> Documentation index: https://packinspo.com/llms.txt · Full documentation: https://packinspo.com/llms-full.txt

# Getting started

> Connect Packinspo MCP to Claude, Codex, Cursor, VS Code, Gemini CLI or ChatGPT.

## Give your AI packaging taste

Models are good at code and copy, but packaging needs references: real formats, closures, label hierarchies, palettes and finishes that work on shelf. Packinspo MCP gives agents that research layer, grounded in shipped products rather than generic defaults.

## What you need

- **Nothing to sign up for.** The server is free, read-only and needs no API key.
- **A tool that supports remote MCP servers** over Streamable HTTP: Claude Code, Claude (web and desktop), Codex, Cursor, VS Code, Gemini CLI, ChatGPT (developer mode) and others.

## Server URL

```text
https://packinspo.com/mcp
```

## Connect your tool

**Claude Code: plugin with MCP and skill**

```text
/plugin marketplace add packinspo/packinspo-skill
/plugin install packinspo@packinspo
```

**Claude Code: MCP only**

```bash
claude mcp add --transport http packinspo https://packinspo.com/mcp
```

**Claude (web and desktop)**: Settings → Connectors → Add custom connector, then paste the server URL.

**Codex**: add to `~/.codex/config.toml`

```toml
[mcp_servers.packinspo]
url = "https://packinspo.com/mcp"
```

**Cursor**: add to `.cursor/mcp.json`

```json
{ "mcpServers": { "packinspo": { "url": "https://packinspo.com/mcp" } } }
```

**VS Code**: add to `.vscode/mcp.json`

```json
{ "servers": { "packinspo": { "type": "http", "url": "https://packinspo.com/mcp" } } }
```

**Gemini CLI**

```bash
gemini extensions install https://github.com/packinspo/packinspo-skill
```

**ChatGPT**: in developer mode, add a connector with the server URL (available on plans that allow custom connectors).

**stdio-only clients**

```bash
npx -y mcp-remote https://packinspo.com/mcp
```

## Check that it works

Ask your agent:

```text
List the Packinspo packaging categories.
```

It should call `list_taxonomy` and return industries, categories, formats and materials.

## Limits

The server allows fair use per connection (about 120 requests a minute). Responses are Markdown with absolute URLs and photo credits.