How manjotsingh20/azure-synapse-mcp works
This is an MCP server that exposes Azure Synapse SQL, pipeline, Spark, and notebook operations to compatible AI clients. The principal workflow is a client tool request, server-side dispatch, execution through the relevant Synapse interface, and a returned tool result. The sampled source shows SQL connection and query helpers plus pipeline and notebook calls; Spark submission is documented and partially sampled, so its internal client wiring is not expanded.
Main entry points: server.py MCP server
Components 13
MCP interface
- MCP server
server.py
SQL access
- SQL tools
server.py - SQL execution
server.py - Azure credentials
server.py - SQL connection
server.py - Synapse SQL
Workspace artifacts
- Artifacts client
server.py - Pipeline tools
server.py - Notebook tools
server.py - Synapse artifacts
Spark jobs
- Spark tools
server.py - Synapse Spark
Other components
- MCP client
Connections 17
- MCP client → MCP server: requests tools
- MCP server → SQL tools: dispatches SQL
- MCP server → Pipeline tools: dispatches pipelines
- MCP server → Notebook tools: dispatches notebooks
- MCP server → Spark tools: dispatches Spark
- SQL tools → SQL execution: executes SQL
- SQL execution → SQL connection: opens connection
- SQL connection → Azure credentials: gets token
- SQL connection → Synapse SQL: connects
- SQL execution → Synapse SQL: queries or writes
- MCP server → Azure credentials: gets credentials
- Pipeline tools → Artifacts client: calls pipeline API
- Notebook tools → Artifacts client: lists notebooks
- Artifacts client → Azure credentials: uses credentials
- Artifacts client → Synapse artifacts: calls workspace API
- Spark tools → Synapse Spark: submits or lists jobs
- MCP server → MCP client: returns results