Weaviate MCP. Query vector data and manage schemas via chat.
Gemini Works with every AI agent you already use
…and any MCP-compatible client
Just plug in your AI agents and start using Vinkius.
Weaviate MCP Server lets your AI client interact directly with a vector database. You can run semantic searches against massive collections, check the health of your cluster nodes, and manage schemas—all through natural conversation.
It bypasses complex console UIs for data discovery and retrieval.
What your AI agents can do
Get class schema
Retrieves the specific property definitions for a single class (collection).
Get cluster nodes
Gets operational details and health status for all connected Weaviate cluster nodes.
Get full schema
Retrieves the complete schema definition, showing all collections available in the instance.
Send a query vector to find the most semantically similar data objects across designated collections.
Fetch either the complete schema for all collections or the specific property definitions for one class.
Retrieve a single, full data object record using its unique identifier (UUID), including internal metadata.
Query the server to get real-time status and resource usage for all connected Weaviate cluster nodes.
Browse and retrieve paginated lists of data objects within a specified class, showing property values.
Ask AI about this MCP
Supported MCP Clients
Gemini Waiting for input…
Weaviate MCP Server: 7 Tools for Data Indexing & Schema Management
Use these tools to audit your database structure, run semantic searches with vectors, or monitor the live operational status of your Weaviate cluster.
019d7620get class schema
Retrieves the specific property definitions for a single class (collection).
019d7620get cluster nodes
Gets operational details and health status for all connected Weaviate cluster nodes.
019d7620get full schema
Retrieves the complete schema definition, showing all collections available in the instance.
019d7620get instance metadata
Returns high-level operational information about the entire Weaviate server instance.
019d7620get object details
Retrieves all metadata and properties for one specific data object using its UUID.
019d7620list objects
Lists paginated objects within a specified class, showing their property values.
019d7620search near vector
Performs a nearest neighbor vector search by comparing an input query vector against indexed data.
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 Weaviate, 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
You connect your AI client straight into Weaviate using this server, giving it direct access to vector search and core administrative functions. Instead of wrestling with a complex console UI or writing specialized query languages, you just talk to the system.
Vector Search & Data Discovery: When you need content based on meaning rather than keywords, you run a nearest neighbor vector search using search_near_vector. You send an input query vector and it compares that against all indexed data across designated collections, pulling back the most semantically similar results. If you're just auditing what collections exist in your whole setup, use get_full_schema to retrieve the complete schema definition showing every available class.
For a deep dive into one specific area, get_class_schema pulls out the precise property definitions for a single collection.
Reading and Auditing Data: You can browse data in two ways. To list objects within a specified class, you use list_objects, which returns paginated results showing all the defined property values for those items. If you know exactly what you're looking for, you don't have to scroll; instead, you pass a unique identifier (UUID) to get_object_details and pull back every piece of metadata and properties associated with that single record.
This allows your agent to pinpoint any specific data object in the entire instance.
System Health & Administration: The server provides tools for checking its operational status, too. You can get high-level info about the whole Weaviate server using get_instance_metadata, which returns general operational details about the instance itself. If you need to check if the cluster is healthy or what resources it's eating up, get_cluster_nodes gives you the real-time status and detailed health metrics for every connected node in your cluster.
This entire setup lets your agent perform complex data actions—everything from semantic querying to auditing schema structure—all through natural conversation. You won’t need specialized database query skills; you'll just ask what you want.
How Weaviate MCP Works
- 1 Subscribe to the server and provide your Weaviate Base URL and API Key.
- 2 Instruct your AI agent (Claude, Cursor, etc.) to perform a task: e.g., 'List all classes in my schema.'
- 3 The agent invokes the necessary tool (
get_full_schema), processes the data response, and gives you the clean answer.
The bottom line is that your AI client handles the API calls to Weaviate for you; you just talk to it like a database admin.
Who Is Weaviate MCP For?
Data engineers who hate writing boilerplate JSON queries, and research teams needing immediate answers from vast knowledge bases. If your job involves figuring out what data exists or finding the most related document in a massive vector store, this is for you.
Testing semantic search queries and verifying that new vectors are ingesting correctly into specific classes.
