Summary
Log streams you can interrogate in SQL, from the same place you are debugging.
Listing streams and reading their schema is what makes the SQL usable — you can find the field names before you guess at them. The project describes itself as early and log-focused, and it ships no auth of its own, so the deployment shape matters as much as the tools.
What it is
A Go MCP server for Parseable, letting an agent list data streams, inspect them, and query them with SQL over either HTTP or stdio.
What you get
- The available Parseable data streams listed, so you can find the one you need without opening the UI
- SQL queries run against a stream, with responses returned as JSON in both text and structured form
- Schema, stats and info for any stream, which is what makes a first query on an unfamiliar stream possible
- Stateless tool calls — no session to keep alive
- Both transports in one binary: HTTP on /mcp by default, or stdio for CLI and agent-to-agent workflows
Requirements
A reachable Parseable instance, with its URL and a username and password supplied through the environment or the matching command-line options. Building from source needs a recent Go toolchain. The server itself has no authentication — put a reverse proxy such as nginx or envoy in front of it if it will be exposed.