# Nyckel ML MCP for AI Agents AI Agent Connect > Nyckel ML lets you connect your machine learning account to your AI agent to handle automated classification and semantic search. You can manage ML functions, track training samples, and update labels directly through a chat interface. It cuts out the manual overhead of switching between your ML platform and your workspace. ## Overview - **Category:** developer-tools - **Price:** Free - **Endpoint:** https://edge.vinkius.com/vk_preview_XHJXGFWpxs8iH5SOEbAtipKgPVXhxGte0LRcKzdL/ai-agent-connect - **Tags:** machine-learning, classification, semantic-search, automated-labeling, predictive-modeling, data-tagging ## Description This Connector brings your Nyckel machine learning account into your daily workflow. Imagine you're in the middle of a project and need to know if a specific piece of content hits your safety filters or if a product image matches your gallery's style. Instead of jumping back into a separate dashboard to run a manual query or check a function's status, you just ask your agent. You can pull up your current ML functions, check the confidence scores on new classifications, or even find similar samples using semantic search without leaving your current screen. If you're managing a large set of training data, you can oversee labels and upload new samples for your models right from the chat. It's about making your machine learning tools feel like a native part of your conversation. By using Vinkius, you can get this connected to your favorite AI client in minutes, giving you a unified way to interact with your models, labels, and training data. ## Tools ### annotate_ml_sample Assign a specific label to a training sample. This helps you organize your training data more effectively. ### create_ml_sample Add a new training sample to your dataset. This makes it easy to expand your training library on the fly. ### delete_ml_function Remove an ML function from your account. Use this to clean up old or unused models. ### get_ml_function Retrieve specific configuration and metadata for an ML function. This provides a clear look at your model settings. ### get_account_info Pull profile and workspace metadata for your authenticated account. This gives you a quick view of your account status. ### invoke_ml_function Classify text or image data using a specific ML function. This provides immediate predictions for your data. ### list_ml_functions List all active ML functions in your Nyckel account. This helps you see every model you have available. ### list_ml_labels Retrieve all available labels and categories for your models. This ensures you use the correct categories every time. ### list_ml_samples View the list of training samples in your account. This lets you see your current training data at a glance. ### semantic_search Find semantically similar samples based on your input data. This helps you discover related content quickly. ## Prompt Examples **Prompt:** ``` Classify this text: 'The delivery was very late and the food was cold' using function ID 'func_123'. ``` **Response:** ``` Invoking ML function func_123... The prediction for your text is **Negative Sentiment** with a confidence score of **98.45%**. Would you like me to check if there are any similar historical samples in your database? | Metric | Result | | :--- | :--- | | Prediction | Negative Sentiment | | Confidence | 98.45% | ``` **Prompt:** ``` Search my product gallery for an image similar to 'https://example.com/shoe.jpg' using function 'func_search_99'. ``` **Response:** ``` Executing semantic search... I've found **3 semantically similar samples** in your gallery. **Top Match:** * Running Shoe - Blue (Confidence: 95.2%) Would you like the metadata for the matching samples? ``` **Prompt:** ``` List all the machine learning functions in my Nyckel account. ``` **Response:** ``` Retrieving ML functions... You have **4 active functions**: * **Sentiment Classifier** (func_123) * **Product Search** (func_search_99) * **Logo Detector** * **Spam Filter** Which function would you like to inspect? ``` ## Capabilities ### Classify text or images Send data to your trained ML functions to get instant predictions and confidence scores. ### Find similar samples Query your search galleries to find semantically similar items based on your input. ### Manage ML functions List, retrieve, and delete ML functions in your account to keep your library organized. ### Track training data View, create, and manage the training samples used to build your models. ### Update labels Retrieve available labels and assign or update classification categories for your samples. ### View account profile Access your profile and workspace metadata for your authenticated account. ## Use Cases ### Automated Content Moderation A moderator asks the agent to check 50 user comments for hate speech using a specific function to see which ones need human review. ### E-commerce Image Discovery A merchant asks the agent to find 'shoes similar to this URL' to populate a new gallery with relevant products. ### Model Metadata Inspection A data scientist asks for the metadata of a specific ML function to check its confidence thresholds and configuration. ### Batch Data Labeling A team lead asks the agent to list all current samples and then assign labels to the ones that are currently missing them. ## Benefits - Skip the dashboard jumping by using invoke_ml_function to get instant predictions directly in your chat. - Manage your entire model library easily with list_ml_functions and get_ml_function to see all configurations. - Keep your training data organized by using create_ml_sample and annotate_ml_sample to build better models. - Find relevant content faster with semantic_search to pull similar samples from your existing galleries. - Keep your labels consistent by using list_ml_labels to see all categories available for your models. - Monitor your training progress directly by using list_ml_samples to see which data points are currently labeled. ## How It Works The bottom line is you get a direct chat interface to your machine learning operations. 1. Subscribe to the Nyckel ML MCP on Vinkius. 2. Enter your Nyckel Client ID and Client Secret in your AI client settings. 3. Start asking your agent to classify data, search samples, or manage your ML functions. ## Frequently Asked Questions **Can Nyckel ML MCP help me moderate comments automatically?** Yes. You can ask your agent to classify comments using your custom models, which allows you to identify and flag problematic content much faster. **How does Nyckel ML MCP find similar images?** It uses your account's semantic search functions to find samples that are visually or contextually similar to an image or URL you provide. **Can I use Nyckel ML MCP to manage my training data?** Yes. You can use it to view your current samples, create new ones, and assign labels to them directly through your AI client. **Is Nyckel ML MCP good for checking my ML model labels?** It is perfect for that. You can quickly list all available categories and check the labels on specific samples without opening a browser. **Can I see my ML functions using Nyckel ML MCP?** Yes. You can ask your agent to list all the functions in your account to see what models are ready for use and what their configurations are. **How do I use Nyckel ML MCP to classify new data?** Just provide the text or image URL in your chat and tell your agent which function to use. It will return the prediction and confidence score instantly. **How do I get my Nyckel credentials?** Log in to your Nyckel dashboard, navigate to your profile or settings, and look for the 'API Keys' section to find your Client ID and Client Secret. **Does this support image classification?** Yes! You can pass a publicly accessible image URL to the `invoke_ml_function` or `semantic_search` tools to classify or search based on visual content. **How accurate are the predictions?** Accuracy depends on the quality and quantity of training samples provided to your Nyckel function. The `invoke_ml_function` tool returns a confidence score for each prediction.