Haystack (deepset Cloud) MCP. Run RAG pipelines and search indexed knowledge.
Gemini Works with every AI agent you already use
…and any MCP-compatible client
Just plug in your AI agents and start using Vinkius.
Haystack (deepset Cloud) MCP Server manages your RAG pipelines and enterprise search context. It lets your AI agent interact with isolated workspaces, run document searches over indexed knowledge, and list metadata for both files and NLP pipelines.
Manage complex, contextual AI workflows without leaving your chat client.
What your AI agents can do
Get file
Retrieves specific metadata for an uploaded file.
Get pipeline
Gets the detailed configuration and structure of a single pipeline.
List files
Lists all files currently uploaded to the system.
List and manage isolated environments where different search contexts are maintained.
View the structure of NLP topologies, including embedding nodes and retriever logic.
Run a pipeline search by dispatching immediate LLM or Retriever invocations.
Trigger dense or sparse vector searches across your indexed enterprise knowledge.
List uploaded files and retrieve specific metadata attached to source document embeddings.
Ask AI about this MCP
Supported MCP Clients
Gemini Waiting for input…
Haystack (deepset Cloud) MCP Server: 7 Tools for RAG & Search
These tools allow your agent to list workspaces, manage pipelines, run searches, and inspect document metadata directly within the chat interface.
019d75aeget file
Retrieves specific metadata for an uploaded file.
019d75aeget pipeline
Gets the detailed configuration and structure of a single pipeline.
019d75aelist files
Lists all files currently uploaded to the system.
019d75aelist pipelines
Lists all available pipelines across the connected workspace.
019d75aelist workspaces
Lists all isolated environments you have set up.
019d75aerun pipeline
Executes a complete search run using a specified pipeline configuration.
019d75aesearch documents
Triggers a dense or sparse vector search against your entire document index.
Choose How to Get Started
Build a custom MCP for your own tools, or connect a ready-made integration from our catalog.
Build Your Own
Turn any API into an MCP. Import a spec, define Agent Skills, or deploy with MCPFusion.
- Import from OpenAPI, Swagger, or YAML specs
- Create Agent Skills with progressive disclosure
- Deploy to edge with MCPFusion framework
- Built in DLP, auth, and compliance on every call
- Real time usage dashboard and cost metering
- Publish to catalog or keep private
Make Your AI Do More
Start with Haystack (deepset Cloud), then connect any of our 4,700+ other servers whenever your AI needs more. One click, no limits.
- Use this MCP plus 4,700+ others, all in one place
- Add new capabilities to your AI anytime you want
- Every connection is secured and compliant automatically
- Track usage and costs across all your servers
- Works with Claude, ChatGPT, Cursor, and more
- New servers added to the catalog every week
What you can do with this MCP connector
Haystack (deepset Cloud) MCP Server manages your whole RAG pipeline and enterprise search context. It lets your AI agent interact with isolated workspaces, run document searches over indexed knowledge, and list metadata for both files and NLP pipelines. You manage complex, contextual AI workflows without ever leaving your chat client.
Audit Workspaces
You'll list all isolated environments you've set up with list_workspaces and then get detailed info on a specific environment with list_pipelines and get_pipeline. Inspect Pipelines
You can view the structure of any NLP topology, checking out embedding nodes and retriever logic using get_pipeline. Execute Search Tests
To test your setup, you run a search using a specified pipeline configuration with run_pipeline.
You can also trigger a direct, dense or sparse vector search against your entire document index using search_documents. Perform Vector Search
Use search_documents to trigger dense or sparse vector searches across your indexed enterprise knowledge. Check File Metadata
To see what files you've uploaded, run list_files to get a list of everything in the system; you can then check specific source document embedding metadata for any file using get_file.
How Haystack (deepset Cloud) MCP Works
- 1 Subscribe to the server and provide your deepset Cloud API credentials.
- 2 Your AI client connects, granting the agent full access to the Haystack tools.
- 3 You prompt the agent (e.g., 'Check the compliance documents'). The agent sequences
list_workspaces->run_pipeline->search_documentsto deliver the answer.
The bottom line is, you manage your entire RAG pipeline lifecycle—from setup to execution—using only natural conversation.
Who Is Haystack (deepset Cloud) MCP For?
This is for the ML Engineer who needs to test a new knowledge retrieval model without touching the cloud console. It's for the Product Manager who needs to prove that the documents are indexed correctly before launch. It's for the Data Scientist who needs to audit complex pipelines rapidly. It lets you treat your entire data stack like a set of chat commands.
