SynapCores
Self-hosted database with SQL, vector search, graphs and MCP access for AI apps
SynapCores is a free MCP server for databases. It runs locally on your machine and is started by your AI client.
- Pricing
- Free
- Platforms
- Linux, macOS, Windows
- Runs
- Locally
- Version
- 1.0.0
- Listing updated
Similar MCP servers
Install
Run in a terminal
claude mcp add synapcores --env SYNAPCORES_URL='<SYNAPCORES_URL>' --env SYNAPCORES_API_KEY='<SYNAPCORES_API_KEY>' -- npx -y @synapcores/mcpAdd to claude_desktop_config.json
{
"mcpServers": {
"synapcores": {
"command": "npx",
"args": [
"-y",
"@synapcores/mcp"
],
"env": {
"SYNAPCORES_URL": "<SYNAPCORES_URL>",
"SYNAPCORES_API_KEY": "<SYNAPCORES_API_KEY>"
}
}
}
}Add to ~/.cursor/mcp.json
{
"mcpServers": {
"synapcores": {
"command": "npx",
"args": [
"-y",
"@synapcores/mcp"
],
"env": {
"SYNAPCORES_URL": "<SYNAPCORES_URL>",
"SYNAPCORES_API_KEY": "<SYNAPCORES_API_KEY>"
}
}
}
}Add to .vscode/mcp.json
{
"servers": {
"synapcores": {
"type": "stdio",
"command": "npx",
"args": [
"-y",
"@synapcores/mcp"
],
"env": {
"SYNAPCORES_URL": "<SYNAPCORES_URL>",
"SYNAPCORES_API_KEY": "<SYNAPCORES_API_KEY>"
}
}
}
}Add to ~/.codeium/windsurf/mcp_config.json
{
"mcpServers": {
"synapcores": {
"command": "npx",
"args": [
"-y",
"@synapcores/mcp"
],
"env": {
"SYNAPCORES_URL": "<SYNAPCORES_URL>",
"SYNAPCORES_API_KEY": "<SYNAPCORES_API_KEY>"
}
}
}
}Add to your client's MCP settings (most use this shape)
{
"mcpServers": {
"synapcores": {
"command": "npx",
"args": [
"-y",
"@synapcores/mcp"
],
"env": {
"SYNAPCORES_URL": "<SYNAPCORES_URL>",
"SYNAPCORES_API_KEY": "<SYNAPCORES_API_KEY>"
}
}
}
}Using several servers? Build one config file with the MCP config generator.
Configuration
SYNAPCORES_URLSynapCores gateway URL, e.g. http://127.0.0.1:8080RequiredSYNAPCORES_API_KEYSynapCores API key (create with POST /v1/api-keys; use a ReadOnly key for read-only access)Required
About SynapCores
The README describes SynapCores as a single self-hosted engine that combines relational tables, vector similarity search, a Cypher graph, in-database AutoML and LLM text generation. Its MCP server runs inside the engine at /v1/mcp, so an assistant reaches the database through the same API key permissions that govern any other client.
The npm package @synapcores/mcp is a stdio connector that forwards JSON-RPC calls to that endpoint. Per the README it exposes 15 tools, including list_tables, query, execute, semantic_search, graph_query, train_model and predict. A ReadOnly API key blocks SQL writes, DDL and Cypher writes through those tools.
The product suits developers and data teams who want an AI assistant or agent to query, search and model data stored in their own infrastructure. The website also describes durable SQL-defined agents with memory and a hash-chained run ledger, though the MCP registry entry covers only the connector.
Tools this server provides (15)
Taken from the project documentation; the running server may differ.
- list_tables
- List the tables in the database
- describe_table
- Columns, types and constraints for a table
- query
- Run a SELECT and return rows
- execute
- Run INSERT / UPDATE / DELETE and DDL
- validate_query
- Check SQL syntax, optionally with the execution plan
- sql_manual
- SynapCores SQL syntax and examples
- semantic_search
- Similarity search over a vector collection
- embed_text
- Vector embedding for any text
- graph_query
- Cypher on the graph
- train_model
- Train an AutoML model on a table column
- predict
- Score rows with a trained model
- list_models
- List trained models
- describe_model
- Algorithm, metrics and feature importance of a model
- generate_text
- Generate text with the engine's configured LLM
- agent_soul
- Inspect learned runs, episodes, goals and plans (administrator key)
What it does
- Relational and vector queries
- Runs SQL over ACID tables alongside vector similarity search in the same engine.
- Cypher graph access
- Supports graph queries over the same data using Cypher, including gds algorithms.
- In-database AutoML
- Trains models on table columns and scores rows with predict from SQL-accessible tools.
- Read-only API keys
- Keys with ReadOnly permission block SQL writes, DDL and Cypher writes through the MCP tools.
- Stdio connector package
- The MIT-licensed npm package connects MCP clients over stdio with zero dependencies on Node.js 18+.
Good for
- Letting an assistant such as Claude Code query and describe tables in a local database
- Running semantic search over support articles or other text stored in vector collections
- Training and scoring a churn-style model on a table from an AI client
- Keeping AI access to sensitive data on self-hosted infrastructure
Pricing
Free
Community Edition is free with no signup; hosting and external AI provider costs are separate. Enterprise options are available.
Prices change. Check synapcores.com before you buy.
SynapCores FAQ
How is the MCP server connected?
The README says the stdio package @synapcores/mcp forwards calls over HTTP to the engine's /v1/mcp endpoint, and Claude Code can also connect to that endpoint directly over HTTP.
Which environment variables are required?
SYNAPCORES_URL and SYNAPCORES_API_KEY are marked required in the registry entry. SYNAPCORES_TIMEOUT_MS is optional with a default of 300000 ms.
Can the assistant write to the database?
By default, FullAccess keys allow writes. The README says a ReadOnly key makes the engine refuse SQL writes and DDL through execute, and Cypher writes through graph_query.
Is SynapCores open source?
The connector package is MIT-licensed. The website states that the Community Edition engine is proprietary software under its own license, not an open-source license.
Is SynapCores free?
Yes. SynapCores is free to use.
How do I add SynapCores to Claude Code?
Run this command in your terminal: claude mcp add synapcores --env SYNAPCORES_URL='<SYNAPCORES_URL>' --env SYNAPCORES_API_KEY='<SYNAPCORES_API_KEY>' -- npx -y @synapcores/mcp
What are the best alternatives to SynapCores?
The closest alternatives to SynapCores are stoolap, Vectros, MultiDB MCP, Mongodb MCP Server and ArcadeDB.
Where this comes from
- Description: written from the project's README, the project's website and its MCP Registry entry.
- Install steps: from the official MCP Registry entry.