Master the AI-Native Database with Natural Language — the seekdb MCP Server

Posted on In Views: Word count in article: 1.9k Reading time ≈ 7 mins.
This article explains how to install and configure the seekdb MCP Server, covering both embedded and client/server deployment modes and the Stdio and SSE connection methods. Through a hands-on case of building a personal note knowledge base, it demonstrates how to use natural language to manage vector collections, perform semantic search, run hybrid search, and use AI memory.

Introduction

Imagine this: you just describe your needs in natural language, and the AI automatically performs the database operations for you — creating document collections, inserting data, running complex queries, and even building a complete knowledge base application. This isn’t the future; it’s something you can do right now.

The seekdb MCP Server is the bridge that makes this vision a reality. Built on the MCP (Model Context Protocol) proposed by Anthropic, it lets AI assistants interact directly with the seekdb database, turning “natural language” into “database operations.”

This article will get you started with the seekdb MCP Server, and through a hands-on case — building an AI application with natural language — let you experience firsthand the appeal of an AI-native database.

seekdb MCP Server

What Is the seekdb MCP Server?

seekdb is an AI-native search database that fuses relational data, vector data, full-text indexing, JSON, and GIS capabilities under a unified architecture, supporting hybrid search and in-database AI workflows.

The MCP Server is the “adapter” that connects AI tools to the database. Through the MCP protocol, AI tools such as Cursor, Claude Code, and Cline can directly access and operate the seekdb database.

Capabilities at a Glance

Capability Category Tool List Description
Vector collection management create_collection, query_collection, add_data_to_collection, etc. Create vector collections, semantic search, document management
Advanced search full_text_search, hybrid_search Full-text search, hybrid search (BM25 + vector)
AI functions ai_complete, ai_rerank, create_ai_model, etc. Call an LLM to generate text, rerank search results
AI memory system seekdb_memory_query, seekdb_memory_insert, etc. Persist memory across sessions so the AI “remembers” you
Data import/export import_csv_file_to_seekdb, export_csv_file_from_seekdb Convert between CSV files and database tables/vector collections

Installing the seekdb Database

Before using the seekdb MCP Server, you need to prepare the seekdb database first. seekdb offers two deployment modes:

Mode 1: Embedded Mode (Zero Configuration, Linux Only)

Embedded mode requires no separate installation of the seekdb database! When the seekdb MCP Server starts, it automatically initializes a local embedded database — ready to use out of the box.

Use cases: personal learning, rapid prototyping, running on edge devices.

⚠️ Note:
macOS and Windows users need to use “client / server mode,” which requires deploying the seekdb database first (Docker is recommended) and then configuring the connection parameters.

If you need to deploy seekdb in a test or production environment, you can choose one of the following methods:

Method 1: Install via yum (RPM systems)

1
2
3
4
5
6
7
8
9
10
11
12
13
14
# 1. Add the seekdb mirror source
sudo yum-config-manager --add-repo https://mirrors.aliyun.com/oceanbase/OceanBase.repo

# 2. Install seekdb and the client
sudo yum install seekdb obclient

# 3. Start seekdb
sudo systemctl start seekdb

# 4. Check the startup status (status "Service is ready" means startup succeeded)
sudo systemctl status seekdb

# 5. Test the connection
mysql -h127.0.0.1 -uroot -P2881 -A oceanbase

Method 2: Use Docker (Fastest)

1
2
3
4
5
6
# Start seekdb with a single command
sudo docker run -d -p 2881:2881 oceanbase/seekdb

# If the pull fails, you can use a backup mirror source:
# sudo docker run -d -p 2881:2881 quay.io/oceanbase/seekdb
# sudo docker run -d -p 2881:2881 ghcr.io/oceanbase/seekdb

System requirements:

  • CPU: at least 1 core
  • Memory: at least 2 GB available
  • Supported operating systems: CentOS 7/8, Ubuntu 20+, Debian 9+, Anolis OS 8, Kylin V10, and more

For more deployment methods, see the seekdb deployment documentation[1].

