MYSQL
Talk to your MySQL in plain English
Self-hosted text-to-SQL for MySQL. Read-only, governed, runs in your infrastructure.
Connect in one step
Ask in plain English
Real MySQL queries Limerence generates from business questions.
“Monthly active users for the last 6 months”
SELECT DATE_FORMAT(occurred_at, '%Y-%m') AS month,
COUNT(DISTINCT user_id) AS mau
FROM sessions
WHERE occurred_at >= DATE_SUB(CURDATE(), INTERVAL 6 MONTH)
GROUP BY month
ORDER BY month;“Pull notification preferences out of the JSON settings column”
SELECT u.id, u.email,
JSON_EXTRACT(u.settings, '$.notifications.email') AS email_pref,
JSON_EXTRACT(u.settings, '$.notifications.digest_frequency') AS digest_freq
FROM users u
WHERE JSON_EXTRACT(u.settings, '$.notifications.email') = CAST('true' AS JSON)
LIMIT 50;“Revenue rollup by region and product, with subtotals”
SELECT region, product, SUM(amount_cents) / 100.0 AS revenue
FROM orders
WHERE created_at >= DATE_SUB(CURDATE(), INTERVAL 90 DAY)
GROUP BY region, product WITH ROLLUP
ORDER BY region, product;How it works with MySQL
- Connects via the mysql2 driver with TLS support.
- Schema discovery uses information_schema.
- Compatible with MySQL 5.7, 8.0, and 8.4.
- JSON columns are queryable — generated SQL uses JSON_EXTRACT and JSON_TABLE.
- Window functions detected and used when the server reports MySQL 8.0+.
MySQL questions
Does this work with Amazon Aurora, Cloud SQL, or PlanetScale?+
Yes. Aurora MySQL and Cloud SQL for MySQL behave like vanilla MySQL. PlanetScale works on a read-only branch using a standard MySQL connection.
Does it understand window functions?+
On MySQL 8.0+, yes. On 5.7 the agent generates equivalent self-join or subquery patterns.
What about generated columns?+
Discovered from information_schema and queryable like any other column.
Will it handle utf8mb4 and emoji safely?+
Yes. The driver negotiates utf8mb4 and the agent treats text columns as Unicode.
Only your schema structure (table and column names, types) is shared with the LLM. Row data is never sent. Generated SQL runs against your database directly and results stay inside your network.