Connect Claude Code, Cursor and VS Code to SynapCores with MCP
SynapCores has an MCP server built into the engine at /v1/mcp, on the same port as the database. Coding agents connect to it directly over HTTP with an API key and get 15 tools: SQL, semantic search, Cypher graph queries, AutoML and LLM generation, all against your own data.
Install free, one command
curl -fsSL https://get.synapcores.com | SYNAPCORES_REF=docs-mcp shFree Community Edition. Linux (x86_64, aarch64) and macOS on Apple Silicon; Windows via WSL2. Docker and other options
You need a running engine (see the quickstart or Docker) and its admin password from the first-boot block.
Create an API key
MCP clients send a fixed header, so give them an API key rather than a session token. Log in once and create a key:
export SYNAPCORES_PASSWORD='your-admin-password'
TOKEN=$(curl -s localhost:8080/v1/auth/login -H 'Content-Type: application/json' \
-d "{\"username\":\"admin\",\"password\":\"$SYNAPCORES_PASSWORD\"}" | jq -r .access_token)
export SYNAPCORES_API_KEY=$(curl -s localhost:8080/v1/api-keys -H "Authorization: Bearer $TOKEN" \
-H 'Content-Type: application/json' -d '{"name":"coding-agents","permission":"FullAccess"}' | jq -r .raw_key)The key starts with aidb_ and is shown only once, so store it where your editor can read it (for example in your shell profile as SYNAPCORES_API_KEY). Use "permission":"ReadOnly" for an agent that should only read.
Check the MCP endpoint
The server speaks JSON-RPC over HTTP with Authorization: Bearer <key>. Initialize and list the tools with curl:
curl -s localhost:8080/v1/mcp -H "Authorization: Bearer $SYNAPCORES_API_KEY" \
-H 'Content-Type: application/json' \
-d '{"jsonrpc":"2.0","id":1,"method":"initialize","params":{"protocolVersion":"2024-11-05","capabilities":{},"clientInfo":{"name":"curl","version":"1.0"}}}' \
| jq -c .result.serverInfo
curl -s localhost:8080/v1/mcp -H "Authorization: Bearer $SYNAPCORES_API_KEY" \
-H 'Content-Type: application/json' \
-d '{"jsonrpc":"2.0","id":2,"method":"tools/list"}' | jq -r '.result.tools[].name'{"name":"synapcores-aidb-mcp","version":"2.5.0-ce"}
describe_table
execute
list_tables
query
sql_manual
validate_query
train_model
predict
list_models
describe_model
semantic_search
embed_text
graph_query
generate_text
agent_soulConnect Claude Code
claude mcp add --transport http synapcores http://localhost:8080/v1/mcp \
--header "Authorization: Bearer $SYNAPCORES_API_KEY"
claude mcp listChecking MCP server health…
synapcores: http://localhost:8080/v1/mcp (HTTP) - ✔ ConnectedClaude Code now lists the tools as mcp__synapcores__query, mcp__synapcores__semantic_search and so on. Add --scope user to the add command to use the server in every project.
Connect Cursor
Add the server to ~/.cursor/mcp.json (all projects) or .cursor/mcp.json in a project. Cursor reads the key from your environment:
{
"mcpServers": {
"synapcores": {
"url": "http://localhost:8080/v1/mcp",
"headers": { "Authorization": "Bearer ${env:SYNAPCORES_API_KEY}" }
}
}
}Connect VS Code (GitHub Copilot)
Add the server to ~/.copilot/mcp-config.json (all workspaces, also read by the Copilot CLI) or to .mcp.json at the root of a workspace. Copilot reads the key from your environment, so set SYNAPCORES_API_KEY before you start VS Code:
{
"mcpServers": {
"synapcores": {
"type": "http",
"url": "http://localhost:8080/v1/mcp",
"headers": { "Authorization": "Bearer ${SYNAPCORES_API_KEY}" },
"tools": ["*"]
}
}
}Try it
Ask your agent for something that needs the database, for example:
- “List the tables in SynapCores and count the rows in each.”
- “Create a
ticketstable with an embedding column, add five sample support tickets, then find the ticket closest in meaning to ‘refund never arrived’.” - “Train a model on my orders table to predict which customers churn, then show the top three features.”
To test a tool call without an editor, use the MCP Inspector CLI:
npx -y @modelcontextprotocol/inspector --cli http://localhost:8080/v1/mcp --transport http \
--header "Authorization: Bearer $SYNAPCORES_API_KEY" \
--method tools/call --tool-name query --tool-arg 'sql=SELECT 1 + 1 AS two'{
"content": [
{
"type": "text",
"text": "{\"columns\":[{\"name\":\"two\",\"type\":\"INTEGER\",\"nullable\":true}],\"rows\":[{\"two\":2}],\"row_count\":1,\"execution_time_ms\":28}"
}
]
}The 15 tools
| Tool | What it does |
|---|---|
list_tables | List the tables in the database |
describe_table | Columns, types and constraints of one table |
query | Run a SELECT and return rows |
execute | Run INSERT, UPDATE, DELETE or DDL (CREATE, ALTER, DROP) |
validate_query | Check SQL syntax without running it, optionally with the plan |
sql_manual | Look up SynapCores SQL syntax and examples |
semantic_search | Similarity search over a vector collection; embeds the query for you |
embed_text | Get the embedding vector for a piece of text |
graph_query | Run Cypher (MATCH, MERGE, CREATE, …) on the graph |
train_model | Train an AutoML model on a table column |
predict | Score rows with a trained model |
list_models | List trained AutoML models |
describe_model | Algorithm, metrics and features of a model |
generate_text | Generate text with the engine’s configured LLM |
agent_soul | Inspect the learned memory, episodes and runs of in-database agents |
Every tool call runs with the permissions of the key, inside its tenant, through the same auth layer as the REST API: a ReadOnly key can query but not write. Revoke a key at any time with DELETE /v1/api-keys/<id>.
Next steps
Full reference
- SQL reference: Every statement and function: EMBED, VECTOR, COSINE_SIMILARITY, GENERATE, Cypher, AutoML, agent memory.
- API reference: The REST API. Your engine also serves its own OpenAPI schema at /v1/openapi.json.
- System requirements: Supported platforms, memory and disk sizing.