# VectorShift MCP for AI Agents AI Agent Connect > VectorShift is an Connector that lets you manage AI pipelines and RAG workflows directly from your AI client. You can build multi-step automations, index documents into vector databases, and deploy chatbots without leaving your workspace. It is built for teams who need to move beyond simple chat and into production-ready AI operations. ## Overview - **Category:** knowledge-management - **Price:** Free - **Endpoint:** https://edge.vinkius.com/vk_preview_EmQ0ykbVFnJJJOx8Wrsr9PBOJ4yiohOhm84Pa96G/ai-agent-connect - **Tags:** rag, ai-workflows, automation, knowledge-base, llm-ops ## Description Imagine you're tired of switching between your IDE, a database manager, and a workflow builder just to see if your RAG pipeline is actually working. With VectorShift, you can handle the entire lifecycle of your AI operations from a single chat interface. You can build out multi-step pipelines that handle data extraction, run them in parallel, or pause them when you need to tweak a step. If you need to ground your agent in specific company data, you can create knowledge bases and index your files or URLs on the fly. It handles the heavy lifting of semantic search and data processing so you can focus on the logic. Because Vinkius hosts this Connector, you can connect it to your favorite tools like Cursor or Claude and keep your entire LLM-Ops stack in one place. It turns your AI agent into a hands-on operator for your VectorShift environment, making it much easier to deploy bots and manage your data assets without jumping through hoops. ## Tools ### delete_knowledge_base_documents Delete specific files from a knowledge base by their ID. This helps you keep your data clean and relevant. ### list_knowledge_bases See all your available knowledge bases. Use this to manage your different data silos. ### upload_chatbot_files Upload files to a specific chatbot session. This lets you give the bot immediate context for a conversation. ### bulk_run_pipeline Run multiple pipeline instances at the same time. This is great for processing large batches of data quickly. ### create_chatbot Set up a new chatbot instance. Use this to quickly spin up a new conversational point of contact. ### create_knowledge_base Build a new vector-based knowledge base. This is your first step for grounding your AI in specific data. ### create_pipeline Build a new multi-step AI workflow. Use this to define the logic for your automated processes. ### create_transformation Create a custom Python or JS transformation. This lets you handle complex data logic before it hits your model. ### delete_chatbot Remove a chatbot from your account. Use this to clean up old or unused conversational instances. ### delete_knowledge_base Remove an entire knowledge base. Use this when you no longer need a specific data set indexed. ### delete_pipeline Remove a specific pipeline from your list. Use this to keep your workspace organized. ### delete_transformation Remove a custom transformation. Use this to clear out old logic or scripts. ### get_chatbot Fetch the details of a specific chatbot. Use this to check the status or configuration of an existing bot. ### get_knowledge_base Fetch details for a specific knowledge base. This helps you verify the status of your indexed data. ### get_pipeline Fetch the details of a specific pipeline. Use this to check the configuration of a workflow you're building. ### get_transformation Fetch the details of a specific transformation. Use this to review the logic of your custom scripts. ### index_knowledge_base Add files or URLs to your knowledge base. This is how you feed your RAG system new information. ### list_chatbots See a list of all your chatbots. Use this to get an overview of your active conversational agents. ### list_knowledge_base_documents Find all the documents currently in a knowledge base. This helps you audit what data your AI can see. ### list_pipelines See a list of all your pipelines. Use this to keep track of your active AI workflows. ### list_transformations See a list of all your custom transformations. Use this to manage your collection of data logic scripts. ### pause_pipeline Pause a pipeline that's currently running. Use this to stop execution without deleting the progress. ### query_knowledge_base Perform a semantic search on a knowledge base. This is how you get grounded answers from your data. ### resume_pipeline Resume a pipeline that was paused. Use this to pick up where the workflow left off. ### run_chatbot Send a message to a chatbot and get a response. Use this to test your bot's behavior in real-time. ### run_pipeline Execute a pipeline with specific inputs. This starts the actual automation process for your data. ### run_transformation Execute a custom transformation with your inputs. Use this to run specific data logic on demand. ### terminate_chatbot End an active chatbot session. Use this to clear out a specific conversation or session. ### terminate_pipeline Stop a pipeline that's currently running. Use this to kill a process that isn't behaving correctly. ## Prompt Examples **Prompt:** ``` Check my active pipelines and tell me if the 'Data Scraper' is running. ``` **Response:** ``` I've checked your active workflows. Here is the status of your 'Data Scraper' pipeline: | Pipeline Name | Status | Last Action | | :--- | :--- | :--- | | Data Scraper | **Running** | Extracting URLs | | Lead Gen | Paused | Waiting for manual resume | Would you like me to resume the Lead Gen task? ``` **Prompt:** ``` I need to add our new HR policy PDF to the internal knowledge base. ``` **Response:** ``` I can do that for you. I'll use the index_knowledge_base tool to add that file to your 'Company Wiki' knowledge base. **Next Steps:** 1. Please provide the file path or URL. 2. I'll confirm once the indexing is complete. 