Db Metadata Extractor Mcp

io.github.Optisol-BusinessPublished by GitHub account Optisol-Businessv0.1.6Updated Oct 24, 2026

Extract database metadata from PostgreSQL, Snowflake, SQL Server, BigQuery, and Oracle.

VerifiedSTDIODesktop onlyDatabasesDeveloper Tools

Overview

AI-generated overview

Local MCP server that lets an assistant extract schema metadata from PostgreSQL, Snowflake, SQL Server, BigQuery and Oracle into JSON files and query it.

What you can do
Extract full schema metadata (tables, columns, primary keys, indexes, constraints) from PostgreSQL, Snowflake, SQL Server, BigQuery or Oracle, optionally limited to a schema or table list, and save it as a local JSON file with summary statistics. Search and filter previously extracted metadata files by table or column name, with pagination for large schemas.
When to use it
Useful when you want an assistant to inventory or document a database schema, compare or browse table and column structures, or produce a local JSON snapshot of metadata without a cloud service.
Requirements
Runs locally as a Python package (Python 3.8+) or Docker image over stdio; per-database drivers such as psycopg2-binary, snowflake-connector-python, pyodbc/pymssql, google-cloud-bigquery or oracledb. Needs network access to the target database and a database user with SELECT, USAGE and CONNECT rights; BigQuery needs a base64-encoded service account key.
Before you install
Database credentials (username, password, Snowflake account/warehouse/role, BigQuery service account key) are passed as tool parameters or environment variables, so they appear in assistant conversations and config files. The database user should be read-only; extraction writes JSON files to a path you specify. The README also documents an HTTP transport option, which exposes the server beyond the local machine.

Installation

In SourceWeft

  1. Open Db Metadata Extractor Mcp in the dashboard and add it to a workspace.
  2. Enable the server for the chats that should use its tools.

Desktop only via STDIO. STDIO servers start a local process, so they need the SourceWeft desktop host.

Other MCP clients

Follow the launch instructions in the repository.

README

mcp-name: io.github.Optisol-Business/db-metadata-extractor-mcp

Database Metadata Extractor MCP Server

A Model Context Protocol (MCP) server that extracts and queries database schema metadata from PostgreSQL, Snowflake, SQL Server, BigQuery, and Oracle databases.

Features

  • ✅ Multi-database support: PostgreSQL, Snowflake, SQL Server (MSSQL), BigQuery, Oracle
  • ✅ Complete schema extraction: Tables, columns, primary keys, indexes, constraints
  • ✅ Local JSON output: Saves metadata directly to local folder (no cloud required)
  • ✅ Query interface: Search and filter metadata by table/column names
  • ✅ Pagination support: Browse large schemas efficiently
  • ✅ VS Code integration: Works with VS Code Agent Mode
  • ✅ CLI customizable: Transport options (stdio, HTTP)

Installation

From PyPI

bash
pip install db-metadata-extractor-mcp

From Source

bash
git clone https://github.com/Optisol-Business/db-metadata-extractor-mcp.gitcd db-metadata-extractor-mcppip install -e .

Quick Start

1. Start the MCP Server

bash
db-metadata-extractor-mcp

The server starts in stdio mode by default and listens for MCP client connections.

2. Configure in Claude Desktop

Add to ~/.config/Claude/claude_desktop_config.json (macOS/Linux) or %APPDATA%\Claude\claude_desktop_config.json (Windows):

json
{  "mcpServers": {    "db-metadata-extractor": {      "command": "db-metadata-extractor-mcp",      "args": [],      "env": {}    }  }}

Restart Claude Desktop.

3. Use in Claude

Tell Claude:

Extract metadata from my PostgreSQL database and save it to /tmp/output

Claude will use the server's tools to extract and query your database schema.

Tools

extract_metadata

Extracts complete schema metadata from a database.

Parameters:

  • db_type (required): postgresql, snowflake, sqlserver, bigquery, oracle
  • output_path (required): Local directory for JSON output
  • database_name: Database/schema name
  • host: Database host (not needed for BigQuery/Snowflake)
  • port: Database port
  • username: Database user
  • password: Database password
  • schema_name: Specific schema (optional)
  • tables: Array of table names to extract (optional)
  • account: Snowflake account ID
  • warehouse: Snowflake warehouse
  • role_name: Snowflake role
  • project_id: BigQuery project ID
  • service_account_key: BigQuery service account JSON (base64 encoded)

Returns:

  • File path where metadata was saved
  • Summary statistics (table count, column count, etc.)

query_metadata

Query previously extracted metadata.

