Skip to main content
Glama

GrokMCP

PyPI version Python License: MIT

A Model Context Protocol (MCP) server for Grok (xAI) — chat/reasoning/vision and Grok Imagine video generation, powered by the AceDataCloud API.

Chat with Grok models, or generate short AI videos from a text prompt or a still image — directly from any MCP-compatible client (Claude Desktop, Claude Code, Cursor, etc.).

Features

  • Chat / Reasoning / Vision — Talk to Grok 4.5 / Grok 4 / Grok 3 models, with image input and tool calling

  • Text to Video — Generate a video clip from a text description

  • Image to Video — Animate a reference image into a video

  • Async task tracking — Submit a job, poll for the result, single or batch

  • stdio & HTTP transports — Local stdio for desktop clients, HTTP for remote hosting

Related MCP server: xbridge-mcp

Tools

Tool

Description

grok_chat_completions

Chat completion (reasoning / vision / tool calling) with Grok chat models.

grok_text_to_video

Generate a video from a text prompt (any model except grok-imagine-video-1.5:official).

grok_image_to_video

Generate a video from an input image (+ optional motion prompt).

grok_get_task

Query the status/result of a single generation task.

grok_get_tasks_batch

Query the status/result of multiple tasks at once.

grok_list_models

List available models and their capabilities.

grok_list_actions

List all tools and example workflows.

grok_get_prompt_guide

Tips for writing effective video prompts.

Models

Chat (grok_chat_completions)

Model

Notes

grok-4.5

Default — latest flagship reasoning model

grok-4

Previous flagship reasoning model

grok-3

Earlier-generation model

Video

Model

Text→Video

Image→Video

Notes

grok-imagine-video-1.5-fast:reverse

✅

✅

Default. Fastest & cheapest. 6-30s, duration-banded billing.

grok-imagine-video:reverse

✅

✅

Standard. 1-15s, billed per output second.

grok-imagine-video:official

✅

✅

Official endpoint, higher fidelity. 1-15s, per second.

grok-imagine-video-1.5:official

❌

✅

Official image-to-video only (requires image_url). Up to 1080p, per second.

Parameters

Parameter

Applies to

Values

prompt

both

Text description (required for text-to-video)

image_url

image-to-video

Input image URL (required for -1.5-preview)

reference_image_urls

image-to-video

Optional list of style/content reference images

aspect_ratio

both

1:1, 16:9 (default), 9:16, 4:3, 3:4, 3:2, 2:3

resolution

both

480p (default), 720p, 1080p

duration

both

grok-imagine-video-1.5-fast:reverse: 6–30s; other models: 1–15s (default 6)

callback_url

both

Optional async webhook

Connect: hosted OAuth, API token, or local stdio

The hosted endpoint is https://grok.mcp.acedata.cloud/mcp. Choose one route for the MCP client:

Route

When to use it

Credential setup

Hosted OAuth

The client supports remote MCP OAuth

Add only the URL, then sign in to AceDataCloud and approve access. No token needs to be pasted into client configuration.

Hosted API token

The client cannot finish OAuth, or you need an explicit integration credential

Send an AceDataCloud API token in the Authorization: Bearer … header. Keep it in a local secret store or environment variable.

Local stdio

The client runs a local MCP process

Install mcp-grok and pass ACEDATACLOUD_API_TOKEN to that process. It still calls the AceDataCloud API.

The hosted service advertises OAuth metadata and Dynamic Client Registration (DCR). DCR registers the client application; it is not an API key. OAuth signs you in and the client sends the resulting Bearer token; it may reuse or create an API credential for the account. Browser sign-in still requires an AceDataCloud account. The hosted service can be metered: review current service documentation and displayed pricing before a real operation. Do not configure both an OAuth login and a fixed Authorization header for the same server.

