Setup guide · Hosted MySQL
Connect PlanetScale, RDS, Cloud SQL or any hosted MySQL to Dataki
Dataki connects to MySQL directly, with a host, a port, a database, a user and a password. It connects from one fixed IP address, over TLS whenever the server offers it, so hosts that refuse unencrypted connections, like PlanetScale, TiDB and Azure, work as they are.
- Your MySQL host
- Direct connection over TLS
- Dataki
- Setup
- About 5 minutes.
- Connects from
34.89.253.13, one fixed address to allow.- Encryption
- TLS whenever the server offers it. The server certificate is not verified, and client certificates are not supported.
- What it reads
- Tables and views in the database you name.
01Before you start
- A database reachable from the internet, or one you can open to a single address.
- A user that can create users and grant privileges, or on PlanetScale, access to
Settings›Passwords.
02Set it up
3 steps, about 5 minutes
- 01
Create a read-only user
Run this with your database's name and a long password.
Run as an admin user CREATE USER 'dataki_reader'@'%' IDENTIFIED BY 'choose-a-long-password'; GRANT SELECT, SHOW VIEW ON your_database.* TO 'dataki_reader'@'%';- PlanetScale does not support
CREATE USER. Create a password with theRead-onlyrole underSettings›Passwordsinstead. - On TiDB Cloud, user names carry your cluster's prefix, e.g.
'4AbC1x.dataki_reader', in the statement and in Dataki.
- PlanetScale does not support
- 02
Let Dataki in
If your host filters connections by address, allow
34.89.253.13. It is the only address Dataki connects from.-
PlanetScale
Connection details
Connect, orSettings›Passwords. Port3306.Letting Dataki in
Open by default. A password can be limited to addresses with
Manage IP restrictions. -
Amazon RDS and Aurora MySQL
Connection details
Connectivity & securitytab. On Aurora, use the reader endpoint.Letting Dataki in
Public accesson, then a security-group inbound rule for TCP 3306 from34.89.253.13/32. -
Google Cloud SQL for MySQL
Connection details
Instance
Overviewfor the public IP.Letting Dataki in
Connections›Networking›Authorized networks: add34.89.253.13/32. -
Azure Database for MySQL
Connection details
The host is
<server>.mysql.database.azure.com.Letting Dataki in
Settings›Networking: a firewall rule for34.89.253.13. -
DigitalOcean
Connection details
Connection Details, public network. The port is25060.Letting Dataki in
Network Access›Trusted sources: add34.89.253.13. -
TiDB Cloud
Connection details
Connect, public endpoint. The port is4000.Letting Dataki in
Open by default. Restrict it under
Settings›Networking›Authorized Networks. -
MariaDB Cloud
Connection details
The service's
Connectpanel.Letting Dataki in
Closed by default: add
34.89.253.13to the allowlist underSettings›Secure Access.
-
- 03
Add the connection in Dataki
Open app.dataki.ai/connect and, under
Database Connections, chooseConnect MySQL. Fill inHost,Port,Database Name,UsernameandPassword, thenTest connectionandSave Connection.
03Check it
Make sure it is right
Two checks, run while logged in as dataki_reader.
Which tables will Dataki see?
This is the list Dataki gives the model.
SELECT table_name, table_type
FROM information_schema.tables
WHERE table_schema = DATABASE() AND table_type IN ('BASE TABLE', 'VIEW')
ORDER BY table_name; Is the user really read-only?
This must fail with an access-denied error.
CREATE TABLE dataki_write_check (id int); 04Worth knowing
What you will run into
- One database per connection
- Dataki reads the tables and views of the database named in the connection. Add a second connection for a second database.
- Servers Dataki cannot reach
- Users created with
REQUIRE X509orREQUIRE SUBJECT, which demand a client certificate, and servers on a private network only, such as Azure servers created with private access.
Checked against
- MySQL: GRANT
- PlanetScale: password roles
- Amazon RDS: requiring TLS for MySQL
- Cloud SQL: authorized networks
- TiDB Cloud: authorized networks
Last checked 15 September 2026.
Free while we are in beta
Connect Hosted MySQL. Ask it something.
Once the data is where Dataki can read it, the first answer is a question away, and anything worth keeping becomes a dashboard with a link that stays live.
- One data source
- Free tier. Connect a second on any paid plan.
- Read-only
- Every query runs read-only. Dataki cannot change your data.
- No card
- There is nothing to cancel if you stop.