Tests new embedding models or retriever logic by calling run_pipeline and verifying the search results against source documents.
Verifies the document indexing status and tracks search performance by listing files and workspaces.
Audits existing RAG pipelines and document sources by calling list_pipelines and search_documents to confirm data coverage.
What Changes When You Connect
- Test search logic instantly. Instead of clicking through a dashboard, use
run_pipelineto dispatch immediate LLM or Retriever invocations and test your RAG setup. - See your data structure clearly. Use
list_pipelinesandget_pipelineto visualize NLP topologies, checking every embedding node and retriever logic without needing the cloud console. - Manage your context isolation.
list_workspaceslets you list and select separate environments, ensuring your search queries hit the right isolated data set. - Audit source files easily. Use
list_filesto see what documents are uploaded, andget_fileto check the metadata attached to specific embeddings. - Deep search capabilities.
search_documentsruns vector searches over your whole indexed knowledge base, going beyond simple keyword matching. - End-to-end visibility. You can chain tools—e.g.,
list_workspacesthenrun_pipeline—to manage the entire lifecycle of an AI search query.
Real-World Use Cases
Auditing a Production Search
A Product Manager needs to confirm if the 'Q3 Compliance' documents are indexed correctly. They use the agent to run list_workspaces to select the right environment, then list_files to confirm the document count. Finally, they call search_documents to validate that the desired snippets appear.
Testing a New RAG Model
An ML Engineer wants to test a new embedding model. They use the agent to call get_pipeline on the staging workspace to check the current topology. They then use run_pipeline to execute the search with the new model, verifying the results before committing the change.
Debugging Pipeline Failures
A Data Scientist finds a pipeline is giving weird results. They use the agent to call get_pipeline to inspect the exact sequence of nodes (e.g., checking the retriever logic). They then use search_documents to narrow the search scope and pinpoint the data source issue.
Onboarding a New Data Source
A team member has uploaded 50 new PDFs. They use the agent to call list_files to confirm the uploads. They then use run_pipeline to run a preliminary search on the new data, checking if the pipeline handles the new file types correctly.
The Tradeoffs
Manual Console Checks
The developer logs into the deepset console, navigates to the workspace, manually checks the file list, and then runs the search, copying results into a spreadsheet.
→
Use the agent to sequence the process. First, call list_workspaces to select the context. Then, use run_pipeline to execute the search, and finally, get_file to pull metadata for the source IDs. It's all in chat.
Ignoring Context
The agent just runs search_documents without knowing which workspace or pipeline to use, leading to ambiguous or stale results.
→
Always start by calling list_workspaces to confirm the correct isolated environment. This ensures your run_pipeline call executes against the intended, governed context.
Assuming Full Functionality
Thinking that just running search_documents is enough when the data requires specific NLP processing steps.
→
Don't just search. First, use list_pipelines to see what topologies exist. Then, use run_pipeline to ensure the data runs through the required pre-processing steps before the search.
When It Fits, When It Doesn't
Use this MCP Server if your core task is managing, testing, and debugging Retrieval-Augmented Generation (RAG) pipelines using external, indexed knowledge. You need to move the entire lifecycle—from checking file metadata (get_file) to running the full search (run_pipeline)—into a single, conversational interface. Don't use this if you simply need basic file operations or general LLM prompting; other general-purpose agents handle that. You also don't need it if you are only running single, isolated vector searches without needing to audit the pipeline structure. If you only need to see what files exist, list_files is enough, but this server lets you connect the files to the pipelines and the search results.
Independent Platform Disclaimer: Vinkius is an independent platform and is not affiliated with, endorsed by, sponsored by, verified by, or otherwise authorized by deepset Cloud. All third-party trademarks, logos, and brand names are the property of their respective owners. Their use on this website is strictly for informational purposes to identify service compatibility and interoperability.
VINKIUS INFRASTRUCTURE
Cloud Hosted
Managed infra
V8 Isolated
Sandboxed per request
Zero-Trust Proxy
No stored credentials
DLP Enforced
Policy on every call
GDPR Compliant
EU data residency
Token Compression
~60% cost reduction
Works with Claude, ChatGPT, Cursor, and more
The Model Context Protocol standardizes how applications expose capabilities to LLMs. Instead of operating in isolation, your AI gains direct access to external platforms, live data, and real-world actions through secure, standardized connections.
