The Grok MCP server is an implementation of the Model Context Protocol that connects your AI agent and assistants like Claude, Cursor, etc directly to your Grok account. It provides structured and secure access to Grok's text, image, video, embedding, model, and account capabilities, so your agent can generate responses, create media, produce embeddings, browse available models, and verify account access on your behalf.
- Text and multimodal responses: Have your agent generate focused responses from text or structured multimodal input, with optional reasoning and search.
- Image creation: Direct your agent to create one or more images from a detailed text prompt using an available Grok image model.
- Video generation and tracking: Let the agent start a video from a text prompt, optionally use an image as context, and check its progress later.
- Text embeddings: Instruct your agent to turn one or more text inputs into vector embeddings for search, matching, or classification tasks.
- Model and account checks: Have your agent review available models and pricing details, verify the connected account, and inspect API key permissions or blocking status.
