# SenseCore Platform MCP for AI Agents AI Agent Connect > SenseCore Platform MCP lets you manage SenseTime's industrial AI infrastructure directly from your AI client. You can trigger chat completions, monitor GPU cluster health, track project quotas, and manage foundation models without leaving your workspace. It bridges the gap between your agent and high-performance compute resources for production-grade AI deployment. ## Overview - **Category:** industry-titans - **Price:** Free - **Endpoint:** https://edge.vinkius.com/vk_preview_WGerX5EoMF3cLaPImYP34A8UFoRYYgmvQyPkRFRu/ai-agent-connect - **Tags:** sensecore, sensetime, sensechat, llm-api, compute-resources, gpu-orchestration, industrial-ai ## Description Managing industrial-grade AI infrastructure usually involves a mess of browser tabs. You're constantly switching between your IDE, your terminal, and the SenseCore Console just to see if your GPU nodes are still alive or if your last training run actually finished. This Connector changes that by letting your AI agent handle the heavy lifting for you. Instead of manually hunting for metrics or checking if a training job is still running, you can just ask your agent for a status update or a list of available models in your project. It connects you to the SenseCore Platform, giving your agent the eyes and hands it needs to manage SenseTime's foundation models and compute resources. You can find this Connector in the Vinkius catalog to get a direct line to your infrastructure. You can set up assistants, manage conversation threads, and oversee long-running inference tasks with simple commands. It's about moving from manual monitoring to active orchestration. You stop worrying about the underlying hardware and start focusing on the results of your model deployments. ## Tools ### get_run_status Check the status of an active assistant run. This is useful for seeing if a long-running task is still processing. ### list_files List uploaded files. This allows you to see all the documents or data files attached to your project. ### list_messages Retrieve the message history of a thread. Use this to pull back all previous interactions for context. ### list_models List all available SenseNova models. This shows you every foundation model you can call from your project. ### create_assistant Define a new AI assistant. This lets you set up specific personalities or instructions for different tasks. ### create_run Execute an assistant on a thread. This triggers the assistant's logic to process a specific conversation. ### create_thread Initialize a new conversation thread. Use this to start a fresh interaction with a clean context. ### get_assistant_details Get complete configuration for an assistant. Use this to check the specific settings and parameters of your active assistants. ### list_assistants List all configured assistants. Use this to see every assistant you've built within your project. ### create_message Add a message to a thread. This helps you append new information or prompts to an ongoing conversation. ### chat_completions Send a message to a SenseCore large language model. Use this to get direct responses from SenseTime's foundation models. ## Prompt Examples **Prompt:** ``` What's the status of my last inference run? ``` **Response:** ``` Checking your SenseCore tasks... Your latest run is currently **Completed**. - **Success Rate:** 100% - **P99 Latency:** 45ms - **Status:** Finished successfully. Would you like to see the full logs? ``` **Prompt:** ``` List all the models I can use in my project. ``` **Response:** ``` I've pulled the list of available models for your project: | Model Name | Type | Status | | :--- | :--- | :--- | | SenseChat-5 | Foundation | Active | | SenseVision | Multimodal | Active | | SenseAudio | Specialized | Active | Which one should we use? ``` **Prompt:** ``` Check if our GPU cluster is still healthy. ``` **Response:** ``` Querying SenseCore health metrics... **Service Status:** Healthy - **Uptime:** 99.98% - **Request Success Rate:** 100% - **Current Latency:** 45ms Everything looks stable. ``` ## Capabilities ### Trigger chat completions Send messages to foundation models with persistent context and history. ### Inspect foundation models List all available models and view their specific technical specifications. ### Monitor GPU clusters Check node availability and track quota consumption across your projects. ### Track service health Get real-time latency and health metrics for your deployed model services. ### Manage async tasks List and track the status of long-running training or inference jobs. ### Configure AI assistants Define, list, and manage custom AI assistants within your project. ## Use Cases ### Checking training job status An ML Ops engineer needs to know if a training job is still running. They ask the agent to check the status of the latest run. ### Finding available models A developer needs to see which models are available for a new project. They ask the agent to list the models in their project. ### Building a custom support bot An enterprise team wants to build a custom support bot. They use the Connector to define the bot's personality and start new threads. ### Monitoring cluster health A team is hitting quota limits and needs to see who's using what. They ask the agent to monitor compute node availability. ## Benefits - Stop jumping between tabs to check your GPU cluster health and node availability. - Track your project's compute quota in real-time to avoid hitting limits during peak usage. - Manage complex conversation histories easily by pulling back full message logs for context. - Deploy and configure custom assistants quickly with dedicated configuration tools. - Automate your inference pipeline by triggering completions directly from your agent. - Monitor real-time latency and success rates for all your deployed model services. ## How It Works The bottom line is you get a programmatic bridge to SenseTime's industrial AI infrastructure. 1. Get your API Key, Secret Key, Organization ID, and Project ID from the SenseCore Console. 2. Add your credentials to the Connector configuration fields. 3. Ask your agent to list models or check GPU health to confirm the connection. ## Frequently Asked Questions **Can I use the SenseCore Platform MCP to manage my GPU cluster?** Yes. You can use it to monitor compute node availability and check real-time health metrics like latency and uptime directly through your AI client. **Does SenseCore Platform MCP work with SenseTime's foundation models?** It does. You can list all available SenseNova models and trigger chat completions using SenseTime's foundation models. **How do I see my project's quota usage with SenseCore Platform MCP?** You can ask your agent to track quota consumption across your organizational projects to ensure you don't hit limits unexpectedly. **Can I create custom assistants using SenseCore Platform MCP?** Yes, the Connector includes tools to define new AI assistants and retrieve their full configurations to help build custom workflows. **Is SenseCore Platform MCP good for tracking long-running training jobs?** It's built for that. You can list and track the status of long-running training or inference tasks on the SenseCore infrastructure. **Can I see my message history through the SenseCore Platform MCP?** Yes. You can retrieve the full history of messages for any conversation thread to keep your AI agent in context. **Can I automatically list all available models in my SenseCore project?** Yes! Use the `list_models` tool. Your agent will retrieve a complete list of all SenseTime foundation models and specialized variants currently active in your account. **How do I check the health status of my deployed model services?** Use the `get_service_health` tool with the specific Service ID. The agent will return real-time metrics on availability, throughput, and average latency. **Can I monitor GPU resource utilization via the AI agent?** Yes! The `get_resource_usage` tool retrieves granular metrics on compute node utilization and remaining quota for your specific project environment.