Installing the seekdb MCP Server

Install the uv Package Manager

1
2
# Install the uv package manager
curl -LsSf https://astral.sh/uv/install.sh | sh

Configuring the AI Tool Connection

Stdio Mode

Taking Cursor as an example, open Settings → Tools & MCP → New MCP Server in Cursor, and choose the configuration method according to your operating system:

Linux Users (Embedded Mode)

1
2
3
4
5
6
7
8
{
  "mcpServers": {
    "seekdb": {
      "command": "uvx",
      "args": ["seekdb-mcp-server"]
    }
  }
}

It’s that simple! Embedded mode requires no configuration at all — when the server starts, it automatically initializes a local seekdb database.

macOS / Windows Users (Server Mode)

macOS and Windows don’t support embedded mode, so you need to deploy the seekdb database first (Docker recommended), then configure the connection parameters:

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
{
  "mcpServers": {
    "seekdb": {
      "command": "uvx",
      "args": ["seekdb-mcp-server"],
      "env": {
        "SEEKDB_HOST": "127.0.0.1",
        "SEEKDB_PORT": "2881",
        "SEEKDB_USER": "",
        "SEEKDB_PASSWORD": "",
        "SEEKDB_DATABASE": "test"
      }
    }
  }
}

Parameter descriptions:

Parameter Description Default
SEEKDB_HOST seekdb server address 127.0.0.1
SEEKDB_PORT seekdb service port 2881
SEEKDB_USER Database username None
SEEKDB_PASSWORD Database password None
SEEKDB_DATABASE Database name None

SSE Mode

Linux Users (Embedded Mode)

Start the SSE server directly:

1
uvx seekdb-mcp-server --transport sse --port 6000

macOS / Windows Users (Server Mode)

First configure the environment variables, then start the server:

1
2
3
4
5
6
7
8
9
# Configure seekdb connection info
export SEEKDB_HOST=127.0.0.1
export SEEKDB_PORT=2881
export SEEKDB_USER=
export SEEKDB_PASSWORD=
export SEEKDB_DATABASE=test

# Start the SSE server
uvx seekdb-mcp-server --transport sse --port 6000

Then configure it on the client:

1
2
3
4
5
6
{
  "sse-seekdb": {
    "type": "sse",
    "url": "http://127.0.0.1:6000/sse"
  }
}

Hands-on Case: Building a Personal Note Knowledge Base Through AI Conversation

Now let’s go through a complete hands-on case to experience the power of the seekdb MCP Server. We’ll build a personal note knowledge base that can:

  • ✅ Store notes in natural language
  • ✅ Semantically search related content
  • ✅ Use hybrid search for precise targeting
  • ✅ Provide intelligent AI Q&A

The Scenario

Suppose you’re a technical professional who regularly records all kinds of study notes and work insights. Now you’d like to use AI to help manage these notes and quickly find what you need through natural language.

Step 1: Create the Note Collection

In the conversation box of an AI tool (such as Cursor), enter:

1
Create a knowledge base collection named "my_notes"

Example AI response:

I’ve successfully created a collection named “my_notes”:

  • Dimension: 384 (vector size)
  • Distance metric: L2 (Euclidean distance)

And just like that, a vector collection that supports semantic search is created!

Step 2: Add Note Content

Continue entering:

1
2
3
4
5
6
7
Add the following notes to my_notes:

1. "seekdb supports hybrid retrieval combining vector search and full-text search, leveraging both keyword matching and semantic understanding at once"
2. "The MCP protocol is an open protocol proposed by Anthropic in November 2024, letting LLMs interact with external tools"
3. "Python's asyncio module provides asynchronous programming capabilities, well suited to I/O-bound tasks"
4. "Docker containerization enables fast application deployment and environment isolation"
5. "RAG (Retrieval-Augmented Generation) is currently the most popular AI application architecture, combining retrieval and generation"

Example AI response:

Done! I’ve successfully added 5 notes to the “my_notes” collection:

ID Content
doc1 seekdb supports hybrid retrieval combining vector search and full-text search…
doc2 The MCP protocol is an open protocol proposed by Anthropic in November 2024…
doc3 Python’s asyncio module provides asynchronous programming capabilities…
doc4 Docker containerization enables fast application deployment…
doc5 RAG (Retrieval-Augmented Generation) is currently the most popular AI application architecture…

Now let’s search the notes using natural language:

1
Search my_notes for notes related to "AI application development"

Example AI response:

Here are the notes related to “AI application development,” sorted by relevance:

Rank Content Relevance
1 RAG (Retrieval-Augmented Generation) is currently the most popular AI application architecture… Highest
2 The MCP protocol is an open protocol proposed by Anthropic in November 2024… High
3 seekdb supports hybrid retrieval combining vector search and full-text search… Medium

Note: we searched for “AI application development,” but that exact phrase doesn’t appear in the notes. seekdb’s vector search can understand semantic similarity and automatically find related content!