Hosted OAuth examples

  • Claude and Claude Desktop chat: Add a remote custom connector in Customize → Connectors → Add custom connector, enter https://grok.mcp.acedata.cloud/mcp, select sign-in, and choose Register automatically if Claude asks how to register its OAuth client. Complete consent. Claude Desktop's local claude_desktop_config.json is a separate setup. Claude connector guide.

  • Claude Code: claude mcp add --transport http --scope user grok https://grok.mcp.acedata.cloud/mcp, then claude mcp login grok. Check /mcp. Claude Code MCP guide.

  • Cursor: Add a remote server with only https://grok.mcp.acedata.cloud/mcp. For a project, merge the entry below into <project>/.cursor/mcp.json; for personal use, use ~/.cursor/mcp.json. Cursor MCP guide.

  • VS Code / Copilot: Run MCP: Add Server, select HTTP, enter https://grok.mcp.acedata.cloud/mcp, then finish the browser sign-in. New portable workspace configs use <project>/.mcp.json; the VS Code-specific format below uses <project>/.vscode/mcp.json or the user profile. Check MCP: List Servers. VS Code MCP setup.

  • Codex: codex mcp add grok --url https://grok.mcp.acedata.cloud/mcp, then codex mcp login grok. Its user settings are in ~/.codex/config.toml. Official Codex MCP guide.

Cursor project config (OAuth):

{
  "mcpServers": {
    "grok": {"url": "https://grok.mcp.acedata.cloud/mcp"}
  }
}

VS Code-specific workspace config (OAuth):

{
  "servers": {
    "grok": {"type": "http", "url": "https://grok.mcp.acedata.cloud/mcp"}
  }
}

Hosted API token

  1. Open Ace Data Cloud → Applications. Sign in, or register an account first.

  2. Find General application. If it has not been created, choose Initialize General Application. A general key can use the services available to your account; a dedicated application's key is limited to its service. Check the required service/model access and current pricing.

  3. Copy the API Key from the application card, or open Keys / Manage Keys → Create to create a separate key for this integration. Give it a recognizable name so you can revoke or rotate it independently.

  4. When creating a key, review Expiration Time, Set Usage Limitation, and Allowed APIs. Enable only the restrictions you need, allow the operations you will call, then click Create and copy the token from the key list. Task lookup does not grant permission to submit a generation. Confirm the application's balance can cover your first request; any trial allowance is whatever your account currently displays.

  5. Copy only the token value, without quotation marks, spaces, or Bearer . An API Key / API Token field expects this raw value; only a manually configured HTTP Authorization header needs the Bearer prefix. A platform-... account-management token cannot replace a generation API key.

Copy an application API key or open Manage Keys

Create a separately named key with optional restrictions

Save the key in the client's credential store or the environment variable shown below. Never put it in a prompt, shared workflow, screenshot, or Git commit. If authentication fails, check the full copied value, application/service access, expiration, usage limit, and balance before creating another key.

A fixed Bearer header bypasses OAuth. If the header is invalid, remove it before trying browser sign-in; keep the saved credential private.

For Claude Code, the shell expands the token when you add the server; treat the saved user MCP config as a secret:

export ACEDATACLOUD_API_TOKEN='YOUR_API_TOKEN'
claude mcp add --transport http --scope user grok https://grok.mcp.acedata.cloud/mcp \
  --header "Authorization: Bearer $ACEDATACLOUD_API_TOKEN"

For a Claude Code project config, put a variable reference in <project>/.mcp.json and set that variable in the environment that launches Claude Code:

{
  "mcpServers": {
    "grok": {
      "type": "http",
      "url": "https://grok.mcp.acedata.cloud/mcp",
      "headers": {"Authorization": "Bearer ${ACEDATACLOUD_API_TOKEN}"}
    }
  }
}

Cursor uses a different environment-variable syntax in ~/.cursor/mcp.json or an uncommitted project config:

{
  "mcpServers": {
    "grok": {
      "url": "https://grok.mcp.acedata.cloud/mcp",
      "headers": {"Authorization": "Bearer ${env:ACEDATACLOUD_API_TOKEN}"}
    }
  }
}

