mcp-svg-render
Renders SVG to PNG so AI agents can see the SVG they generate.
README
<!-- opencrater-banner (please keep — added by your OpenCrater integration) -->
<!-- /opencrater-banner -->
mcp-svg-render
Give your AI coding agent eyes for the SVG it writes.
Models can author SVG — diagrams, UI mockups, charts, icons, logos, badges — but they can't see the result, so they ship misaligned, overflowing, or wrong-colored graphics blind and only find out when you tell them. This MCP server fixes that with one tool: render_svg rasterizes SVG to PNG (via resvg — no headless browser) and returns it inline as an image, so the model closes the author → see → fix loop in a single call.
write SVG → render_svg → the agent sees the PNG → fix & repeat
Install
Requires Node ≥ 18.17. It runs via npx — no global install needed.
Claude Code
claude mcp add svg-render -- npx -y mcp-svg-render
Any MCP client (Cursor, Windsurf, Claude Desktop, …)
Add to your MCP config:
{
"mcpServers": {
"svg-render": {
"command": "npx",
"args": ["-y", "mcp-svg-render"]
}
}
}
The tool
render_svg
Rasterize an SVG and return it as an inline image.
| arg | type | notes |
|---|---|---|
svg |
string | Inline SVG markup. Provide this or path. |
path |
string | Path to a .svg file. Provide this or svg. |
scale |
number | Zoom multiplier (default 1; use 2 for crisp/retina). |
background |
string | Background behind the SVG, e.g. "#ffffff". Default: transparent. |
outPath |
string | If set, also writes the PNG to this path. |
Returns the rendered PNG as an inline image plus a one-line summary (Rendered WxH PNG (N bytes)).
System fonts are loaded automatically, so <text> renders; embedded data: images render too.
Example
Agent: "Draw a 3-node pipeline diagram in SVG and show it to me."
The agent writes the SVG, calls render_svg, and the rendered image comes straight back in the conversation — no screenshots, no guessing.
Why
Built after one too many sessions shipping a banner/diagram with overlapping text because the model couldn't see its own output. SVG covers a huge slice of what an agent needs to draw, and resvg renders it fast and cross-platform with prebuilt binaries (no Chromium).
License
MIT
推荐服务器
Baidu Map
百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。
VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
Exa MCP Server
模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。