A Model Context Protocol (MCP) server that provides SQLite database interaction capabilities. This server allows AI agents (like Gemini CLI, Claude, etc.) to query and modify SQLite databases.
- Read Data: Execute
SELECTqueries to retrieve data. - Write Data: Execute
INSERT,UPDATE,DELETE,CREATE, andDROPqueries. - Schema Inspection: List tables and describe table schemas.
- Logging: Automatically logs queries to daily files in the
log/directory.
This MCP server implements a Context Preservation Pattern designed for autonomous AI agents. Since agents lose context between sessions, this server enforces a self-documenting database structure.
The server encourages agents to maintain a "meta-table" called _architecture_notes.
- Architect Agents (who create tables) MUST insert notes explaining why a table exists and how it works.
- Worker Agents (who process data) MUST read these notes to understand the system rules without needing a huge prompt context.
This turns the SQLite database into a self-contained, self-documented artifact.
- Node.js: v14 or higher.
To use this server with your MCP client, you need to configure it to run the index.js script.
-
Clone the repository:
git clone https://github.com/u1pns/sqlite-mcp.git cd sqlite-mcp -
Install dependencies:
npm install
-
Configure Environment Variables:
Add this to your
mcpServersobject in the.gemini/settings.jsonfile:{ "mcpServers": { "sqlite-mcp": { "command": "node", "args": ["/absolute/path/to/sqlite-mcp/index.js"], "env": { "SQLITE_DB_PATH": "/absolute/path/to/your/database.sqlite" } } } }If
SQLITE_DB_PATHis not provided, it defaults todatabase.sqlitein thesqlite-mcpdirectory.
You can run the server directly from GitHub without cloning the repo.
{
"mcpServers": {
"sqlite-mcp": {
"command": "npx",
"args": ["-y", "github:u1pns/sqlite-mcp"],
"env": {
"SQLITE_DB_PATH": "/absolute/path/to/my/db.sqlite"
}
}
}
}Note: The -y flag is important to auto-accept the package installation.
Connect to a specific SQLite database file.
db_path(string): The absolute path to the SQLite database file.- Note: If a relative path is provided, the database will be created inside the MCP server's directory, not the user's current directory. Always prefer absolute paths.
- Feature: Automatically returns custom instructions and best practices (from
instructions.md) upon successful connection.
Execute a SELECT query.
query(string): The SQL query (must start with SELECT or PRAGMA).
Execute a write query.
query(string): The SQL query.
List all tables in the database.
Get schema info for a specific table.
table_name(string): Name of the table.
Returns the full CREATE TABLE statements for the entire database.
- Use this instead of guessing table structures. It provides the ground truth of the database schema.
Returns the content of instructions.md. Useful for re-reading best practices without reconnecting.
MIT License
Copyright (c) 2024 u1pns
Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the "Software"), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions:
The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software.
THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.