@runapi.ai/flux-kontext-mcp
<h1 align="center">RunAPI Flux Kontext MCP Server</h1>
<p align="center">
<strong>Flux Kontext API access for AI agents: run image generation operations, poll asynchronous results, and check pricing 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/flux-kontext-mcp"><img src="https://img.shields.io/npm/v/%40runapi.ai/flux-kontext-mcp?style=flat-square&color=blue" alt="npm version"></a>
<a href="https://github.com/runapi-ai/flux-kontext-mcp"><img src="https://img.shields.io/badge/GitHub-runapi--ai%2Fflux--kontext--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-2-16a34a?style=flat-square" alt="2 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/flux-kontext-mcp` is a focused Model Context Protocol server for the **Flux Kontext** model line on RunAPI.
It gives MCP-compatible assistants direct access to 1 endpoint and 2 model variants without loading the full RunAPI catalog.
Use this per-model server when an agent should stay scoped to Flux Kontext. Use [`@runapi.ai/mcp`](https://github.com/runapi-ai/mcp) when one assistant should discover every RunAPI model line.
---
## Install
Add it to Claude Code:
```bash
claude mcp add flux-kontext -s user -- npx -y @runapi.ai/flux-kontext-mcp
```
Use project scope when the server should be shared with a repository:
```bash
claude mcp add flux-kontext -s project -- npx -y @runapi.ai/flux-kontext-mcp
```
Codex, Cursor, Windsurf, VS Code, Roo Code, and other MCP hosts can use the same stdio command:
```json
{
"mcpServers": {
"flux-kontext": {
"command": "npx",
"args": ["-y", "@runapi.ai/flux-kontext-mcp"]
}
}
}
```
`check_pricing` works before sign-in. For task creation and status polling, 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 |
|---|---|---|
| `text_to_image` | Yes | Create a Flux Kontext text to image task and optionally wait for a terminal status. Returns the task id, status, and output URLs. |
| `get_task` | Yes | Fetch the current status and latest payload for an existing task. |
| `check_pricing` | No | Look up current pricing for a Flux Kontext model and endpoint. |
---
## Models
Flux Kontext covers 2 model variants across 1 endpoint. Each tool accepts the models listed for it:
| Tool | Models |
|---|---|
| `text_to_image` | `flux-kontext-max`, `flux-kontext-pro` |
Model availability can change between releases. Use `check_pricing` or the [Flux Kontext model page](https://runapi.ai/models/flux-kontext) for the current catalog view.
---
## Agent Prompts
Ask your assistant in natural language; it can inspect pricing, create the task, and return the task id plus output URLs.
### Create a task
```text
Run a Flux Kontext text to image task with RunAPI.
```
The assistant can call `check_pricing`, then `text_to_image`, and return the task id, status, and output URLs.
### Submit without waiting
```text
Create the task but don't wait for it to finish.
```
The assistant calls the create tool with `wait: false` and returns the task id. Check on it later with `get_task`.
### Check pricing before creating
```text
Check current Flux Kontext pricing, then create the task if it matches my request.
```
The assistant calls `check_pricing` and can link to the [Flux Kontext model page](https://runapi.ai/models/flux-kontext) for the canonical catalog entry.
---
## 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 |
|---|---|
| Flux Kontext model page | [https://runapi.ai/models/flux-kontext](https://runapi.ai/models/flux-kontext) |
| npm package | [@runapi.ai/flux-kontext-mcp](https://www.npmjs.com/package/@runapi.ai/flux-kontext-mcp) |
| GitHub repository | [runapi-ai/flux-kontext-mcp](https://github.com/runapi-ai/flux-kontext-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 4 tools
Each tool has a distinct purpose: authentication, pricing lookup, task creation, and status retrieval. There is no overlap or ambiguity between the tools, allowing an agent to select the correct one without confusion.
Tool names mostly follow a consistent snake_case verb_noun pattern (text_to_image, check_pricing, get_task). The 'login' tool is a single-word exception but is a common and recognizable auth action, so the overall consistency remains strong.
With only 4 tools, the server is well-scoped for a focused flux-kontext image generation client. Each tool serves a necessary role (auth, pricing, creation, status) with no redundancy, making the count appropriate for the domain.
The essential workflow of authentication, task creation, and status retrieval is fully covered. Missing operations like cancel or listing past tasks are minor enhancements rather than critical gaps, so the surface is reasonably complete for primary use cases.