3. You can then ask me questions about the new policy immediately. ``` **Prompt:** ``` Run the data cleaning script on the latest user logs. ``` **Response:** ``` I'll start that now. I'm using the run_transformation tool to execute your 'User Log Cleaner' script. **Process:** - **Input:** `logs_v2.csv` - **Action:** Running Python transformation - **Output:** `cleaned_logs.csv` I'll let you know as soon as the transformation is finished. ``` ## Capabilities ### Run multi-step AI pipelines Execute complex workflows with specific inputs and monitor their progress. ### Index documents into vector databases Add files or URLs to a knowledge base to ground your AI in your own data. ### Deploy and manage chatbots Create, run, and manage conversational bots directly from your chat. ### Execute custom Python or JS transformations Run custom scripts to clean or prep data before it enters your workflow. ### Query knowledge bases with semantic search Get grounded answers from your indexed data using natural language queries. ### Manage pipeline execution states Pause, resume, or stop running processes without losing your current progress. ## Use Cases ### Automating a high-volume data extraction An ops engineer needs to pull data from 500 URLs. They ask their agent to use bulk_run_pipeline to trigger the extraction and list_pipelines to monitor the progress. ### Updating a company wiki for a RAG bot A product manager wants the chatbot to know about a new refund policy. They ask the agent to use index_knowledge_base to upload the new PDF and then run_chatbot to test it. ### Cleaning messy user data before processing A developer needs to strip PII from logs. They ask the agent to create_transformation using a Python script and then run_transformation to clean the data before the pipeline starts. ### Managing a fleet of customer support bots An admin needs to see all active bots. They use list_chatbots to get an overview and terminate_chatbot to shut down any sessions that are behaving incorrectly. ## Benefits - Skip the dashboard. Use list_pipelines and run_pipeline to manage your entire automation suite from your chat window instead of clicking through tabs. - Ground your AI instantly. Use index_knowledge_base to feed your agent new files or URLs so it has the latest context for every query. - Handle complex logic easily. Create custom Python or JS logic with create_transformation to clean and prep data before it hits your model. - Manage high-volume tasks. Use bulk_run_pipeline to process large batches of data in parallel without having to manually trigger each one. - Control your workflows. Use pause_pipeline and resume_pipeline to manage long-running processes without losing your place or restarting from scratch. - Deploy bots faster. Use create_chatbot and upload_chatbot_files to spin up and context-load new conversational agents in seconds. ## How It Works The bottom line is you get a direct command line to your entire VectorShift automation suite. 1. Subscribe to the Connector and enter your VectorShift API Key. 2. Connect the Connector to your preferred AI client like Claude or Cursor. 3. Start building, indexing, and running pipelines via natural conversation. ## Frequently Asked Questions **Can I use VectorShift MCP to manage my RAG pipelines?** Yes. This Connector gives your AI agent the ability to create, run, and monitor RAG pipelines directly. You can manage the entire lifecycle of your data workflows without leaving your chat interface. **How does the VectorShift MCP handle my company's private data?** It connects to your existing VectorShift account. You can use it to index your private files and URLs into knowledge bases, which your agent then queries to provide grounded, accurate responses. **Can I run custom code with the VectorShift MCP?** Yes, you can. The Connector includes tools to create and execute custom Python and JavaScript transformations, allowing your agent to perform complex data processing as part of an automated workflow. **Is the VectorShift MCP good for bulk processing?** It is designed for that. You can use the bulk_run_pipeline tool to execute multiple instances of a pipeline in parallel, making it ideal for high-volume data tasks. **Can I pause a running workflow using VectorShift MCP?** Yes, you can. If a pipeline hits a point where you need to intervene, you can use the pause_pipeline tool to stop it and resume_pipeline once you're ready to continue. **Does the VectorShift MCP support multiple knowledge bases?** Yes, you can create and manage multiple knowledge bases. This allows you to separate different types of data, like HR docs, product specs, and customer support logs, into distinct silos. **How do I search for specific information within my VectorShift knowledge base?** Use the `query_knowledge_base` tool with your Knowledge Base ID and the search query. The agent will perform a semantic search and return the most relevant data chunks. **Can I trigger a specific AI workflow with custom parameters?** Yes! Use the `run_pipeline` tool. Provide the Pipeline ID and a JSON object mapping your input names to their respective values to start the execution. **Is it possible to add new documents to a knowledge base through the agent?** Absolutely. Use the `index_knowledge_base` tool to add data (such as URLs or file content) to an existing knowledge base for real-time RAG updates.