decor-cli
Enables decoration of images and videos with backgrounds, gradients, containers, text, arrows, and more through MCP tools like render_decor and list_templates. Supports both CLI and MCP server modes with configurable options.
README
decor-cli

decor-cli decorates images and videos from a CLI or MCP server. It adds backgrounds, gradients, rounded containers, shadows, padding, crop, alignment, text, arrows, shapes, counters, spotlight, and template presets without distorting the original media ratio.
How It Works

Features

- Backgrounds: solid colors, gradients, specific image files, random folders, and hosted default backgrounds.
- Container styling: padding, rounded corners, nine-point alignment, opacity, blur, and drop shadow controls.
- Input sources: local files, web URLs, Base64 payloads, images, and videos.
- Crop and composition: crop rectangles, preserve original ratio, and export without distortion.
- Text: font family, size, color, coordinates, wrap width, outline, shadow, and rounded text boxes.
- Annotations: curved arrows, rectangles, circles, lines, transparent fills, and numbered counters.
- Spotlight: brighten a selected region while dimming surrounding context.
- Templates: reusable named presets for common screenshot and visual styles.
- Agent-ready surfaces: CLI, MCP server, and companion agent skill.
Install
npm install -g decor-cli
decor install-backgrounds
Video rendering uses system ffmpeg and ffprobe. Install them with your OS package manager or set DECOR_FFMPEG_PATH and DECOR_FFPROBE_PATH.
Quick Start
decor render --input screenshot.png --output output.png --template clean-gradient --text "Release notes" --overwrite
decor render --input screenshot.png --output output.png --background-folder ~/.decor-cli/backgrounds --padding 104 --radius 38 --overwrite
decor render --input demo.mp4 --output demo-decorated.mp4 --padding 96 --radius 36 --overwrite
decor install-backgrounds --dir ./backgrounds
decor list-templates
decor doctor --json
URL and Base64 inputs are supported:
decor render --input-url https://example.com/image.png --output output.png
decor render --input-base64 "$DATA_URI" --output output.png
Private, loopback, link-local, and metadata URL targets are blocked by default. Use --allow-private-network only for trusted local fixtures.
Background Gallery
Run decor install-backgrounds on a fresh machine to download this hosted background pack into ~/.decor-cli/backgrounds.

Config
Most advanced features are configured through JSON:
{
"template": "clean-gradient",
"input": { "type": "path", "path": "screenshot.png" },
"output": { "path": "output.png", "quality": 92, "overwrite": true },
"container": { "padding": 96, "radius": 36, "alignment": "center" },
"annotations": [
{ "type": "text", "text": "Step 1", "x": 80, "y": 90, "fontSize": 44, "shadow": true },
{ "type": "counter", "value": 1, "x": 110, "y": 160, "size": 42 },
{ "type": "arrow", "from": { "x": 120, "y": 180 }, "to": { "x": 260, "y": 220 } }
]
}
Run it:
decor render --config decor.config.json --output output.png
MCP
decor-mcp --transport stdio
DECOR_MCP_TOKEN=example-token decor-mcp --transport http --port 8080
MCP tools: render_decor, preview_decor, validate_decor, list_templates, doctor, and config_resolve.
Release
Conventional commits drive releases with semantic-release:
main-> stable GitHub release and npmlatestdev-> beta prerelease and npmbeta
GitHub releases also include runtime archives named like decor-cli-vX.Y.Z-linux-amd64.tar.gz for installers that resolve assets by OS and architecture.
Publish jobs are separated from PR CI and should be protected with the release environment. Set the repository NPM_TOKEN secret before the first live publish.
Default Backgrounds
decor install-backgrounds downloads the hosted default background pack from Cloudflare R2 into ~/.decor-cli/backgrounds. The installer verifies every file by byte length and SHA-256 before replacing local files. Use --dir <path> to install elsewhere and --force to redownload files that already match the manifest.
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。