In VS Code, run MCP: Open User Configuration and merge this server plus its masked input; ${input:...} is for VS Code's user/workspace format and is not portable to the Agent Host .mcp.json format:

{
  "inputs": [
    {"id": "acedata-grok-token", "type": "promptString", "description": "AceDataCloud API token", "password": true}
  ],
  "servers": {
    "grok": {
      "type": "http",
      "url": "https://grok.mcp.acedata.cloud/mcp",
      "headers": {"Authorization": "Bearer ${input:acedata-grok-token}"}
    }
  }
}

For Cline, use its MCP configuration UI or CLI file ~/.cline/data/settings/cline_mcp_settings.json; its remote transport value is streamableHttp. For JetBrains AI Assistant, add a remote URL from Settings → Tools → AI Assistant → Model Context Protocol (MCP). For Zed, use a context_servers entry with the URL only for OAuth or add a local Bearer header. These clients have different configuration schemas; follow their current UI rather than copying another client's JSON. Cline · JetBrains · Zed.

Local stdio

Install the package and give the local process an API token:

python -m pip install mcp-grok
export ACEDATACLOUD_API_TOKEN='YOUR_API_TOKEN'
mcp-grok

For Claude Desktop local MCP, merge this entry into the file opened by its developer settings (~/Library/Application Support/Claude/claude_desktop_config.json on macOS). uvx requires uv on PATH:

{
  "mcpServers": {
    "grok": {
      "command": "uvx",
      "args": ["mcp-grok"],
      "env": {"ACEDATACLOUD_API_TOKEN": "YOUR_API_TOKEN"}
    }
  }
}

Keep this user-level file private. Self-hosted HTTP uses mcp-grok --transport http --port 8000; expose it only with suitable network and TLS controls. Local execution still calls the AceDataCloud API.

Check before using the service

  1. https://grok.mcp.acedata.cloud/health returning {"status":"ok"} checks endpoint reachability only.

  2. Confirm that the MCP client loads tools. grok_list_models is a reference tool; it does not verify downstream API access or balance.

  3. If you need a full API check, call grok_chat_completions with your own valid input after reviewing current service documentation and displayed pricing. If the result contains a task ID, call grok_get_task on that same ID until terminal success or failure. Do not resubmit the operation just to check progress.

For 401, check which auth route the client used and whether the token or OAuth session is valid. A 403 may mean an account permission or content moderation failure; read the returned error. Insufficient balance and downstream service failures need their own diagnosis. A listed tool or submitted task does not prove a successful result.

Development

pip install -e ".[dev,test]"
pytest --cov=core --cov=tools
ruff check .

Documentation

Documentation

License

MIT — see LICENSE.

Chat response mode

Chat tools return one complete JSON response. Omit stream or pass false. stream: true is rejected locally before an API request is sent. For native SSE streaming, use the HTTP API directly.

Available Tools

8 tools
grok_chat_completionsAInspect

Create a Grok (xAI) chat completion via the AceDataCloud Grok API.

Sends messages to a Grok chat model and returns the generated response in the
OpenAI-compatible chat completion format.

Use this when:
- You want to chat/reason with a Grok model (grok-4 / grok-3 family)
- You need vision/image understanding
- You need tool/function calling with Grok

For generating videos, use grok_text_to_video / grok_image_to_video instead.

Returns:
    JSON response containing the chat completion result.

Inputs: messages (required), n, seed, stop, +25 more. Returns: result

grok_get_prompt_guideAInspect

Get guidance on writing effective prompts for Grok Imagine video generation.

Shows how to structure prompts for best video generation results.

Returns:
    Complete guide with prompt structure, examples, and tips.

Inputs: no named parameters. Returns: result

grok_get_taskAInspect

Query the status and result of a video generation task.

Use this to check if a generation is complete and retrieve the resulting
video URLs and metadata.