Parameters:

  • filepath (required): Path to metadata JSON file
  • table_name: Filter by table name (substring match)
  • field_name: Filter by column name (substring match)
  • page: Page number (default: 1)
  • page_size: Results per page (default: 20)

Returns:

  • Paginated table results matching filters

Examples

PostgreSQL

bash
# Via Claude"Extract all tables from my dev PostgreSQL database at localhost:5432"

Parameters Claude will use:

json
{  "db_type": "postgresql",  "host": "localhost",  "port": 5432,  "database_name": "dev_db",  "username": "postgres",  "password": "your_password",  "output_path": "/tmp/db_metadata"}

Snowflake

bash
"Extract schema from Snowflake account XYZ123"

Parameters:

json
{  "db_type": "snowflake",  "account": "XYZ123",  "username": "your_user",  "password": "your_password",  "warehouse": "COMPUTE_WH",  "role_name": "ANALYST",  "database_name": "PRODUCTION",  "output_path": "C:/metadata"}

BigQuery

bash
"Extract metadata from BigQuery project my-project-123"

Parameters:

json
{  "db_type": "bigquery",  "project_id": "my-project-123",  "service_account_key": "base64_encoded_json_key",  "output_path": "/tmp/bq_metadata"}

Advanced Usage

Custom Transport

Start with HTTP transport:

bash
db-metadata-extractor-mcp --transport streamable-http --port 3000

Environment Variables

bash
# Set database credentials via envexport DB_HOST=localhostexport DB_USER=postgresexport DB_PASSWORD=secret
db-metadata-extractor-mcp

Output Format

The extracted metadata is saved as a JSON file with structure:

json
{  "source": {    "db_type": "postgresql",    "extracted_at": "2026-04-09T14:30:00",    "host": "localhost"  },  "schemas": [    {      "schema_name": "public",      "tables": [        {          "table_name": "users",          "columns": [            {              "column_name": "id",              "data_type": "int",              "is_nullable": false,              "is_primary_key": true            },            {              "column_name": "email",              "data_type": "varchar",              "is_nullable": false            }          ],          "indexes": [            {              "index_name": "users_email_idx",              "columns": ["email"]            }          ]        }      ]    }  ]}

Requirements

  • Python 3.8+
  • For PostgreSQL: psycopg2-binary
  • For Snowflake: snowflake-connector-python
  • For SQL Server: pyodbc, pymssql
  • For BigQuery: google-cloud-bigquery
  • For Oracle: oracledb

Troubleshooting

Connection Errors

Problem: "Unable to connect to database"

Solution: Verify credentials and network access:

bash
# Test PostgreSQL connectionpsql -h localhost -U postgres -c "SELECT 1"
# Test Snowflakesnowsql -a XYZ123 -u your_user

Permission Errors

Problem: "Access denied" or "insufficient permissions"

Solution: Ensure database user has:

  • SELECT on tables
  • USAGE on schemas
  • CONNECT on databases

Large Schema Timeouts

Problem: Extraction times out on large databases

Solution: Extract specific schema/tables:

json
{  "schema_name": "public",  "tables": ["users", "orders"]  // Specify subset}

License

MIT License - See LICENSE file

Contributing

Contributions welcome! Please:

  1. Fork the repository
  2. Create feature branch
  3. Submit pull request

Support

Links

Source: README.md at commit 34ce92c

Tools

0
SourceWeft has not started this package in its sandbox yet, so its tools are not known.

Version history

1
  1. v0.1.6LatestSep 16, 2026

Score

40% of the applicable checks met

How complete this listing is, computed from the registry entry, the endpoint and public sources. Not a rating of the server and not a statement that it is safe or working today: reachability is a dated check, and no installs or reviews go into the score.

  • Reached in the last 30 days25 pts

    Keep the endpoint answering the MCP handshake; SourceWeft probes it every 30 days.

  • Auth declared10 pts

    Declare the credential headers in the registry entry, or keep the endpoint answering the probe so its auth can be classified.

  • Tools listed15 pts

    Answer tools/list without a credential, or ship a package that starts in a sandbox without one.

  • DescriptionRequired15 pts
  • Logo10 pts
  • Documentation15 pts
  • License10 pts

    Declare a license in the repository or the registry entry.