Your prompts, tools, and agent configurations are scattered across files, Notion pages, and your memory. ResourceX gives you:
- One format to package any AI resource
- Version control like npm packages
- MCP integration so Claude can directly use your resources
Think of it as npm for AI resources.
Claude Desktop (~/.claude/claude_desktop_config.json):
{
"mcpServers": {
"resourcex": {
"command": "npx",
"args": ["@resourcexjs/mcp-server"]
}
}
}VS Code (Claude extension settings):
{
"claude.mcpServers": {
"resourcex": {
"command": "npx",
"args": ["@resourcexjs/mcp-server"]
}
}
}You: Search for code review resources
Claude: [calls search("code review")] Found the following resources...
You: Use code-review:1.0.0 to review this code
Claude: [calls use("code-review:1.0.0")] Let me review your code with this prompt...
That's it. Claude can now access and use ResourceX resources.
Create Your Own Resource
# Create resource directory
mkdir my-prompt && cd my-prompt
# Create metadata
cat > resource.json << 'EOF'
{
"name": "my-prompt",
"type": "text",
"version": "1.0.0"
}
EOF
# Create content
echo "You are a helpful assistant specialized in..." > content
# Add to local storage
npx @resourcexjs/cli add .Manage Resources with CLI
# Install
npm install -g @resourcexjs/cli
# Common commands
rx add ./my-prompt # Add local resource
rx list # List all resources
rx use name:1.0.0 # Use a resource
rx search keyword # Search resources
rx push name:1.0.0 # Publish to registry
rx pull name:1.0.0 # Pull from registryUse SDK in Code
npm install resourcexjsimport { createResourceX } from "resourcexjs";
const rx = createResourceX();
// Add resource
await rx.add("./my-prompt");
// Use resource
const result = await rx.use("my-prompt:1.0.0");
console.log(result.content);
// Search resources
const results = await rx.search("code review");Self-host Registry
import { createRegistryServer } from "@resourcexjs/server";
const app = createRegistryServer({
storagePath: "./data",
});
export default app; // Deploy to any platform that supports HonoTutorials - Step-by-step lessons
- Quick Start - 2 minutes to get started
- Create Your First Resource - Create, test, and use locally
- Publish Your Resource - Share via registry
Guides - How-to for specific tasks
- MCP Integration - Configure for AI agents
- Development Workflow - Rapid iteration with link
- Versioning - Manage resource versions
- Self-Hosting Registry - Deploy your own registry
Reference - Technical reference
- CLI Reference -
rxcommand reference - SDK Reference -
resourcexjsAPI reference - MCP Tools Reference - MCP tools reference
Concepts - Background and explanations
- Locators - Resource identifiers
- Resource Types - Built-in and custom types
- Registry - How registries work
| Package | Description |
|---|---|
@resourcexjs/cli |
rx command-line tool |
@resourcexjs/mcp-server |
MCP Server for AI Agents |
resourcexjs |
SDK |
@resourcexjs/server |
Self-hosted Registry Server |
Internal Packages (for developers)
| Package | Description |
|---|---|
@resourcexjs/core |
Core primitives (RXL, RXM, RXA, RXR) |
@resourcexjs/storage |
Storage backends |
@resourcexjs/registry |
Registry implementations |
@resourcexjs/loader |
Resource loading |
@resourcexjs/type |
Type system |
@resourcexjs/arp |
Low-level I/O protocol |
Part of the Deepractice AI infrastructure:
- AgentX - AI Agent framework
- AgentVM - AI Agent runtime
- ResourceX - Resource management (this project)