How to connect to PostgreSQL, MySQL, MongoDB and Redis in VS Code

Short answer: install a database client extension, fill in the connection form (or paste a connection string), test it, and browse your tables from the sidebar. This guide uses the free Database Client in DevSnip Pro, which supports PostgreSQL, CockroachDB, MySQL, MariaDB, SQL Server, Oracle, SQLite, ClickHouse, MongoDB and Redis from one panel.

During development most database work is small and frequent: check that a migration added the column, look at the row your API just wrote, fix a bad test record, or try a query before putting it in code. Doing it in the editor removes the trip to a separate application.

1. Connect with the form or a connection string

Run DevSnip Pro: Database Client and click +. In Form mode pick the database type and fill in host, port, user, password and database. Or switch to Connection string and paste one:

postgresql://app:secret@localhost:5432/app_dev
mysql://root:secret@localhost:3306/shop
mariadb://root:secret@localhost:3306/shop
Server=localhost,1433;Database=crm;User Id=sa;Password=…;TrustServerCertificate=true
oracle://hr:secret@localhost:1521/FREEPDB1
cockroachdb://root@localhost:26257/defaultdb?sslmode=disable
clickhouse://default:secret@localhost:8123/default
sqlite:///Users/me/project/dev.db
mongodb://localhost:27017/app
redis://localhost:6379/0

Every driver ships with the extension. Oracle uses Thin mode, so you don't need the Oracle Client. The Database Client reference has more examples, including TLS, client certificates, SIDs and JDBC URLs.

As you type, the panel shows what it detected: the database type, host, database, user, whether TLS is on, and warnings such as TLS being off for a remote host. Press Test, then Save & connect.

The connection string is stored in your operating system's keychain (VS Code SecretStorage). The password is never displayed again and never appears in error messages.

Connection problems? The message says why: connection refused, host not found, authentication failed or TLS. For a local server with a self-signed certificate add sslmode=no-verify (PostgreSQL), TrustServerCertificate=true (SQL Server) or tlsAllowInvalidCertificates=true (MongoDB). SQL Server needs a SQL login; Windows authentication is not supported.

2. Browse and edit data safely

• Browse: databases, schemas, tables, views, collections or Redis keys, with each table's columns, types, keys and defaults under Structure.
• Find rows: sort, search across columns, or add filter conditions such as status = active. In MongoDB use a filter document such as { plan: "pro", seats: { $gt: 5 } }; in Redis a key pattern such as session:*.
• Edit: the row editor knows column types, NULL and DEFAULT. MongoDB documents accept Extended JSON (ObjectId(), ISODate()). Redis keys get an editor per type (strings, hashes, lists, sets, sorted sets) with TTLs.

Safety nets:

• Deleting rows and destructive statements (DROP, TRUNCATE, DELETE/UPDATE without WHERE, deleteMany, FLUSHDB) ask first.
• Dropping a table asks you to type its name.
• Mark a production connection read-only: DevSnip Pro refuses writes. On PostgreSQL, CockroachDB, MySQL, MariaDB, Oracle, SQLite and ClickHouse the server session is read-only as well.
• Rows of tables without a primary key are edited through ctid (PostgreSQL), ROWID (Oracle) or the hidden rowid (CockroachDB, SQLite). On MySQL, MariaDB and SQL Server they can be inserted but not edited, rather than risk changing the wrong row. Views are read-only.

3. Run queries

The query console runs SQL (including Oracle PL/SQL blocks), db.collection.find({...}) calls, or Redis commands, with syntax highlighting and line numbers. Press Ctrl/Cmd+Enter to run the selection or the whole buffer.

• Format tidies SQL.
• A syntax error is underlined where the server found it.
• A slow query can be cancelled on the server.
• You can sort and filter the results, then copy them as JSON or CSV for a test fixture or a bug report.

Related tools in the Database section of the sidebar:

• SQL Formatter & Linter: consistent formatting, plus warnings about SELECT *, missing WHERE and similar mistakes.
• SQL Query Helper: parameterised SELECT, INSERT, UPDATE, UPSERT, DELETE and pagination queries.
• Database Connection Strings: builds or fixes PostgreSQL, MySQL, MongoDB and Redis URLs, with driver setup.
• SQL → MongoDB Query: translates a SQL query into its MongoDB equivalent.

A development loop that works

1. Keep one saved connection per environment, with a colour (green for local, red for production) and read-only on anything shared.
2. After running a migration, open the table's Structure to confirm it.
3. After calling an endpoint in the REST API Client, check the row it wrote.
4. Copy query results as JSON straight into a test fixture.

For heavy administration (users and roles, backups, query plans at scale) a dedicated database tool still goes further; for the everyday loop, staying in the editor is faster.