# Posecode MCP server

> Allows an AI client to author, validate, and render 3D human movement using a text-based kinematic language.

- Repository: https://github.com/posecode-dev/posecode
- Page: https://hostedmcp.org/servers/posecode-dev/posecode
- Transport: stdio
- Stars: 131
- License: Apache-2.0
- Latest version: 0.5.0

## Install

### Claude Code

```sh
claude mcp add posecode -- npx -y posecode-mcp
```

### Cursor (`~/.cursor/mcp.json`)

```json
{
  "mcpServers": {
    "posecode": {
      "command": "npx",
      "args": [
        "-y",
        "posecode-mcp"
      ]
    }
  }
}
```

### VS Code (`.vscode/mcp.json`)

```json
{
  "servers": {
    "posecode": {
      "type": "stdio",
      "command": "npx",
      "args": [
        "-y",
        "posecode-mcp"
      ]
    }
  }
}
```

### VS Code CLI

```sh
code --add-mcp '{"name":"posecode","type":"stdio","command":"npx","args":["-y","posecode-mcp"]}'
```

### Claude Desktop (`claude_desktop_config.json`)

```json
{
  "mcpServers": {
    "posecode": {
      "command": "npx",
      "args": [
        "-y",
        "posecode-mcp"
      ]
    }
  }
}
```

### Codex

```sh
codex mcp add posecode -- npx -y posecode-mcp
```

### Codex config.toml (`~/.codex/config.toml`)

```toml
[mcp_servers.posecode]
command = "npx"
args = ["-y", "posecode-mcp"]
```

### Gemini CLI

```sh
gemini mcp add -s user posecode npx -- -y posecode-mcp
```

### Gemini settings.json (`~/.gemini/settings.json`)

```json
{
  "mcpServers": {
    "posecode": {
      "command": "npx",
      "args": [
        "-y",
        "posecode-mcp"
      ]
    }
  }
}
```

## Environment variables

| Name | Required | Secret | Description |
|---|---|---|---|
| `POSECODE_BASE_URL` | no | no | Playground origin used for render permalinks (defaults to https://posecode.org). |

## FAQ

### What is Posecode?

Posecode is an inspectable, editable movement format that uses small text documents to describe, validate, and render deterministic human motion.

### How does Posecode differ from diffusion text-to-motion?

Unlike resource-intensive neural systems, Posecode is lightweight, programmable, and provides editable semantic instructions rather than black-box trajectories.

### Can an AI agent use this server?

Yes, the MCP server is designed to be agent-friendly by exposing tools for generation, validation, critique, and sharing.