Enables an AI agent to run read-only SQL over local CSV, Parquet, and JSON files through MCP tools, with directory sandboxing, statement gating, timeouts, and query timeouts that keep access confined to permitted folders. Results are shaped for small context windows with row caps, truncation notices, and markdown/CSV/JSON output.
Enables AI agents to run SQL analytics directly on CSV files in place via MCP, avoiding imports and token-heavy data pasting. It provides read-only querying, schema inspection, and file listing while keeping data local.
Exposes local data profiling, PII classification, and plan-based cleansing against DuckDB, Postgres, or SQLMesh tables through four tools (list_tables, profile, propose_cleansing_plan, apply_cleansing_plan). All reasoning stays aggregate-only — format signatures, counts, and match rates rather than raw cell values — and cleaning plans execute only after validation against a fixed operation vocabulary and explicit confirmation.
Enables AI agents to ingest CSV/TSV files, profile and categorize tables, generate SQL DDL for multiple engines, and launch background data workflows whose completion and artifacts can be independently verified.
Zero-dependency MCP server and CLI for token-efficient inspection of local CSV/JSON/JSONL files, providing schema, samples, and paginated filtered queries to AI agents.