This server provides 7 capabilities that interface natively with Claude, ChatGPT, Cursor, and any MCP client. No middleware. No custom integration required.
Available Capabilities
Manually auditing search pipelines is a nightmare.
Right now, testing a search pipeline means opening the deepset console. You have to switch tabs to check the file list, click into the workspace to see which pipelines are active, and then manually trigger a run. You copy results from one screen, and then you go to another screen to check the source file metadata. It takes a decade and a dozen clicks.
With this server, you just tell your agent, 'Test the compliance search.' The agent coordinates `list_workspaces` and `run_pipeline` behind the scenes. You get the answer, and you can ask it to pull the source file IDs using `get_file`—all in the chat window. It's a single, coordinated operation.
Haystack (deepset Cloud) MCP Server: Search and Validate
You eliminate the need to jump between the deepset console, the file browser, and the pipeline visualization tool. You never have to manually run a search and then manually check the source document metadata in a different view.
This tool lets you treat your entire complex data stack—the pipelines, the workspaces, and the documents—as a single, queryable entity. It's instant validation for your ML teams.
Common Questions About Haystack (deepset Cloud) MCP
How do I use the `list_workspaces` tool with Haystack (deepset Cloud) MCP Server? +
The list_workspaces tool shows you all isolated environments you've set up. You use this first to confirm the specific context (e.g., 'staging' or 'production') before running any search or pipeline operations.
What is the difference between `search_documents` and `run_pipeline`? +
search_documents executes a raw vector search over the index. run_pipeline executes a complete, configured workflow (the pipeline) which includes embedding nodes and retriever logic, making it more comprehensive.
Can I check metadata for a file using `get_file`? +
Yes, the get_file tool lets you retrieve specific metadata for an uploaded file. This is crucial for auditing, as it tells you exactly what information is attached to the source document embedding.
How do I see what pipelines are available? Use `list_pipelines`. +
Simply calling list_pipelines lists all available NLP topologies in the current workspace. This shows you which workflows (like 'default-rag') are ready to be tested via run_pipeline.
How do I check which files I can manage using the `list_files` tool? +
The list_files tool shows all files currently uploaded. It lets you see the names and basic status of your document repository.
What is the purpose of the `get_file` tool? +
The get_file tool retrieves specific metadata about a file. You use it when you need details beyond just the file name, like creation dates or authors.
If I want to see all available search methods, should I use `list_pipelines` or `search_documents`? +
Use list_pipelines first. This shows you the available NLP topologies and search setups. Once you pick a pipeline, you can run the search using run_pipeline.
How do I run a specific knowledge search using the `search_documents` tool? +
The search_documents tool runs a direct vector search against your indexed knowledge. You pass it the query and the target index to get immediate context.
Can I test my RAG pipelines directly via my AI agent? +
Yes. Use the run_pipeline tool to dispatch a query to any registered pipeline in your workspace. Your agent will return the response from the NLP topology, allowing you to verify retriever performance and LLM grounding without leaving your workspace.
How can I audit my document indexing status? +
Ask your agent to list files in your workspace. You can then get specific metadata for any file to ensure embeddings and attributes are correctly attached. This is essential for debugging retrieval issues in production environments.
Is it possible to manage multiple deepset Cloud workspaces? +
Absolutely. The agent provides high-level workspace listing, allowing you to navigate across tenant boundaries and isolation zones easily. You just need to provide the workspace name to any pipeline or search command.
Use it with your favorite AI tools
Connect this server to Cursor, Claude, VS Code, and more.
More in this category
Firecrawl
Crawl and scrape entire websites into clean LLM-ready markdown with a single API call that handles JavaScript rendering.
Crawlbase
Scrape and crawl via Crawlbase — perform HTML extraction, handle JS-rendered pages, bypass CAPTCHAs, and scrape social profiles directly from any AI agent.
Context7
Empower AI agents via Context7 — pull up-to-date documentation and code examples for any library or framework directly into your workspace.
You might also like
Bear
Manage your Bear notes via AI — search, create, and organize your markdown knowledge base autonomously.
Start.gg
Manage esports tournaments via Start.gg — fetch event IDs, track standings, and report match results directly from any AI agent.
Fastly
Manage edge computing and CDN via Fastly — monitor and activate service versions, manage domains and backends, and purge cache directly from any AI agent.