ImageKit Plugin

ImageKit Skills

Reusable AI agent skills for ImageKit.io — install them with the skills CLI to enhance your coding agent's capabilities.

Skills

Skill	Description
mcp-preflight	Mandatory routing guide — tells the agent which MCP server to call for what, before every ImageKit tool invocation
search-docs	Search ImageKit documentation with optimized queries and source selection
transformation-builder	Build ImageKit image/video transformations — AI editing, background removal, resize, crop, overlays, and more
upload-files	Upload files to ImageKit media library with folder paths, tags, and metadata
ai-tasks	Apply AI-powered analysis to images for business-specific tagging, metadata extraction, and quality checks using controlled vocabularies

Installation

You can install skills and MCP servers using either method:

• Plugin Method (Recommended): Use the ImageKit plugin for your platform for guided, one-click setup of skills and MCP servers
• Manual Method: Install skills via CLI and add MCP servers through configuration files

Choose the method that works best for your workflow. Below, each platform shows both options.

Claude

Plugin Method (Recommended)

Follow these steps to install the ImageKit plugin in Claude Code:

1. Open Plugin Settings — Click on Customize in the left sidebar

2. Add Marketplace — Click the "+" button and select Create Plugin → Add Marketplace

3. Enter Plugin URL — Add imagekit-developer/skills in the marketplace URL field

4. Install Plugin — Find and install the ImageKit Skills plugin

5. Install MCP Servers — Click on Connectors in the installed plugin and install the MCP servers (imagekit_devtools and imagekit_api)

6. Complete Authentication — Complete authentication for the imagekit_api server when prompted

Once complete, all ImageKit skills and MCP servers are ready to use in Claude Code.

Manual Method

Install Skills

npx skills add imagekit-developer/skills --all

Run the following command in your terminal:

claude mcp add imagekit_devtools --transport http https://devtools-mcp.imagekit.io/mcp
claude mcp add imagekit_api --transport http https://api-mcp.imagekit.in/mcp

Or edit your Claude Desktop configuration file:

• macOS: ~/Library/Application Support/Claude/claude_desktop_config.json
• Windows: %APPDATA%\Claude\claude_desktop_config.json

{
  "mcpServers": {
    "imagekit_devtools": {
      "command": "npx",
      "args": ["-y", "mcp-remote@latest", "https://devtools-mcp.imagekit.io/mcp"]
    },
    "imagekit_api": {
      "command": "npx",
      "args": ["-y", "mcp-remote@latest", "https://api-mcp.imagekit.in/mcp"]
    }
  }
}

Note: Restart Claude Desktop / ClaudeCode after making changes for the MCP servers to take effect.

Codex

Install Skills

npx skills add imagekit-developer/skills --all

Add MCP servers via CLI:

codex mcp add imagekit_devtools --url https://devtools-mcp.imagekit.io/mcp
codex mcp add imagekit_api --url https://api-mcp.imagekit.in/mcp

Or edit ~/.codex/config.toml directly:

[mcp_servers.imagekit_devtools]
url = "https://devtools-mcp.imagekit.io/mcp"

[mcp_servers.imagekit_api]
url = "https://api-mcp.imagekit.in/mcp"

Note: Restart Codex after adding MCP servers for them to take effect.

VS Code Copilot

Plugin Method (Recommended)

Follow these steps to install the ImageKit plugin in VS Code:

1. Open Command Palette — Press ⇧⌘P and run Install Plugin from Source

2. Add Plugin Repository — Enter imagekit-developer/skills in the plugin source field

3. Complete Installation — Continue with the installation process as prompted

4. Restart VS Code — Restart VS Code for all skills and MCP servers to take effect

Once complete, all ImageKit skills and MCP servers are ready to use in VS Code.

Manual Method

Install Skills

npx skills add imagekit-developer/skills --all

Install MCP servers via the command line:

code --add-mcp "{\"name\":\"imagekit_devtools\",\"type\":\"http\",\"url\":\"https://devtools-mcp.imagekit.io/mcp\"}"
code --add-mcp "{\"name\":\"imagekit_api\",\"type\":\"http\",\"url\":\"https://api-mcp.imagekit.in/mcp\"}"

Or install via the VS Code UI: