---
title: Query And Seed PostgreSQL
url: "/docs/databases/postgres/sql/"
canonical_url: https://docs.blossom88.com/docs/databases/postgres/sql/
markdown_url: https://docs.blossom88.com/docs/databases/postgres/sql.md
summary: Create a sample table and inspect PostgreSQL data from your AI app or terminal
categories:
- data
- databases
- postgresql
---

# Query And Seed PostgreSQL

Create tables, add sample records and inspect your PostgreSQL data through the
Blossom CLI or a [connected AI app](/docs/ai-agents/connect-ai-apps/).
Choose Full access for SQL tasks, then select the database inside your instance.

    blossom database:list --database_instance database_instance-100

## Add Sample Data

> Create a test_contacts table with a name and city, add Alex in Seattle and
> Jamie in Austin, then show me the rows.

The agent submits one transaction. Save sql.yml:

```yaml
sql:
  statements:
    - CREATE TABLE public.test_contacts (id bigint GENERATED ALWAYS AS IDENTITY PRIMARY KEY, name text NOT NULL, city text NOT NULL)
    - INSERT INTO public.test_contacts (name, city) VALUES ('Alex', 'Seattle'), ('Jamie', 'Austin')
```

    blossom database:execute --database database-100 --request_key seed_contacts_001 --data sql.yml --follow
    blossom database:query --database database-100 --request_key show_contacts_001 --sql 'SELECT * FROM public.test_contacts ORDER BY id' --follow

Each request key identifies one execution. Reuse it with the same request if
your connection drops. `--follow` waits for results; `database:sql:show --id`
opens a Run later. Results remain available for one hour.

## Query Your Data

> Show me how many test contacts are in each city.

    blossom database:query --database database-100 --request_key contacts_by_city_001 --sql 'SELECT city, count(*) FROM public.test_contacts GROUP BY city' --follow

Queries use read-only transactions. Each request can contain up to 20 statements,
run for up to 30 seconds, and return up to 1,000 rows. Use `--max_rows` and
`--timeout_seconds` to choose limits.

In an AI app, the matching tools are `database_query`, `database_execute` and
`database_sql_show`. For a change to existing data, describe the exact change
and authorize it before the agent submits `yes: true`.

## What You Get

A populated sample table, query results in your conversation or terminal,
and a SQL Run you can inspect by its ID.