Typesense Vector Search MCP. Run semantic searches & manage collections from 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.
Typesense Vector Search lets your AI agent handle complex vector operations right from chat. You can index documents, create new semantic collections, and run combined text-filtered similarity searches without writing a single API call or running CURL commands.
What your AI agents can do
Create collection
Creates a new search collection by taking the schema details as a JSON object you provide.
Delete document
Removes a document from a specific collection using its ID. This action cannot be undone.
Get collection details
Retrieves the full schema and metadata for any named collection in your instance.
You can create new semantic search collections and view details on existing ones using create_collection or get_collection_details.
The agent adds or updates records in any collection. You just provide the JSON data, and it handles the indexation via index_document.
Use list_vector_collections to get a list of every active search collection running in your Typesense instance.
Run combined text-filtered vector similarity searches using the search_vectors tool, passing both keywords and the required embedding query string.
Permanently remove a document from any collection with the delete_document tool. This action is irreversible.
Ask AI about this MCP
Supported MCP Clients
Gemini Waiting for input…
Typesense Vector Search: 6 Tools for Advanced Vector Indexing
These six tools give your agent the full power to manage the data lifecycle in a Typesense vector store, handling everything from collection setup to advanced semantic retrieval.
019d7617create collection
Creates a new search collection by taking the schema details as a JSON object you provide.
019d7617delete document
Removes a document from a specific collection using its ID. This action cannot be undone.
019d7617get collection details
Retrieves the full schema and metadata for any named collection in your instance.
019d7617index document
Adds or updates a document record within a specified search collection, given the name and JSON payload.
019d7617list vector collections
Lists all existing collections currently set up across your entire Typesense environment.
019d7617search vectors
Performs a combined vector similarity search and optional text filtering using the collection name, query string, and embedding vector.
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 Typesense Vector Search, 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
Typesense Vector Search MCP Server
Your AI client lets your agent run complex vector operations right from the chat window. You won't gotta write a bunch of API calls or mess around with CURL commands; it handles all the heavy lifting inside your Typesense environment.
You need to set up your data first. To start, you can use list_vector_collections to pull up a list of every search collection currently running in your Typesense setup. If you're building something new, you can call create_collection, which takes the schema details as a JSON object that you provide; it builds out the whole semantic structure for you.
If you need to check what's already there—like confirming the exact fields and metadata of an existing dataset—just run get_collection_details on any named collection. You’ll know exactly what schema geometry you're dealing with, which is critical before adding data or running a query.
Keeping your data current. When you have records ready to go, the agent uses index_document. You just hand over the JSON payload and specify the name of the search collection; it handles both adding brand-new documents and updating existing ones in that collection automatically. If you need to clear out some old junk, the delete_document tool lets your agent permanently remove a document from any specific collection using its ID.
Just know: this action is irreversible.
Running the search. The main event is running queries with search_vectors. This tool performs combined vector similarity searches while letting you apply optional text filtering—you pass both keywords and the required embedding query string, and it runs the whole thing for you using a specific collection name. It's one shot: you get your semantic results backed up by actual text matching.
How Typesense Vector Search MCP Works
- 1 First, subscribe to this MCP server and provide your Typesense Host URL and Admin API Key.
- 2 Next, prompt your agent with the goal: e.g., 'I need a new collection for product reviews.' The agent uses
create_collectionbased on your schema details. - 3 Finally, ask the agent to run a search or index data. It executes the necessary tool (
search_vectorsorindex_document) and returns the results directly in chat.
The bottom line is that you talk to your agent like a developer talks to their database—it just handles the API calls for you.
Who Is Typesense Vector Search MCP For?
This is for the data scientist who needs immediate, reliable vector search capabilities without writing backend boilerplate. It’s for the application builder stuck on manual data ingestion pipelines. If your job involves connecting unstructured knowledge to a searchable database, this toolset saves you hours of API work.
You prompt your agent to build semantic collections and automatically handle the embedding logic required for complex data types like float[].
You instruct the AI to ingest missing reference documents into a running collection, eliminating manual REST calls for data updates.
You perform sanity checks or run text-filtered semantic searches to verify exact relevance scores and collection schemas without writing test scripts.
What Changes When You Connect
- You run complex queries combining keywords and vectors using
search_vectors. You don't need to write separate code for filtering—it’s all in one prompt. - Setup is instant. Use
create_collectionto build a new semantic schema dataset just by providing the required JSON structure, skipping manual setup steps. - Data updates are simple. Instead of hitting a REST endpoint, you tell your agent to run
index_document, and it handles inserting or updating the whole JSON payload seamlessly. - No more debugging payloads. The
get_collection_detailstool instantly shows you the collection's schema geometry, so you always know which fields are available for querying. - You can keep your data clean. If a record is bad, running
delete_documentpermanently removes it by ID. It’s direct and efficient.
Real-World Use Cases
Onboarding a New Data Source
A developer finds a new set of technical manuals that need to be searchable but aren't indexed yet. They ask their agent: 'I need a collection for these manuals.' The agent uses create_collection, setting up the schema, and then runs index_document batch-by-batch until all files are ingested.
Finding Specific Product Info
A customer service agent needs to find a product that matches both 'wireless earbuds' (text filter) and is semantically close to 'best for gym use' (vector query). They prompt the agent, which executes search_vectors, returning highly precise results instantly.
Updating Old Records
An engineer updates a product description that was previously indexed incorrectly. Instead of deleting and re-uploading, they ask their agent to run index_document with the corrected JSON payload on the 'products' collection.
Schema Audit
A developer is unsure if a specific collection has an int32 field or a float[]. They ask their agent to run get_collection_details, which returns the exact schema geometry, preventing runtime errors.
