@runapi.ai/luma-mcp
<h1 align="center">RunAPI Luma MCP Server</h1>
<p align="center">
<strong>Luma historical task access for AI agents: query existing tasks through one focused MCP server.</strong>
</p>
<p align="center">
<sub>Works with Claude Code, Codex, Cursor, Windsurf, VS Code, Roo Code, and any MCP-compatible host.</sub>
</p>
<p align="center">
<a href="https://www.npmjs.com/package/@runapi.ai/luma-mcp"><img src="https://img.shields.io/npm/v/%40runapi.ai/luma-mcp?style=flat-square&color=blue" alt="npm version"></a>
<a href="https://github.com/runapi-ai/luma-mcp"><img src="https://img.shields.io/badge/GitHub-runapi--ai%2Fluma--mcp-24292f?style=flat-square" alt="GitHub repository"></a>
<a href="LICENSE"><img src="https://img.shields.io/badge/License-Apache_2.0-blue?style=flat-square" alt="Apache-2.0 license"></a>
<img src="https://img.shields.io/badge/Type-MCP_Server-blue?style=flat-square" alt="MCP Server">
<img src="https://img.shields.io/badge/Models-0-16a34a?style=flat-square" alt="0 models">
</p>
<p align="center">
<a href="#install">Install</a> |
<a href="#tools">Tools</a> |
<a href="#models">Models</a> |
<a href="#agent-prompts">Agent Prompts</a> |
<a href="#configuration">Configuration</a> |
<a href="#links">Links</a>
</p>
---
## Why This Package?
`@runapi.ai/luma-mcp` is a focused Model Context Protocol server for the **Luma** model line on RunAPI.
It gives MCP-compatible assistants direct access to 0 endpoints and 0 model variants without loading the full RunAPI catalog.
Use this per-model server when an agent should stay scoped to Luma. Use [`@runapi.ai/mcp`](https://github.com/runapi-ai/mcp) when one assistant should discover every RunAPI model line.
`luma-modify-video` was retired on 2026-10-14T00:00:00Z. New requests return HTTP 410 before task creation or balance deduction. Existing tasks can still be queried. This model ID is not mapped to another model.
---
## Install
Add it to Claude Code:
```bash
claude mcp add luma -s user -- npx -y @runapi.ai/luma-mcp
```
Use project scope when the server should be shared with a repository:
```bash
claude mcp add luma -s project -- npx -y @runapi.ai/luma-mcp
```
Codex, Cursor, Windsurf, VS Code, Roo Code, and other MCP hosts can use the same stdio command:
```json
{
"mcpServers": {
"luma": {
"command": "npx",
"args": ["-y", "@runapi.ai/luma-mcp"]
}
}
}
```
No models in this package accept new tasks. `check_pricing` returns no supported model. For historical task lookup, ask your assistant to call the `login` tool. It opens a browser login and saves credentials to `~/.config/runapi/config.json`, the same file used by `runapi login`.
Headless and CI hosts can still set `RUNAPI_API_KEY` before starting the MCP host.
Ready-made examples are in [`examples/`](examples/) for Claude, Cursor, Windsurf, VS Code, and Roo Code.
---
## Tools
| Tool | Auth | Purpose |
|---|---|---|
| `get_task` | Yes | Fetch the current status and latest payload for an existing task. |
| `check_pricing` | No | Returns no supported model because this model line is retired. |
---
## Models
Luma covers 0 model variants across 0 endpoints. Each tool accepts the models listed for it:
| Tool | Models |
|---|---|
Model availability can change between releases. Use `check_pricing` or the [Luma model page](https://runapi.ai/models/luma) for the current catalog view.
---
## Agent Prompts
Query an existing task with `get_task`. New requests for this retired model return HTTP 410.
```text
Check the status of my existing Luma task with RunAPI.
```
The assistant calls `get_task` with the task id and returns its status and available result.
---
## Configuration
The server resolves auth in this order:
1. `RUNAPI_API_KEY` environment variable, useful for headless and CI hosts
2. `~/.config/runapi/config.json`, created by the MCP `login` tool or `runapi login`
3. No key, which still allows `check_pricing`
The config file is normally managed by login. A pre-provisioned headless config can use:
```json
{
"apiKey": "your_runapi_key"
}
```
Do not commit real API keys.
---
## Links
| Resource | URL |
|---|---|
| Luma model page | [https://runapi.ai/models/luma](https://runapi.ai/models/luma) |
| npm package | [@runapi.ai/luma-mcp](https://www.npmjs.com/package/@runapi.ai/luma-mcp) |
| GitHub repository | [runapi-ai/luma-mcp](https://github.com/runapi-ai/luma-mcp) |
| RunAPI MCP overview | [runapi.ai/mcp](https://runapi.ai/mcp) |
| RunAPI docs | [runapi.ai/docs](https://runapi.ai/docs) |
---
## License
Licensed under the [Apache License, Version 2.0](LICENSE).
TDQS
Scored across 3 tools
The three tools have clearly distinct purposes: pricing lookup, task status retrieval, and authentication. There is no overlap or realistic chance of an agent confusing one for another.
check_pricing and get_task follow a consistent verb_noun snake_case pattern, and login is also snake_case but is a bare verb with no noun. The minor deviation is conventional and still readable.
Three tools is borderline thin for a model-serving MCP; one slot is consumed by login and another by pricing, leaving only a single task-status tool. A minimal client could work with this, but it feels under-scoped.
The surface has a major dead end: get_task retrieves results but no tool submits or creates a Luma task, so the core generate/run workflow cannot be initiated. Pricing and login are covered, but the primary lifecycle operation is missing.