# Haystack (deepset Cloud) MCP for AI Agents AI Agent Connect

> Haystack (deepset Cloud) MCP lets your AI client talk directly to your deepset Cloud account. It gives your agent the ability to run RAG pipelines, manage workspaces, and query your indexed enterprise documents without you having to switch tabs or manually navigate a dashboard.

## Overview
- **Category:** friends-mcp
- **Price:** Free
- **Endpoint:** https://edge.vinkius.com/vk_preview_Tf2g8zLZroyDlliFU39whBZWkEDykSKQVz38t3zr/ai-agent-connect
- **Tags:** rag-pipelines, llm-framework, enterprise-search, nlp-topologies, embedding-models, document-retrieval

## Description

You're building a RAG system, and half your day is spent jumping between your code editor, your cloud console, and your documentation. You need to know if a specific pipeline is actually live, check which files made it into the index, or just see if a search query is hitting the right retriever. This Connector changes that by putting your deepset Cloud environment right inside your chat interface. Instead of hunting for buttons, you can just ask your agent to show you the status of a workspace or trigger a search to see how a new embedding model is performing. It makes the ops side of AI development feel like a conversation rather than a chore. Since Vinkius hosts this, you can get it hooked up to your favorite tools quickly. It's about getting eyes on your data and pipelines where you're already working, making it much easier to iterate on your search logic and verify that your knowledge base is actually doing what you told it to do.

## Tools

### list_workspaces
List all workspaces. Use this to keep different projects' search contexts separate.

### list_pipelines
List pipelines. It helps you see what's available to run in your cloud account.

### run_pipeline
Run a pipeline search. This lets you test RAG logic instantly with a natural language query.

### get_pipeline
Get pipeline details. Use this to check out the exact configuration of a retriever.

### list_files
List uploaded files. It's the easiest way to check if your documents actually made it to the cloud.

### get_file
Get file metadata. Use this to see the details attached to your source document embeddings.

### search_documents
Search documents in index. This triggers the actual vector search over your enterprise knowledge.

## Prompt Examples

**Prompt:** 
```
What pipelines do I have in my production workspace?
```

**Response:** 
```
I found 3 pipelines in your production workspace:

| Pipeline Name | Type | Status |
| :--- | :--- | :--- |
| `default-rag` | Hybrid Retriever | Active |
| `semantic-search` | Dense Vector | Active |
| `hybrid-retriever` | Sparse Vector | Active |

Which one would you like to inspect?
```

**Prompt:** 
```
Run a search for 'how do I reset my password' using the default-rag pipeline.
```

**Response:** 
```
Running search on `default-rag`...

**Results Found:**
- **Snippet 1:** To reset your password, navigate to the settings page and click 'Forgot Password'.
- **Snippet 2:** For security reasons, password reset links expire after 15 minutes.

**Source Documents:**
- `user_guide_v2.pdf` (ID: 9823)
- `security_faq.md` (ID: 1029)

Would you like to see the full metadata for these files?
```

**Prompt:** 
```
List the files in my knowledge-base workspace.
```

**Response:** 
```
You have 12 files in your 'knowledge-base' workspace. Here are the most recent uploads:

*   `q4_report.pdf`
*   `security_guidelines.md`
*   `api_docs.txt`

I can fetch the metadata for any of these if you need to see their embedding details.
```

## Capabilities

### Run RAG pipeline searches
Trigger immediate LLM or Retriever invocations to test your RAG logic.

### Manage isolated workspaces
List and organize different environments for various search contexts.

### Inspect document metadata
View the metadata attached to your source document embeddings.

### Trigger vector searches
Execute dense or sparse vector searches over your indexed knowledge.

### Visualize NLP topologies
Get details on embedding nodes and retriever logic within your pipelines.

## Use Cases

### Verifying document indexing
A developer needs to check if a new PDF was indexed. They ask the agent to list files in the 'HR' workspace.

### Testing retriever logic
An ML engineer wants to see if a hybrid retriever is working. They ask the agent to run a search using the 'hybrid-search' pipeline.

### Audit production pipelines
A product manager wants to see all available search options. They ask the agent to list all pipelines in the production workspace.

### Inspecting metadata
A developer needs to see the metadata of a specific document. They ask the agent to get the file info for 'q4_report.pdf'.

## Benefits

- Run RAG searches directly in your chat window without switching to the cloud console.
- Keep your projects organized by managing separate search contexts for different environments.
- Verify your data integrity quickly to see what's currently indexed in your cloud account.
- Debug your NLP topologies faster by checking the specific configuration of any retriever.
- Test new embedding models on the fly by triggering searches with custom queries.
- Get instant visibility into document metadata without opening the deepset Cloud dashboard.

## How It Works

The bottom line is you can manage your entire RAG infrastructure through a chat window instead of a complex dashboard.

1. Subscribe to the Connector and grab your deepset Cloud API URL and API Key.
2. Plug those credentials into your AI client's connection settings.
3. Ask your agent to list pipelines, check files, or run a search on your indexed data.

## Frequently Asked Questions

**What can I do with the Haystack (deepset Cloud) MCP?**
You can manage your deepset Cloud RAG pipelines, view your workspaces, and query your indexed documents directly through your AI agent.

**Can I run actual searches on my indexed data using the Haystack (deepset Cloud) MCP?**
Yes. You can ask your agent to run a search on a specific pipeline, and it will return the relevant snippets and source documents from your index.

**How does the Haystack (deepset Cloud) MCP help with RAG development?**
It simplifies the 'ops' side of RAG by letting you audit pipelines, check file statuses, and test search results in a conversational interface.

**Can I see which documents are actually uploaded with the Haystack (deepset Cloud) MCP?**
Yes, you can ask your agent to list the files in any workspace to verify that your source documents have been successfully indexed.

**How do I manage different search environments using the Haystack (deepset Cloud) MCP?**
You can use the workspace tools to list and switch between different isolated environments for different projects or search contexts.

**Is the Haystack (deepset Cloud) MCP good for testing NLP topologies?**
It's great for that. You can quickly pull the details of any pipeline to see how your embedding nodes and retrievers are configured.

**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.