Hybrid search combines keyword matching and semantic understanding for more precise results:

1
2
3
Run a hybrid search on my_notes:
- Full-text keyword: "Python"
- Semantic query: "asynchronous programming and concurrency handling"

Example AI response:

Rank Content
1 Python’s asyncio module provides asynchronous programming capabilities, well suited to I/O-bound tasks
2 Docker containerization enables fast application deployment and environment isolation

Hybrid search uses the RRF (Reciprocal Rank Fusion) algorithm to combine keyword-matching (BM25) and vector-similarity scores, producing more accurate rankings.

Step 5: Let the AI Remember Your Preferences

The seekdb MCP Server also provides an AI memory system that lets the AI remember your information across sessions:

1
Remember that I'm a Python developer who mainly uses seekdb to build vector search applications

Example AI response:

Noted! I’ll remember:

  • You’re a Python developer
  • You mainly use seekdb to build vector search applications

Next time you chat, you can simply ask:

1
Based on my technical background, recommend some learning resources

The AI will automatically retrieve your preferences and give personalized recommendations!

Features at a Glance

Vector Collection Management

Tool Function
create_collection Create a vector collection
list_collections List all collections
has_collection Check whether a collection exists
peek_collection Preview documents in a collection
add_data_to_collection Add documents (vectors generated automatically)
update_collection Update documents
delete_documents Delete documents
query_collection Vector similarity search
delete_collection Delete a collection
Tool Function
full_text_search Full-text search (keyword-based)
hybrid_search Hybrid search (combining full-text and vector search)

AI Model Tools

Tool Function
create_ai_model Register an AI model (embedding, text generation, or reranking)
create_ai_model_endpoint Create an endpoint connecting the model to an API service
drop_ai_model Remove a registered AI model
drop_ai_model_endpoint Remove an AI model endpoint
ai_complete Call an LLM for text generation
ai_rerank Use an AI model to rerank documents by relevance
get_registered_ai_models List all registered AI models
get_ai_model_endpoints List all AI model endpoints

AI Memory System

The seekdb MCP Server provides a powerful AI memory capability that lets the AI assistant remember information across sessions:

Tool Function
seekdb_memory_query Semantically search memories
seekdb_memory_insert Store a new memory
seekdb_memory_update Update a memory
seekdb_memory_delete Delete a memory

Use cases:

  • The AI remembers your tech-stack preferences (e.g., “I usually use Python”)
  • The AI remembers project information (e.g., “This project uses FastAPI”)
  • The AI remembers personal preferences (e.g., “I like a clean code style”)

Data Import/Export

Tool Function
import_csv_file_to_seekdb Import a CSV file
export_csv_file_from_seekdb Export data to CSV

SQL Operations

Tool Function
execute_sql Execute a SQL query
get_current_time Get the database’s current time

Exploring More Tools

Beyond the features covered in this article, the seekdb MCP Server also supports:

  • AI function calls
    • Use an AI model to analyze the sentiment of this text: “The weather is great today, and I’m in a wonderful mood!”
  • CSV data import
    • Import /path/to/products.csv as a vector collection, using column 2 (product description) as the document

FAQ

Q: Do I need to install seekdb?

A: No! The seekdb MCP Server uses embedded mode — seekdb is already included, with no separate installation required.

Q: Where is the data stored?

A: Data is stored on the local file system, by default under the current user’s home directory. Your data stays entirely local and is never uploaded to any cloud.

Q: Which operating systems are supported?

A: Currently Linux (glibc >= 2.28), supporting the x86_64 and aarch64 architectures.

Q: How do I upgrade?

A: When using uvx, the latest version is used automatically.

Conclusion

The seekdb MCP Server makes database operations easier than ever:

Traditional Way The MCP Way
Learn SQL syntax Describe your needs in natural language
Write code to call APIs The AI performs operations automatically
Manually manage vector embeddings Automatic generation and indexing
Handle search logic separately Hybrid search in a single sentence

Whether you want to quickly build a RAG application or give your AI assistant “long-term memory,” the seekdb MCP Server is your best choice.

Start your AI-native database journey! 🚀

References

[1] seekdb deployment documentation: https://www.oceanbase.ai/docs/deploy-overview/