cleo

A unified micro-harness + fine-tuned Qwen3.5-2B (a "hardel") for SQL-based analytical workflows that you point at your own database connections. Cleo's been trained on probing the data read-only to discover real values, codes, and conventions deep in the data, repair broken queries in-flight, and treat clarity & observability as first-class features.

No server, no pre-staging. Hand it a live DB-API connection and ask:

from cleo import Cleo
import psycopg2

cleo = Cleo.from_gguf()   # downloads + caches model (CPU-friendly); or Cleo.from_hf(...)
conn = psycopg2.connect("postgresql://...") # your existing connection: Postgres, SQLite, DuckDB, ...

ans = cleo.ask("How many employees are currently in each department?", conn)
print(ans.sql)            # the final read-only SELECT
print(ans.rows)           # executed result (rows), or None if Cleo asked to clarify
print(ans.clarification)  # set when the question is ambiguous / out-of-schema
print(ans.discovered)     # real values Cleo found while probing

conn is any DB-API 2.0 connection (psycopg2, sqlite3, duckdb, a SQLAlchemy engine.raw_connection()), or a callable executor(sql, limit) -> (columns, rows, truncated).

Safety

Every statement Cleo issues is validated read-only (single SELECT/WITH, AST-checked) and run in a rolled-back transaction. It never writes.

Big databases

Schema is introspected from the connection. Scope it so the prompt stays focused:

cleo.ask("...", conn, tables=["employees", "departments"])   # only these tables
cleo.ask("...", conn, schema=my_ddl_string)                  # or hand it the DDL yourself

CLI

cleo "total revenue by region" --db warehouse.duckdb
cleo "active users this week?" --db postgresql://me@host/db --tables users,sessions --json

As an MCP tool

cleo mcp --db "$DATABASE_URL" serves a natural-language query_database tool (pip install "cleo-sql[gguf,mcp]"). Claude Code / Claude Desktop config:

"cleo": { "command": "cleo", "args": ["mcp", "--db", "postgresql://..."] }

Install

pip install "cleo-sql[gguf]"       # llama-cpp-python backend (CPU/Mac/CUDA)
pip install "cleo-sql[hf]"         # transformers backend (GPU)
pip install "cleo-sql[gguf,mcp]"   # MCP server extras
# Weights download + cache themselves on first use: Cleo.from_gguf() pulls the current champion
# Q8_0 GGUF (bf16 parity) from HF. If the repo is private, run `hf auth login` first.

Run the tests (no model/GPU needed): pip install -e ".[test]" && pytest

API

• Cleo.from_gguf(path=None, *, n_ctx=4096, n_threads=8, n_gpu_layers=0): no path downloads/caches the current champion
• Cleo.from_hf(model="dreeseaw/cleo", *, device=None)
• cleo.ask(question, conn, *, schema=None, tables=None, max_gather=3, max_repair=2, execute_final=True, row_limit=1000, dialect=None, schema_fks=False, schema_samples=0) -> Answer: auto-detects dialect, transpiles from DuckDB SQL, and retries failed finals up to max_repair.
• Answer(sql, rows, columns, clarification, gathers, discovered, error): truthy when answered.