The Tradeoffs
Trying to guess the right query syntax
Writing a complex search prompt that mixes up vector syntax with simple text filters, resulting in an ambiguous or malformed request for search_vectors.
→ Use your agent. Tell it: 'Search my collections for X using this embedding.' The tool handles the combined filtering and vector format automatically.
Running bulk operations manually
Manually looping through hundreds of records in code to decide which ones need updating, wasting time writing repetitive API logic.
→
Use list_vector_collections first to confirm the target collection name. Then, instruct your agent to run index_document for each record you want to update.
Relying on limited data visibility
Assuming a collection exists because it contains records, without checking its structure first.
→
Always check the schema before writing. Run get_collection_details to validate that your target collection has the required field types (e.g., confirming the presence of an embedding field).
When It Fits, When It Doesn't
Use this MCP Server if your core problem is connecting unstructured knowledge (documents, manuals, articles) to a searchable database via semantic vectors. The toolset excels at defining and managing the data lifecycle: you use create_collection for setup, index_document for writing, and search_vectors for reading.
Don't use this if your primary need is structured relationship querying (e.g., 'find all customers who bought Product A AND live in State B'). For those relational graphs, you should look at a dedicated graph database connector instead. This server assumes data can be flattened into JSON documents and vectorized; it doesn't manage complex one-to-many relationships inherently.
Independent Platform Disclaimer: Vinkius is an independent platform and is not affiliated with, endorsed by, sponsored by, verified by, or otherwise authorized by Typesense Vector Search. 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 6 capabilities that interface natively with Claude, ChatGPT, Cursor, and any MCP client. No middleware. No custom integration required.
Available Capabilities
Indexing knowledge shouldn't feel like writing a microservice.
Today, getting new documents searchable means multiple steps: first, you write code to connect to the database; then, you manually format the JSON payloads into chunks; after that, you run an API call just for indexing. It's tedious boilerplate—you spend more time managing the plumbing than actually adding knowledge.
With this MCP server, your agent handles all of it. You simply prompt: 'Make these 50 PDFs searchable.' The agent runs `create_collection` if needed, then executes `index_document` automatically for every file. You just get the result—a fully indexed dataset.
Typesense Vector Search MCP Server: Index documents & manage collections
The manual steps that disappear are writing schema validation logic, handling connection credentials in code, and manually triggering the index update sequence. You don't write a single function call for these core operations.
Now your agent manages the entire data flow autonomously. It’s not just an API wrapper; it’s a conversational control panel for your vector database.
Common Questions About Typesense Vector Search MCP
How do I start using `create_collection` with my own schema? +
You provide the full schema details as a JSON object to the agent. You'll need to specify data types and dimensions, especially for vector fields like float[].
What is the difference between `index_document` and `search_vectors`? +
index_document writes or updates data into a collection. search_vectors, conversely, reads that data by running a vector similarity search combined with text filters.
Can I delete documents using the `delete_document` tool if I don't know the ID? +
No. The delete_document tool requires you to provide the exact document ID for removal. You must retrieve this ID first, perhaps by running a search or getting collection details.
What does `get_collection_details` show me? +
It returns the complete schema and metadata for a specific collection name. This tells you exactly what fields exist and what data types they accept, which is key for writing correct queries.
What security requirements are needed for my AI agent to run `list_vector_collections`? +
The server requires an Admin API Key with read permissions across all collections. If the key lacks this scope, you'll get a 403 permission error when running any management tool like listing or creating collections.
If I try to run `index_document` with embedded data that doesn't match the schema dimensions, what happens? +
The operation fails immediately, returning a schema mismatch error. Before indexing, always verify your required vector dimension count using get_collection_details to prevent costly retries.
When I use `search_vectors`, does the complexity of my text filter impact performance? +
Yes, complex or highly specific text filters narrow down results faster. However, overly broad filters can slow queries because the system still performs large vector calculations across many candidates.
After using `list_vector_collections`, how do I confirm a collection is ready for new data? +
Running list_vector_collections only provides names. You must execute get_collection_details next to verify the schema structure and ensure the correct vector embedding type (float[]) is present.
Can the agent perform vector plus text-filtering search combined natively? +
Yes. Provide the agent with the collection name alongside the text payload and tell it the exact vector structure. It leverages internal filters querying natively and returns the nearest neighbors with exact accuracy scores.
How do I make the AI create a semantic collection ready for embeddings (OpenAI 1536 dims)? +
Ask the agent to use 'create_collection'. Provide standard JSON declaring the name, the field structure, and explicitly define the float[] field tracking the 1536 dims length. The cluster will spin the framework up instantly.
Can it delete problematic vectors holding bad geometry data manually? +
Absolutely. Supplying the explicit collection target and the item 'id' to the delete_document prompt securely wipes out all traces from the dataset. Use this sparingly as it can't be undone easily.
Use it with your favorite AI tools
Connect this server to Cursor, Claude, VS Code, and more.
More in this category
Postman
Design, test, and document APIs collaboratively with the world most popular API development platform used by millions of developers.
Typesense Cloud
Automate search cluster workflows via Typesense Cloud — monitor performance metrics, check cluster health, manage aliases, and execute multi-searches.
UUID & ULID Generator
Stop LLMs from hallucinating fake or repeated IDs. Generate mathematically guaranteed v4 UUIDs and time-sortable ULIDs natively.
You might also like
Mailify (Sarbacane)
Manage email and SMS marketing campaigns, address books, and contacts via the Mailify API.
Buildium
Manage rental properties via Buildium — track units, tenants, and maintenance tasks directly from any AI agent.
Saysimple
Automate omnichannel messaging via Saysimple — manage chats, send WhatsApp/SMS, and use templates with AI.