Built an MCP server for data work with memory and rules. Use cases: - Engineers: query your data from Claude/Cursor, debug issues, build with analytics in dev flow (like [1] but with memory and observability built in) - Data teams: chat with your DB, define rules for how AI should query, share dashboards and analysis Works with Postgres, Snowflake, BigQuery, Redshift, and more. Any LLM. Swap or mix instantly What's different: - Memory – stores context, preferences, usage down to table/column level. Learns over time. - Rules – instructions, terms, guardrails with versioning. Git sync with dbt, markdown, code. - Observability – traces, plans, evals, feedback. See exactly what happened. Would love to receive feedback! |
No comments yet