# Metatext MCP for AI Agents AI Agent Connect > Metatext MCP lets you run inference and manage NLP models without leaving your favorite AI client. It connects your Metatext account to your agent so you can handle datasets, model deployments, and model searches through plain conversation. It's a no-code way to manage your full NLP lifecycle, from training data to live production deployments. ## Overview - **Category:** developer-tools - **Price:** Free - **Endpoint:** https://edge.vinkius.com/vk_preview_77VAYStySeROKtVdW5OxQRJpWkfEWaDXQ23SvhBu/ai-agent-connect - **Tags:** nlp, model-training, inference, machine-learning, no-code, data-pipelines ## Description Metatext MCP connects your account to any AI agent to give you full control over your NLP models and data pipelines. You spend way too much time jumping between tabs to check if a model is live or to see if a dataset has enough records for training. This Connector changes that by letting your AI agent do the heavy lifting. Instead of manually navigating a dashboard, you can just ask your agent to find a specific model by name or check the status of your active deployments. It handles the grunt work of listing models, fetching metadata, and running predictions on the fly. If you're building data pipelines, you can tell your agent to create new records in a dataset or pull a list of existing ones to verify your inputs. It's about turning your ML workflow into a conversation. You can find this in the Vinkius catalog to get your NLP operations running in minutes without touching a single line of boilerplate code. This means you can iterate on your models faster, move from training to production with fewer hurdles, and keep your data organized without the constant friction of manual entry. Whether you are checking a sentiment classifier or a complex entity extractor, your agent acts as a direct bridge to your Metatext environment. It's the difference between clicking through twenty menus and just saying what you need done. ## Tools ### create_dataset_record Use create_dataset_record to add a new data point to a dataset for training or evaluation. It removes the need for manual CSV editing. ### get_dataset_details Use get_dataset_details to see specific information about a dataset to check its properties. This helps you verify your data before starting a training run. ### get_account_info Use get_account_info to check your account status and general information. This helps you keep track of your overall usage. ### get_model_details Use get_model_details to see the full specs and metadata for a specific NLP model. It helps you understand the training status and model type. ### list_nlp_datasets Use list_nlp_datasets to see all the datasets you've created in your Metatext account. This provides a high-level view of your data assets. ### list_model_deployments Use list_model_deployments to view a list of all your currently active deployments. This gives you a quick overview of what's live in production. ### list_nlp_models Use list_nlp_models to see every trained NLP model available in your Metatext account. This is great for auditing your available assets. ### list_dataset_records Use list_dataset_records to pull a list of all records within a specific dataset. It helps you check the volume and variety of your training data. ### run_model_inference Use run_model_inference to send data to a model and get a real-time prediction. This lets you test model accuracy on the fly. ### search_nlp_models Use search_nlp_models to find specific models by name when you need to grab one quickly. It saves you from scrolling through long lists. ## Prompt Examples **Prompt:** ``` What models do I have for sentiment analysis? ``` **Response:** ``` I've checked your Metatext account. You have the following models available for sentiment analysis: | Model Name | Status | Type | | :--- | :--- | :--- | | **Sentiment_Classifier_v2** | Live | Production | | **Customer_Feedback_Analysis** | Training | Evaluation | Would you like to see the metadata for either of these? ``` **Prompt:** ``` Run the sentiment model on 'This was a great experience!' ``` **Response:** ``` Running inference now... **Result:** - **Label:** Positive - **Confidence:** 99.4% - **Model Used:** Sentiment_Classifier_v2 Is there anything else you'd like me to analyze? ``` **Prompt:** ``` Add a new record to my 'Support_Tickets' dataset with the text 'The login button is broken' and label 'Bug'. ``` **Response:** ``` I've added that record for you. - **Dataset:** Support_Tickets - **Text:** "The login button is broken" - **Label:** Bug - **Status:** Successfully created Your training set is now updated. ``` ## Capabilities ### Run model predictions Metatext MCP allows you to execute inference on your deployed models to see how they handle real-world data. ### Search for NLP models Metatext MCP lets you find specific models by name to quickly identify which ones to use for different tasks. ### Create new dataset records Metatext MCP enables you to add new entries to your training or evaluation datasets via chat. ### Monitor active deployments Metatext MCP gives you a clear view of what models are currently live and how they're performing. ### Fetch model metadata Metatext MCP lets you pull detailed stats and training statuses for your entire library of NLP models. ### List dataset contents Metatext MCP lets you see every record in your datasets to ensure your training data is ready. ## Use Cases ### Testing sentiment analysis I need to test if my sentiment model is working. I ask my agent to run inference on a list of 50 customer reviews using run_model_inference and get the results in a table. ### Checking production status Is our entity extractor live? I ask my agent to list_model_deployments to see the current production status of our NLP tools without leaving my workspace. ### Expanding training sets We need more training data for the support bot. I tell my agent to create_dataset_record for several new customer queries to expand our training set for next week. ### Model discovery Which models do we have for translation? I ask my agent to search_nlp_models to find all relevant translation tools available in our Metatext account. ## Benefits - Stop tab-switching by checking model deployments and account info directly in your chat window. This saves you from jumping between your IDE and the web dashboard every time you need a status update. - Speed up data prep by using create_dataset_record to add new entries via chat. You can populate your training sets without ever opening a CSV file or a manual web form. - Test your models instantly with run_model_inference to see how they handle specific inputs before you go live. It gives you immediate feedback on your predictions in real-time. - Find the right tools faster by using search_nlp_models to filter your library of trained models. You can quickly identify the best model for a task without scrolling through a list. - Keep your training data organized by using list_nlp_datasets and list_dataset_records to verify your inputs. This ensures your data pipeline is consistent and ready for the next run. - Get a clear picture of your training progress by fetching metadata with get_model_details. You get a high-level view of training statuses and model specs in a single conversational turn. ## How It Works The bottom line is you get a direct conversational interface for your entire Metatext NLP infrastructure. 1. Subscribe to the Metatext MCP on Vinkius. 2. Add your Metatext API Key to your agent's configuration. 3. Start asking your agent to manage models, run inference, or update datasets. ## Frequently Asked Questions **How can I use Metatext MCP to manage my NLP models?** You can use it to list all your trained models, check their training statuses, and see which ones are currently live in production through your AI agent. **Can Metatext MCP run inference on my deployed models?** Yes, it allows your agent to send data to your models and return real-time predictions or classifications directly in the chat. **How do I add data to my datasets using the Metatext MCP?** You can simply tell your agent to create new records for a specific dataset. It handles the data entry for you so you don't have to use a web form or CSV. **Can I check my model deployment status with Metatext MCP?** Yes, your agent can pull a list of all active deployments to show you exactly what is live and what is still in the training phase. **Is there a way to search for specific models in Metatext MCP?** You can ask your agent to find models by name. It will search your Metatext account and list the relevant models that match your query. **How does Metatext MCP help with my ML data pipeline?** It streamlines your workflow by letting you manage datasets and run inference through a conversational interface, reducing the need for manual data handling. **How do I find my Metatext API Key?** Log in to Metatext and navigate to your account settings to find and copy your API Key. **Can I run inference on any model type?** Yes, as long as the model is fully trained and deployed, you can use the `run_model_inference` tool. **Is my AI data secure?** Absolutely. Your token is encrypted at rest and injected securely at runtime.