# Oracle Vector DB MCP for AI Agents AI Agent Connect

> Oracle Vector DB MCP lets you run vector similarity searches and manage indexes directly in Oracle 23ai. Connect your AI agent to perform VECTOR_DISTANCE queries, inspect schemas, and execute SQL via ORDS. It's built for teams moving production RAG workloads into a robust enterprise database.

## Overview
- **Category:** industry-titans
- **Price:** Free
- **Endpoint:** https://edge.vinkius.com/vk_preview_Bs3fBgLmPqQPV1iPDfqOKsvG2g10fcqpkDZdhwp8/ai-agent-connect
- **Tags:** vector-database, sql, rag, machine-learning, database-administration, similarity-search

## Description

You're trying to move your production data into a vector store that actually scales with your business. Oracle 23ai gives you that enterprise-grade foundation, and this Connector makes it usable from your AI client. Instead of copy-pasting results between your database console and your chat window, you can just ask your agent to find the closest neighbors to a specific embedding. It handles the native similarity queries for you, whether you're using cosine or Euclidean metrics. You can also do the heavy lifting of schema inspection and index management without leaving your conversation. Because Vinkius handles the connection, you get a clean way to manage HNSW and IVF indexes while keeping your ML pipeline moving. It's about getting your production data into a RAG workflow without the friction of jumping between different apps.

## Tools

### describe_table
Get the full schema for a table, including data types. You can use this to find specific VECTOR columns.

### execute_sql_query
Run custom SQL queries against your Oracle database via ORDS. Keep your row counts low to ensure the AI stays stable.

### table_stats
Get row counts and optimizer statistics for your tables. It's great for planning your capacity and tuning performance.

### get_database_version
Check the exact Oracle DB runtime version banner. Use this to confirm your instance supports 23ai vector features.

### list_tables
See all the tables available in your current Oracle schema. This helps you quickly identify where your data lives.

### vector_search
Execute native VECTOR_DISTANCE queries to find similar neighbors. This works with both cosine and Euclidean metrics.

### list_vector_indexes
See a list of all HNSW and IVF vector indexes. Use this to check your index status across different tables.

## Prompt Examples

**Prompt:** 
```
Show me all tables in my schema that have VECTOR columns.
```

**Response:** 
```
I've checked your schema and found 3 tables with VECTOR columns:

| Table Name | Vector Column | Row Count |
| :--- | :--- | :--- |
| **DOCUMENTS** | EMBEDDING (1536) | 82,400 |
| **PRODUCTS** | FEATURES_VECTOR (768) | 5,200 |
| **FAQ_ENTRIES** | ANSWER_EMBEDDING (384) | 1,100 |

Would you like me to run a similarity search on one of these?
```

**Prompt:** 
```
Find the 5 most similar documents to this embedding using cosine distance.
```

**Response:** 
```
Searching the **DOCUMENTS** table using cosine distance... Here are the top 5 matches:

1. **Doc #4521** — 'API Rate Limits Overview' (Distance: 0.034)
2. **Doc #1892** — 'Authentication Flow Guide' (Distance: 0.089)
3. **Doc #7301** — 'Error Handling Reference' (Distance: 0.112)
4. **Doc #2244** — 'SDK Quickstart' (Distance: 0.145)
5. **Doc #9010** — 'Webhook Setup' (Distance: 0.178)
```

**Prompt:** 
```
What version of Oracle is running and does it support vectors?
```

**Response:** 
```
Your instance is running **Oracle Database 23ai (23.4.0.24.05)** on the Free tier. 

Here is the status of your vector features:
* **VECTOR Data Type:** Supported
* **VECTOR_DISTANCE Function:** Supported
* **HNSW Index Support:** Yes

Your database is fully ready for vector similarity search operations.
```

## Capabilities

### Run vector similarity searches
Perform native similarity queries against your vector columns to find the nearest neighbors.

