DeepSeek Visionary
An MCP server that integrates DeepSeek's web-native multimodal vision model into Zed, featuring automatic browser login and image analysis tools. It enables vision-powered tasks like image understanding and multi-turn conversations directly within the editor.
README
DeepSeek Visionary
在任意支持 MCP 的 AI agent(Zed、OpenCode、Codex、Claude Code、Cursor、Claude Desktop)中使用 DeepSeek 网页版的原生多模态视觉模型,支持浏览器自动登录(无需手动复制 token)。
这是 Python 版 deepseek-vision-mcp 的 Rust 全量重写:单原生二进制,多平台分发。
架构
graph TD
subgraph 宿主[任意 MCP 宿主]
AG["Zed / OpenCode / Codex / Claude Code / Cursor / Claude Desktop"]
AG -->|spawn 独立进程| SRV
end
subgraph visionary-server 原生二进制
SRV["MCP stdio 服务<br/>vision / status / login / logout 工具<br/>init / doctor CLI"]
CFG["~/.deepseek-visionary/config.json<br/>token + smidV2 + cf_clearance + 会话"]
SRV --> CFG
end
SRV -->|HTTPS| DS["DeepSeek 网页后端"]
SRV -->|CDP 启动 + 监听| BRO["Chrome 系浏览器<br/>仅登录时出现"]
- visionary-server:标准 MCP stdio 服务,实现完整 vision 流水线(PoW → 上传 → fork → HIF 签名 → SSE 流式 completion)与 CDP 自动登录;同时提供
init/doctorCLI 引导接入 - visionary-zed-ext:Zed 扩展壳(仅 Zed 需要),按平台从 GitHub Releases 下载/缓存 visionary-server 并启动
安装
1. 安装二进制
# 一键脚本(macOS / Linux)
curl -LsSf https://github.com/xlight/deepseek-visionary/releases/latest/download/visionary-server-installer.sh | sh
# 或 Homebrew(发布后可用)
brew install <tap>/visionary-server
# 或 npm
npm install -g <npm-package>
也可以直接从 GitHub Releases 下载对应平台的
visionary-server-<target-triple>裸二进制加入 PATH。 Windows 用户可使用 PowerShell 安装脚本(发布后可用)。
2. 接入你的 AI agent
# 一键检测并接入(列出已安装 agent)
visionary-server init
# 接入指定 agent
visionary-server init opencode
visionary-server init codex
visionary-server init claude
visionary-server init cursor
visionary-server init claude-desktop
# 批量接入多个 agent(免交互)
visionary-server init --opencode --codex --yes
# 先预览将写入的配置(不落盘)
visionary-server init opencode --dry-run
各 agent 的详细接入文档见 docs/integrations/:
| Agent | 文档 | 一键命令 |
|---|---|---|
| Zed | zed.md | 扩展市场安装(见下) |
| OpenCode | opencode.md | visionary-server init opencode |
| Codex | codex.md | visionary-server init codex |
| Claude Code | claude-code.md | visionary-server init claude |
| Cursor | cursor.md | visionary-server init cursor |
| Claude Desktop | claude-desktop.md | visionary-server init claude-desktop |
新兴通道:Microsoft Agent Package Manager 用户可直接
apm install --mcp io.github.xlight/deepseek-visionary(复用 MCP Registry 标识)。
3. 登录
调用 MCP 工具的 deepseek_vision_login 完成自动登录:
- 会打开浏览器窗口并导航到 chat.deepseek.com
- 在浏览器中登录后,工具自动抓取 token 并保存
手动兜底:登录 chat.deepseek.com 后,DevTools → Application → Local Storage →
userToken→ 复制JSON.parse(value).value,写入~/.deepseek-visionary/config.json:{ "user_token": "你的 token" }
4. 使用
调用 deepseek_vision 传入图片路径或 base64 即可识图。
Zed 扩展安装
如果你只用 Zed,也可以直接从扩展市场安装:
- Zed 命令面板(
Cmd+Shift+P)→zed: extensions→ 搜索 DeepSeek Visionary → Install - 扩展壳自动下载/缓存
visionary-server二进制并启动 MCP 服务 - 授权工具权限(见 docs/integrations/zed.md)
CLI 工具
| 命令 | 说明 |
|---|---|
visionary-server(无参数) |
进入 MCP stdio serve 模式(所有 agent 配置的默认入口) |
visionary-server --version |
输出版本号 |
visionary-server doctor |
诊断环境:config 路径/权限、浏览器、token 有效性、平台 |
visionary-server init [agent] |
检测并接入已安装的 AI agent(--dry-run / --yes / 多选 flags) |
MCP 工具
| 工具 | 说明 |
|---|---|
deepseek_vision |
上传本地图片(路径 / base64)并用 DeepSeek 视觉模型分析。参数:image(必填)、prompt、thinking、continue_conversation、session_id |
deepseek_vision_status |
检查登录状态与 token 有效性(含真实校验探针) |
deepseek_vision_login |
浏览器自动登录并抓取凭据 |
deepseek_vision_logout |
清除保存的凭据 |
会话续聊
deepseek_vision 支持多轮对话:
continue_conversation=true:复用上一次会话,可对比多张图片session_id:显式切换到指定会话线程
会话状态持久化在 ~/.deepseek-visionary/session.json。
环境变量
| 变量 | 说明 |
|---|---|
DEEPSEEK_USER_TOKEN |
覆盖 config.json 中的 token(可选) |
DEEPSEEK_SMIDV2 / DEEPSEEK_CF_CLEARANCE |
覆盖对应 cookie(可选) |
DEEPSEEK_BASE_URL |
API 基地址(默认 https://chat.deepseek.com) |
DEEPSEEK_LOGIN_TIMEOUT |
登录等待超时秒数(默认 600) |
开发
# 构建原生服务
cargo build -p visionary-server --release
# 构建扩展壳(wasm32-wasip2)
rustup target add wasm32-wasip2
cargo build -p visionary-zed-ext --release --target wasm32-wasip2
# 测试
cargo test -p visionary-server
平台支持
- macOS(Apple Silicon / Intel)
- Linux(x86_64 / aarch64)
- Windows(x86_64)
需要 Chrome / Chromium / Edge 之一用于自动登录。
工作原理(要点)
- PoW:wasmtime 加载 DeepSeek 站内
sha3_wasm_bg.*.wasm(随仓库分发),调用wasm_solve求解upload_file与completion的 challenge - TLS 指纹:completion 端点与 Python 版(curl_cffi chrome131)对齐;Rust 侧默认普通 reqwest,若被 403 再启用指纹模拟(见 design.md spike 记录)
- 登录:CDP 控制 Chrome 系浏览器(专用 profile
~/.deepseek-visionary/browser/),读取localStorage.userToken与smidV2/cf_clearancecookie - 凭据安全:
~/.deepseek-visionary/config.json权限 0600,浏览器 profile 0700
License
MIT
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.