getting-started.md•1.58 kB
# Getting Started
This guide covers the minimal setup to run the Claude Skills MCP server in its current v1.0.0 two-package architecture.
## Quick Start (Cursor)
Add to your Cursor MCP settings (`~/.cursor/mcp.json`):
```json
{
"mcpServers": {
"claude-skills": {
"command": "uvx",
"args": ["claude-skills-mcp"]
}
}
}
```
Restart Cursor. The lightweight frontend starts instantly and displays tools. On the first tool use, the backend downloads in the background. Subsequent uses are instant.
## Quick Start (CLI)
Run the server locally with defaults:
```bash
uvx claude-skills-mcp
```
## Configuration (Current Version)
If you need a config file with the current defaults:
```bash
uvx claude-skills-mcp --example-config > config.json
```
Run with your config:
```bash
uvx claude-skills-mcp --config config.json
```
Notes (v1.0.0):
- Frontend (`claude-skills-mcp`) is a lightweight MCP stdio server and proxy.
- Backend (`claude-skills-mcp-backend`) provides vector search over skills and runs over HTTP.
- On first use, the backend is auto-installed and started in the background.
## Troubleshooting (Current)
- First tool call shows "Loading backend...": expected on first run while backend downloads.
- Ensure Python 3.12 is available and `uv` manages the environment.
- For verbose logs:
```bash
uvx claude-skills-mcp --verbose
```
## Next Steps
- Architecture details: see `architecture.md`.
- API reference (tools and parameters): see `api.md`.
- Usage patterns and examples: see `usage.md`.
- Testing instructions: see `testing.md`.
