super

This Super MCP Server provides a Lambda-deployable data querying engine with an embedded SuperDB. It runs as a local MCP server for testing and as a Lambda-compatible service, and it exposes three core tools for interacting with data: query_data, list_files, and analyze_data. query_data executes SuperSQL queries against data files or URLs, enabling you to retrieve, filter, and transform datasets using a SQL-like syntax tailored for semi-structured data. list_files enumerates data files within directories, helping you discover available datasets. analyze_data inspects data structures to infer schema and type information, which is especially useful when working with JSON, CSV, Parquet, NDJSON, or other common formats. In local development, you can connect Claude Desktop to this server to run natural language questions that are translated into these tools under the hood. In Lambda, the embedded Super binary is extracted to /tmp on first invocation, enabling data queries without external dependencies.

Prerequisites:

• Go (required for building the server locally)
• Access to the repository with the Super binary script and setup script as described below

Installation steps:

1. Download and prepare the Super binary (local development):

chmod +x scripts/setup.sh
./scripts/setup.sh

2. Install Go module dependencies and build for local testing:

go mod tidy
go build -o super-mcp-server

3. Build for AWS Lambda (requires Go):

GOOS=linux GOARCH=amd64 go build -o bootstrap main.go
zip lambda-deployment.zip bootstrap

4. (Optional) Verify local server launch target:

./super-mcp-server

5. For Claude Desktop integration, configure the MCP server using the absolute path to the built binary (see Local Usage configuration).

Tips and considerations:

• The embedded Super binary is packaged with the Go executable and is extracted to /tmp on first Lambda invocation, so cold starts may require extra time for extraction.
• Data sources can be files on disk, URLs, or S3 paths (if Lambda has permissions). Ensure correct permissions when accessing remote data.
• For Claude Desktop integration, use absolute paths in the mcpServers configuration to avoid path resolution issues.
• If you encounter issues with the Super binary not found, verify that the setup script ran successfully and that the binaries/super-linux-amd64 file exists and is executable.
• When running Lambda, adjust timeout and memory settings to accommodate large queries or datasets (e.g., 30+ seconds and 512MB+ RAM).
• The server supports multiple data formats (JSON, CSV, Parquet, NDJSON, etc.). Use analyze_data to understand data shapes before crafting queries.