Use this when:
- You want to check if a generation has completed
- You need to retrieve video URLs from a previous generation
- You want to get the full details of a generated video

Task states:
- 'pending'/'processing': Generation is still in progress
- 'succeeded': Generation finished successfully
- 'failed': Generation failed (check error message)

Returns:
    Task status and generated video information including URLs and state.

Inputs: task_id (required). Returns: result

grok_get_tasks_batchAInspect

Query multiple video generation tasks at once.

Efficiently check the status of multiple tasks in a single request.
More efficient than calling grok_get_task multiple times.

Use this when:
- You have multiple pending generations to check
- You want to get status of several videos at once
- You're tracking a batch of generations

Returns:
    Status and video information for all queried tasks.

Inputs: task_ids (required). Returns: result

grok_image_to_videoAInspect

Generate AI video from a reference image using Grok Imagine.

This animates your input image into a video clip. Provide a prompt to
describe the motion you want, and optional reference images to guide style.

Use this when:
- You have a specific image you want to animate
- You want consistent visual style from a reference
- You want to turn a still photo into a short video

For video generation from text only, use grok_text_to_video instead.

Returns:
    Task ID and generated video information including URLs and state.

Inputs: image_url (required), model, prompt, duration, +4 more. Returns: result

grok_list_actionsAInspect

List all available Grok Imagine tools and workflows.

Reference guide for what each tool does and how they fit together.

Returns:
    Categorized list of all tools and example workflows.

Inputs: no named parameters. Returns: result

grok_list_modelsAInspect

List all available Grok models (chat + video) and their capabilities.

Shows the chat models and the Grok Imagine video models with their features
and input rules. Use this to understand which model to choose.

Returns:
    Tables of all models with their capabilities.

Inputs: no named parameters. Returns: result

grok_text_to_videoAInspect

Generate AI video from a text prompt using Grok Imagine.

This creates a video from scratch based on your text description. Grok
Imagine will interpret your prompt and generate a matching video clip.

Use this when:
- You want to create a video from a text description
- You don't have a reference image to use
- You want maximum creative freedom

Only the 'grok-imagine-video-1.5-fast:reverse', 'grok-imagine-video:reverse',
and 'grok-imagine-video:official' models support text-to-video. For generating
a video from a reference image, use grok_image_to_video instead.

Returns:
    Task ID and generated video information including URLs and state.

Inputs: prompt (required), model, duration, resolution, +2 more. Returns: result

Tool Schema Changelog

Recent tool additions, removals, and schema changes observed during successful MCP inspections.

Loading tool schema changelog…

    TDQS

    A4.1/5.0

    Scored across 8 tools

    Disambiguation4/5

    The core split between chat, text-to-video, image-to-video, and task polling is clear, and descriptions explicitly redirect to the right sibling tool. The only fuzzy pair is grok_get_task vs grok_get_tasks_batch, which is largely resolved by the batch tool's 'more efficient' framing, and grok_list_actions vs grok_list_models overlap slightly in scope (workflows vs models).

    Naming Consistency4/5

    All eight tools share a grok_ prefix and snake_case, with consistent verb_noun shapes (list_actions, get_task, get_tasks_batch, get_prompt_guide). The generation tools drop the verb (grok_text_to_video, grok_image_to_video) and grok_chat_completions is noun-style, but the deviation is predictable and readable.

    Tool Count4/5

    Eight tools is well within a comfortable range for this surface and each maps to a real operation. Three of them (list_actions, get_prompt_guide, list_models) are essentially documentation helpers, which is slightly more self-referential overhead than ideal but not disproportionate.

    Completeness3/5

    Chat plus both video generation modes, model discovery, prompt guidance, and single/batch task polling cover the main workflow. However, there is no way to list recent tasks or task history, so an agent that loses a task ID is stuck, and there is no cancel/delete operation for in-flight or stored generations.

    Maintenance

    ActivityActive
    ResponsivenessNo recent community issues

    Related MCP Connectors

    Related MCP Servers