Auditing the database schema (get_full_schema) or monitoring cluster health to debug data pipelines.
Quickly surfacing relevant documents and data points from massive, disparate vector collections through conversational chat.
What Changes When You Connect
- Semantic Search: Stop searching by keywords. Use
search_near_vectorto find content based on meaning, getting much better results for knowledge retrieval. - System Visibility: Need to know if the cluster is healthy? Call
get_cluster_nodes. You get immediate operational status and resource usage without logging into a dashboard. - Schema Discovery: Forget guessing what data fields exist. Use
get_full_schemato pull the entire database map—all collections, all properties—in one go. - Targeted Auditing: Find that single record? Pass its UUID to
get_object_details. You get a full dump of its metadata and internal state instantly. - Structured Browsing: Use
list_objectswhen you need to browse a collection like an index. It gives you paginated results for quick data exploration.
Real-World Use Cases
Debugging Missing Data
A developer suspects the 'Article' class is missing records, but they aren't sure if it was a write failure or just a bad query. They ask their agent to run list_objects against 'Article'. The agent confirms recent pagination limits are in effect and shows 50 existing items, solving the mystery without manual console queries.
Understanding System Boundaries
A new data engineer needs to know what collections exist. Instead of navigating through five different admin tabs, they ask their agent for the full schema using get_full_schema. The agent returns a clean list of every class ('Product', 'Article', 'UserFeedback'), letting them start their work immediately.
Finding Related Knowledge
A research team needs to find documents related to 'quantum computing ethics'. They provide an embedding vector and ask the agent to use search_near_vector. The agent returns the top 5 most relevant articles, complete with their similarity scores—a much faster result than keyword search.
Checking Infrastructure Status
The DevOps team gets an alert about slow performance. They use get_cluster_nodes via their agent to check the live resource usage across all nodes. The report shows Node-3 is hitting 90% CPU, pointing directly to the bottleneck.
The Tradeoffs
Blindly listing everything
Just calling list_objects repeatedly without knowing which class you're in and how many records exist. You waste time paginating through thousands of irrelevant results.
→
First, use get_full_schema to confirm the collection name (e.g., 'Product'). Then, run list_objects specifically on that class to see a manageable sample.
Ignoring object identifiers
Trying to find a record based only on its content or partial metadata. The system fails because it can't pinpoint the exact source.
→
If you know what you’re looking for, use get_object_details and provide the UUID. This is the most direct way to fetch an object.
Assuming everything works
Running a complex search query without checking system health first. You might waste quota or get misleading results because a node is actually overloaded.
→
Always start by calling get_cluster_nodes to verify the operational status and resource allocation before running high-volume queries.
When It Fits, When It Doesn't
Use this server if your primary need is semantic search or deep data auditing. You want an AI agent to manage complex database interactions that traditionally require writing boilerplate JSON calls, like search_near_vector or inspecting the full structure via get_full_schema.
Don't use it if you just need basic CRUD operations—if all you need is to create a record or update a simple field, stick to your standard API client. This server excels at reading and discovering. If you are constantly getting vague errors like 'Data not found,' run get_object_details with the UUID first; that's usually where the problem lies.
If performance is critical, remember that while list_objects lets you paginate, checking system limits using get_instance_metadata keeps your queries efficient.
Independent Platform Disclaimer: Vinkius is an independent platform and is not affiliated with, endorsed by, sponsored by, verified by, or otherwise authorized by Weaviate. 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 figuring out what data exists in a massive vector store is a nightmare.
Today, if you need to know the structure of your database—what collections exist and how they are built—you're stuck clicking through console tabs. You manually check schema definitions one by one, then copy-paste partial JSON snippets just to verify a field type or see if a collection was actually created.
With this MCP server, you simply ask: 'What is the full schema?' Your AI agent runs `get_full_schema` and spits out everything in readable markdown. No more clicking through tabs; you get the map instantly.
Weaviate MCP Server: Semantic Search with `search_near_vector`
The old way meant taking a document, running it through an embedding model locally, getting the vector array (e.g., `[0.12, -0.05...]`), and then constructing a massive query body just to run the search. This was tedious, error-prone JSON templating.
