# FastGPT MCP for AI Agents AI Agent Connect > FastGPT. Manage your FastGPT knowledge bases, automate data ingestion, and run semantic searches directly through your AI agent. It lets you build and maintain RAG pipelines without leaving your chat interface. You can push content, create datasets, and monitor app configurations in one place. It is the primary way to handle knowledge base lifecycle management through an agent. This makes it easier to keep your data accurate and accessible. ## Overview - **Category:** data-management - **Price:** Free - **Endpoint:** https://edge.vinkius.com/vk_preview_5ych6Vwp9PYj4HCpHT5jojSMb56573gZQlH3hDOl/ai-agent-connect - **Tags:** fastgpt, rag-engine, knowledge-base, semantic-search, llm-orchestration, open-source ## Description This Connector lets you take full control of your FastGPT knowledge base from your AI agent. Instead of jumping back and forth between a dashboard and a terminal, you can handle the entire lifecycle of your RAG pipelines in one place. You can build out new datasets, push content into them, and manage your AI applications without ever leaving your current workspace. It handles the heavy lifting of indexing and vectorization by interacting with the platform for you. If you're building complex knowledge-based apps, this makes it much easier to iterate on your data. It's one of the many tools available in the Vinkius catalog that turns a static knowledge base into a dynamic resource your agent can actually manipulate. You can check your app configurations, search for specific information across your datasets, and even trigger chat completions that keep track of your session context. It's about making your RAG workflows more hands-on and less of a chore. You can manage data items, delete old records, and create new knowledge bases as your project grows. This connection bridges the gap between your data storage and your agent's ability to act on that data. Whether you're managing thousands of documents or just a few specific datasets, your agent can now query, update, and organize your information with high precision. ## Tools ### get_dataset_detail View the configuration and metadata for a specific knowledge base dataset. It helps you see how your data is structured. ### list_apps Get a complete list of all AI applications currently hosted on your FastGPT instance. It helps you keep track of your active tools. ### list_dataset_data View all individual data items currently stored within a specific dataset. This lets you see exactly what your agent can access. ### list_datasets List all your knowledge base datasets with options to filter by parent ID or keyword. This helps you organize large data inventories. ### push_dataset_data Add new content or update existing records in your knowledge base datasets. This is the primary way to ingest data via your agent. ### search_dataset_data Perform semantic searches against your datasets to test RAG relevance and accuracy. Use this to debug your retrieval quality. ### update_dataset_data Modify specific records within a dataset to keep your information current. This ensures your agent always has the latest facts. ### get_app_detail Pull specific configuration details for a single AI application. Use this to inspect your current app settings. ### chat_completions Send messages to a FastGPT app while maintaining session context and viewing intermediate steps. This is useful for interacting with your RAG-powered chat apps. ### get_embeddings Generate text embeddings for use in semantic search tasks outside of the FastGPT platform. This helps with custom vector work. ### create_dataset Create a new knowledge base dataset to house your documents or data chunks. This is the first step in building a new RAG pipeline. ### delete_dataset_data Remove specific data points from a dataset to keep your knowledge base clean. Use this to prune outdated information. ## Prompt Examples **Prompt:** ``` Create a new dataset for our 2024 marketing guidelines. ``` **Response:** ``` I've created a new dataset named '2024_Marketing_Guidelines'. You can now start pushing content to it using the push tool. ``` **Prompt:** ``` Search the HR dataset for our remote work policy. ``` **Response:** ``` Searching the HR dataset... I found 3 relevant results. The most relevant one says: 'Employees can work remotely up to 3 days a week with manager approval.' Would you like to see more details? ``` **Prompt:** ``` List all my apps and tell me which one handles customer support. ``` **Response:** ``` You have 3 active applications: 1. Customer Support, 2. Internal HR, 3. Sales Assistant. The 'Customer Support' app is the one you're looking for. ``` ## Capabilities ### Create knowledge base datasets Build new datasets to house your documents or data chunks via your agent. ### Push text content to datasets Add new content or update existing records in your knowledge base datasets. ### Run semantic searches on data Perform semantic searches against your datasets to test RAG relevance and accuracy. ### Manage AI application settings List and inspect AI applications to monitor their configurations and linked datasets. ### Generate text embeddings Generate text embeddings for use in semantic search tasks outside of the FastGPT platform. ### Track chat session context Send messages to an app while maintaining session context and viewing intermediate steps. ### Update existing dataset items Modify specific records within a dataset to keep your information current. ## Use Cases ### Bulk document ingestion A knowledge engineer needs to update 50 internal PDFs. They ask the agent to use `push_dataset_data` to ingest the files into the 'HR_Docs' dataset. ### Retrieval debugging A developer wants to see why a RAG response was wrong. They ask the agent to use `search_dataset_data` to check the retrieval score for a specific query. ### Application auditing An admin needs to audit all active apps. They ask the agent to use `list_apps` and `get_app_detail` to generate a summary of every active tool. ### Dynamic dataset creation A user wants to start a new project. They ask the agent to use `create_dataset` to set up a new knowledge base for 'Project_X_Research'. ## Benefits - Automate data ingestion by using `push_dataset_data` to send documents directly to your knowledge base. - Debug your RAG accuracy faster by running `search_dataset_data` to see how your agent retrieves information. - Keep your apps organized by using `list_apps` and `get_app_detail` to monitor configurations without opening a browser. - Manage large-scale content with `create_dataset` to build new knowledge bases for different projects in seconds. - Maintain data integrity by using `delete_dataset_data` and `update_dataset_data` to prune old or incorrect information. - Get precise search results by using `get_embeddings` for custom semantic tasks outside of the standard chat flow. ## How It Works The bottom line is you get a direct line of command to your FastGPT RAG infrastructure from your favorite AI client. 1. Subscribe to the Connector and get your API key from your FastGPT app settings. 2. Provide your Base URL and API key to your AI client. 3. Use your agent to create datasets, push data, or query your knowledge base. ## Frequently Asked Questions **What can I do with the FastGPT MCP?** This Connector lets you manage your FastGPT knowledge base directly through your AI agent. You can create datasets, push new information, and search your data without leaving your chat interface. **Can I use this to update my knowledge base?** Yes, you can use your agent to update existing records or push new content into your datasets to keep your information current. **How does this help with RAG accuracy?** It allows you to run semantic searches to see exactly how your agent retrieves information, making it much easier to debug and optimize your RAG results. **Can I create new datasets through my AI agent?** Absolutely. You can ask your agent to create a new knowledge base dataset for any project, and it will handle the setup for you. **Does this support session tracking?** Yes, when you use the chat tools, your agent can maintain session context and show you the intermediate steps the application takes to generate a response. **Can I use this for semantic search?** Yes, you can perform semantic searches against your datasets to test relevance and accuracy for specific queries. **How do I connect this to my FastGPT account?** You just need to provide your FastGPT Base URL and your API key from your app settings to your AI client to get started. **Can I use this with my self-hosted FastGPT instance?** Yes! Simply provide your custom domain in the **Base URL** field (e.g., `https://fastgpt.mycompany.com`). The Connector will route all requests to your specific instance. **How do I add new data to an existing dataset?** Use the `push_dataset_data` tool. You can send a JSON array of objects containing the text content to be indexed. FastGPT will handle the chunking and embedding automatically. **What is the 'goods_sign' used for in Pinduoduo tools?** Wait, this is the FastGPT FAQ. For FastGPT, you mostly need Dataset IDs and App IDs. The 'goods_sign' is specific to the Pinduoduo MCP. Always check you are using the correct tools for the specific platform.