### Inspect database schemas
See your table structures and identify which columns are set up for vector embeddings.

### Execute ad-hoc SQL queries
Run arbitrary SQL commands against your Oracle runtime for quick data analysis.

### Manage vector indexes
View and track all HNSW and IVF indexes across your database tables.

### View table statistics
Get row counts and optimizer stats to help with capacity planning and performance tuning.

### Verify database version
Check your Oracle runtime version to confirm it supports 23ai vector features.

## Use Cases

### Testing RAG Accuracy
An ML engineer asks the agent to find the 5 most similar documents to a new embedding using `vector_search` to verify their retrieval logic.

### Schema Auditing
A data engineer asks the agent to list all tables and describe them to find every column currently using the VECTOR data type.

### Performance Tuning
A DBA asks the agent to pull optimizer stats for a specific table using `table_stats` to see why a query is running slowly.

### Version Verification
A developer uses `get_database_version` to confirm the production instance is actually running Oracle 23ai before deploying a vector update.

## Benefits

- Stop switching tools by using `execute_sql_query` to run ad-hoc SQL analysis directly in your chat.
- Speed up RAG development with `vector_search` to test your embeddings against real Oracle data instantly.
- Simplify your schema audits by using `describe_table` to find and verify your vector-enabled columns.
- Get better performance insights by pulling optimizer stats with `table_stats` to find slow queries.
- Track your vector infrastructure more easily with `list_vector_indexes` to see all your HNSW and IVF setups.
- Confirm your environment setup quickly with `get_database_version` to make sure you're on the right runtime.

## How It Works

The bottom line is you get a direct line from your AI client to your Oracle 23ai vector store.

1. Subscribe to the Oracle Vector DB MCP via the Vinkius catalog.
2. Provide your Oracle ORDS URL, schema, and credentials.
3. Ask your AI client to search vectors, check schemas, or run SQL queries.

## Frequently Asked Questions

**Does the Oracle Vector DB MCP work with my existing Oracle database?**
Yes, it works with Oracle Database 23ai. It connects to your existing setup via Oracle ORDS, allowing your AI agent to interact with your current tables and vector columns.

**Can I use the Oracle Vector DB MCP for RAG applications?**
Absolutely. It's designed to help you build RAG pipelines by letting your agent perform native vector similarity searches and retrieve relevant context from your Oracle data.

**Do I need to move my data to a new database to use this?**
No, you don't have to move anything. This Connector lets you use your existing Oracle 23ai instance as your vector store, keeping your data centralized and secure.

**How does the Oracle Vector DB MCP handle similarity searches?**
It uses Oracle's native VECTOR_DISTANCE functions. You can ask your agent to find similar items using either cosine or Euclidean metrics directly on your vector columns.

**Can I use this Connector to manage my database indexes?**
Yes, you can use it to list and inspect HNSW and IVF vector indexes. This helps you keep track of your AI search infrastructure without leaving your AI client.

**Is it safe to run SQL queries through an AI agent?**
Yes, the Connector connects via Oracle ORDS. You can run ad-hoc queries for analysis, and the tool includes reminders to limit row counts to keep your AI agent's responses stable.

**Does it work with Oracle Autonomous Database?**
Yes. Oracle Autonomous Database on OCI has ORDS enabled by default. Use the ORDS URL from your ADB instance (e.g., https://xxxxx.adb.us-ashburn-1.oraclecloudapps.com/ords), your schema name (typically ADMIN), and the admin credentials. The VECTOR type is available on all 23ai-compatible instances.

**Can I run arbitrary SQL — not just vector searches?**
Yes. The execute_sql_query tool accepts any valid Oracle SQL statement and returns results through ORDS. Add FETCH FIRST N ROWS ONLY to keep payloads manageable. This makes the agent useful for relational queries too, not just vector operations.

**Which distance metrics are available for vector search?**
Oracle 23ai supports COSINE and EUCLIDEAN (L2) distance metrics natively via VECTOR_DISTANCE. Specify the metric when running a search — cosine is recommended for most text embedding use cases, while L2 works better for image or audio embeddings.