Now, you just ask: 'Find articles similar to this topic.' The agent handles packaging your vector and running the `search_near_vector` tool. You get relevant results with similarity scores—it's clean, direct, and immediately actionable.
Common Questions About Weaviate MCP
How do I see all collections in Weaviate using get_full_schema? +
Call get_full_schema directly through your agent. It returns a listing of every single class name (e.g., 'Article', 'Product') and their property structures, giving you the entire database map.
What is the difference between list_objects and get_object_details? +
list_objects shows a paginated list of items in a collection when you don't know the UUID. get_object_details requires the exact UUID to pull all data for one specific record.
Can I check if my cluster nodes are healthy with get_cluster_nodes? +
Yes, running get_cluster_nodes provides real-time operational metrics. You can verify CPU usage and RAM consumption across all connected nodes to spot bottlenecks.
What is the input format for search_near_vector? +
You must provide a class name and the query vector as a JSON array of floats (e.g., [[0.1, -0.2]]). The agent expects this structured data to perform the comparison.
When should I use get_class_schema to check a collection's structure? +
Use get_class_schema when you only need the definition for one specific class, which is much faster than running get_full_schema. It returns the property definitions and data types for that single collection, helping you validate your input before querying.
If I get an empty list from search_near_vector, does it mean there's a problem with my API connection? +
No. An empty result set means the database found no objects matching that vector similarity score based on your criteria. If you receive a technical error code instead of an empty array, then you should check your credentials or network access.
How does get_instance_metadata help me audit my overall Weaviate server setup? +
This tool pulls high-level configuration details. You can use it to verify the current running version, see which modules are enabled, and check general operational status without needing access to specific data collections.
What extra information does get_object_details provide beyond just listing object properties? +
It gives you a deep audit view of the object. You retrieve not only the stored property values but also internal metadata, UUIDs, and configuration details that help data engineers trace provenance or check compliance.
Can I perform a vector search using float arrays through the agent? +
Yes. The search_near_vector tool allows you to perform semantic searches by providing a query vector as a JSON array of floats. Your AI agent will return the most similar objects from your Weaviate collection.
How do I see the data structure of my Weaviate collections? +
You can use the get_full_schema tool to see all classes and properties defined in your instance, or get_class_schema if you want to focus on a specific collection's definition.
Is it possible to monitor cluster health via chat? +
Absolutely. Use the get_cluster_nodes tool to retrieve operational data for all nodes in your Weaviate cluster, including their current status and resource utilization metrics.
Multi-server workflows that include Weaviate MCP
Find Codebase Duplications Using MCP Servers
Your codebase has 4 different implementations of date formatting, 3 versions of the retry logic, and 2 competing validation libraries , but nobody knows because grep only finds exact matches and these duplicates are semantic
Improve RAG Search Quality Using MCP Servers
Your RAG retrieves 10 documents but the answer is in #7 , Cohere reranking moves it to #1 and accuracy jumps from 68% to 94% without changing a single embedding
MCP Servers for Self-Updating Research Bases
You spend 3 hours reading 40 articles to write one research brief , an AI agent with Firecrawl reads all 40 in 90 seconds, stores them semantically in Weaviate, and writes the brief in Notion with every source linked and every claim verified
Search Your Entire Codebase Using MCP Servers
Code indexed, patterns detected, architecture documented, onboarding guides generated , build a living knowledge base from your codebase
Use it with your favorite AI tools
Connect this server to Cursor, Claude, VS Code, and more.
More in this category
DigitalOcean
Deploy and manage cloud infrastructure with simple virtual servers, managed databases, and Kubernetes clusters built for developers.
Linear
Streamline issue tracking and project management via Linear — list teams, query issues, create comments and inspect cycles directly from any AI agent.
Vercel
Bring your Vercel deployment infrastructure into chat. Control project domains, trigger manual builds, and inspect deployment status natively.
You might also like
Cloud DNS (Anycast DNS Provider API)
Manage your Anycast DNS infrastructure — list zones, configure records, and automate domain management directly through AI.
Forj
Manage community members, groups, and activity via AI agents with Forj (formerly Mobilize).
Fieldly
Connect Fieldly to automate construction management — manage work items, bookings, and invoices directly from your AI agent.