{"$schema":"https://wellknown.network/schemas/agent-record-v1.json","schemaVersion":"1","id":"ag_nxj2fgh8wtw7","handle":"vl-mcp-server","url":"https://wellknown.network/agents/vl-mcp-server","links":{"self":"https://wellknown.network/agents/vl-mcp-server/record.json","html":"https://wellknown.network/agents/vl-mcp-server","markdown":"https://wellknown.network/agents/vl-mcp-server/record.md","api":"https://wellknown.network/api/v1/agents/vl-mcp-server","status":"https://wellknown.network/api/v1/agents/vl-mcp-server/status","claim":"https://wellknown.network/agents/vl-mcp-server/claim","claimApi":"https://wellknown.network/api/v1/claims","claimDescriptor":"https://wellknown.network/agents/vl-mcp-server/claim.json","badge":"https://wellknown.network/agents/vl-mcp-server/badge.svg","openapi":"https://wellknown.network/openapi.json","history":"https://wellknown.network/api/v1/agents/vl-mcp-server/history","tools":"https://wellknown.network/api/v1/agents/vl-mcp-server/tools"},"ard":{"identifier":"urn:air::server:vl-mcp-server","type":"application/mcp-server-card+json"},"kind":"mcp_server","declared":{"name":"vl-mcp-server","summary":"VL 视觉理解 MCP 服务 - 基于 tokendance 的 Qwen2.5-VL 模型","description":"# VL MCP Server\n\n基于 MCP（Model Context Protocol）协议的视觉理解服务，通过 tokendance 平台调用 Qwen2.5-VL 模型，实现图片内容分析功能。\n\n## 功能特性\n\n- **自动识别图片来源**：无需指定类型，自动区分 URL / 本地文件 / Base64 / data URI\n- **兼容 Claude Code 粘贴图片**：在对话框直接粘贴截图即可分析，无需提供文件路径\n- 基于 Qwen2.5-VL-7B-Instruct 模型\n- 兼容 Cherry Studio、Claude Code、Cursor 等 MCP 客户端\n- 可托管部署到魔搭 MCP 广场\n\n## 安装使用\n\n### 环境要求\n\n- Python >= 3.10\n- uv（推荐）或 pip\n\n### 配置\n\n设置环境变量：\n\n```bash\nexport TOKENDANCE_API_KEY=\"your-api-key\"\n```\n\n### MCP 客户端配置\n\n**Cherry Studio / Claude Desktop / Cursor / Claude Code：**\n\n```json\n{\n  \"mcpServers\": {\n    \"vl-image-analyzer\": {\n      \"command\": \"uvx\",\n      \"args\": [\"vl-mcp-server\"],\n      \"env\": {\n        \"TOKENDANCE_API_KEY\": \"your-api-key\",\n        \"VL_MODEL\": \"qwen2.5-vl-7b-instruct\"\n      }\n    }\n  }\n}\n```\n\n> `VL_MODEL` 为可选项，默认使用 `qwen2.5-vl-7b-instruct`。  \n> 如需使用更大参数模型，可修改为 `qwen2.5-vl-72b-instruct`。\n\n## 开发\n\n### 本地开发\n\n```bash\n# 克隆仓库\ngit clone <repo-url>\ncd vl-mcp-server\n\n# 使用 uv 安装依赖\nuv sync\n\n# 运行服务\nuv run vl-mcp-server\n```\n\n### 调试\n\n使用 MCP Inspector：\n\n```bash\nnpx @modelcontextprotocol/inspector uv run vl-mcp-server\n```\n\n## 工具说明\n\n### analyze_image\n\n分析图片内容并返回描述。**图片来源自动识别，无需指定类型。**\n\n**参数：**\n\n| 参数 | 类型 | 必填 | 默认值 | 说明 |\n|-----|------|-----|-------|------|\n| `image` | string | ✅ | - | 图片来源，支持多种格式（见下方）|\n| `prompt` | string | ❌ | \"请详细描述这张图片的内容\" | 自定义分析提示词 |\n\n**`image` 参数支持的格式：**\n\n| 格式 | 示例 | 使用场景 |\n|-----|------|---------|\n| 本地文件路径 | `/Users/xxx/image.png` | 直接指定本地文件 |\n| HTTP/HTTPS URL | `https://example.com/img.jpg` | 网络图片 |\n| data URI | `data:image/png;base64,...` | **Claude Code 粘贴图片自动使用** |\n| 裸 Base64 | `iVBORw0KGgo...` | 第三方系统传入的 base64 |\n\n**支持的图片格式：** png, jpg, jpeg, gif, webp, bmp, tiff\n\n**示例：**\n\n**本地文件方式：**\n```json\n{\n  \"image\": \"/Users/username/Pictures/photo.jpg\",\n  \"prompt\": \"描述这张图片的主要内容\"\n}\n```\n\n**URL 方式：**\n```json\n{\n  \"image\": \"https://example.com/image.jpg\",\n  \"prompt\": \"描述这张图片的主要内容\"\n}\n```\n\n**Claude Code 粘贴图片方式（由 Claude 自动处理）：**\n- 在对话框粘贴图片后，直接说\"使用 analyze_image 工具分析这张图片\"\n- Claude 会自动将图片转为 `data:image/png;base64,...` 格…","publisher":null,"homepage":null,"repository":null,"version":"0.1.2","license":"MIT","protocols":["mcp"],"tags":["mcp","qwen","tokendance","vision","vl"],"pricing":null,"endpoints":[{"url":"pypi:vl-mcp-server","type":"package_pypi","auth":null,"probeable":false}],"skills":null,"tools":null,"extra":null,"attribution":{"kind":"pypi","name":"pypi","license":"pypi","summary":"pypi","version":"pypi","description":"pypi"}},"derived":{"capabilities":[{"slug":"media.image-understanding","name":"Image Understanding","confidence":1,"provenance":"declared"}],"categories":["media"],"language":"en"},"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":"vl-mcp-server","registry":"pypi","observedAt":"2026-10-22T23:26:19.121Z","publishedAt":"2026-04-14T09:28:17.574192Z","latestVersion":"0.1.2"},"toolSurface":null,"endpointFacts":[]},"verification":{"claimed":false,"claimedAt":null,"proofs":[]},"provenance":{"sources":[{"source":"pypi","key":"vl-mcp-server","url":"https://pypi.org/project/vl-mcp-server/","firstSeenAt":"2026-09-10T15:23:47.500Z","fetchedAt":"2026-10-22T23:24:48.320Z","normalizedAt":"2026-10-22T23:24:48.320Z"}]},"firstSeenAt":"2026-09-10T15:23:47.500Z","updatedAt":"2026-10-22T23:26:19.121Z"}