io/badge/License-MIT-green.svg\" alt=\"MIT License\"></a> \n<img src=\"https://img.shields.io/badge/transport-stdio-blueviolet.svg\" alt=\"MCP stdio transport\"> \n \n</div> \n \nOpenEye 是一个 stdio 类型的 [MCP](https://modelcontextprotocol.io) 服务器：把视觉模型（GPT / Claude / Gemini / 通义 / 智谱 / 本地 Ollama 等）包装成 6 个 MCP 工具，接进 Claude Code、Cursor、Cline 等 coding agent，原本只能处理文本的模型就获得了识图能力——分析 UI 截图、照抄报错文字、读表格、按图还原前端。 \n \n## 功能特性 \n \n- **6 个识图工具**：图像描述、OCR 取字、视觉问答、UI 布局结构化分析、表格转 Markdown、多图批量分析 + 跨图汇总 \n- **5 类视觉后端**：OpenAI 兼容协议（默认）/ Anthropic Messages / OpenAI Responses / Gemini generateContent / Gemini Interactions，一个 `VISION_PROVIDER` 切换 \n- **任意 OpenAI 兼容厂商**：改 `OPENAI_BASE_URL` 就换厂商——通义、智谱、阶跃星辰、Kimi、DeepSeek、本地 Ollama / vLLM 都能直接接 \n- **三种图片来源**：本地路径、`http(s)` 公网 URL、`data:image/...;base64,` data URI，同一入口自动识别 \n- **结构化输出**：布局与表格走 JSON 模式，端点不支持 `response_format` 时自动降级并用容错解析兜底 \n- **省 token 与提速**：图片等比压缩（JPEG 化）、结果缓存（LRU + TTL）、已解析图片缓存，重复问同一张图不再重复上传 \n- **进度上报**：支持 MCP `progressToken` 的客户端能看到「正在请求视觉后端（第 N 次）」这类实时进度 \n- **默认安全的取图链路**：SSRF 防护（DNS 固定 + 逐跳重定向校验）、图片大小上限、凭据脱敏、可选本地目录白名单、单次调用总时间预算 \n- **零 SDK 依赖**：只用 `httpx` 手写协议，不装各厂商 SDK \n \n## 安装 \n \n需要 Python 3.11+、一个支持图片输入的视觉模型 API Key（纯文本模型无法看图），以及任一支持 MCP 的客户端（Claude Code / Cursor / Cline / Windsurf / Codex CLI / Zed / Continue / Roo Code / opencode 等）。 \n \n```bash \npip install openeye-mcp \n``` \n \n**虚拟环境安装**（推荐；客户端配置里用该环境的绝对路径）： \n \n```bash \npython -m venv .venv \nsource .venv/bin/activate        # Windows: .venv\\Scripts\\activate \npip install openeye-mcp \n``` \n \n**从源码安装**（开发与贡献）： \n \n```bash \ngit clone https://github.com/ouli-1242/openeye-mcp.git \ncd openeye-mcp \npip install -e \".[dev]\"          # edit…","publisher":null,"homepage":"https://github.com/ouli-1242/openeye-mcp","repository":"https://github.com/ouli-1242/openeye-mcp/releases","version":"0.3.1","license":null,"protocols":["mcp"],"tags":["mcp","vision","ocr","openai","llm","model-context-protocol","gemini","deepseek","visual-question-answering","image-processing"],"pricing":null,"endpoints":[{"url":"pypi:openeye-mcp","type":"package_pypi","auth":null,"probeable":false}],"skills":null,"tools":null,"extra":null,"attribution":{"kind":"pypi","name":"pypi","repoUrl":"pypi","summary":"pypi","version":"pypi","description":"pypi","homepageUrl":"pypi"}},"derived":{"capabilities":[{"slug":"media.image-understanding","name":"Image Understanding","confidence":1,"provenance":"declared"},{"slug":"documents.ocr","name":"OCR","confidence":1,"provenance":"declared"},{"slug":"dev.version-control","name":"Version Control","confidence":0.745,"provenance":"derived"}],"categories":["dev","documents","media"],"language":"fr"},"observed":{"status":"unknown","statusReason":"Distributed as a package to run locally; no network endpoint to check.","lastOkAt":null,"lastProbedAt":null,"statusComputedAt":null,"reliability30d":null,"latestObservations":[],"tools":null,"package":{"name":"openeye-mcp","registry":"pypi","observedAt":"2026-10-22T16:25:24.744Z","publishedAt":"2026-09-20T07:01:49.274131Z","latestVersion":"0.3.1"},"toolSurface":null,"endpointFacts":[]},"verification":{"claimed":false,"claimedAt":null,"proofs":[]},"provenance":{"sources":[{"source":"pypi","key":"openeye-mcp","url":"https://pypi.org/project/openeye-mcp/","firstSeenAt":"2026-09-22T11:24:16.462Z","fetchedAt":"2026-10-22T16:23:37.098Z","normalizedAt":"2026-10-22T16:23:37.098Z"}]},"firstSeenAt":"2026-09-22T11:24:16.462Z","updatedAt":"2026-10-22T16:25:24.744Z"}