PostgreSQL
App · 21 actions
Query and change data in PostgreSQL, including Amazon RDS, Cloud SQL, Supabase and Neon: run SQL with safe parameters; find, count, select, insert, upsert, update and delete rows without writing SQL; create and drop tables; call procedures; look up schemas, tables, columns and indexes; and start a workflow when a row is added or changed.
Actions
| Action | What it does |
|---|---|
| Call procedure | Call a stored procedure or function with arguments and get back the rows it returns. |
| Count rows | How many rows in a table match the given values, or all of its rows when none are given. |
| Create table | Create a table from a list of columns. |
| Delete rows | Remove the rows matching the given values. |
| Describe table | A table's or view's columns in order, with type, nullability and default, and its primary key. |
| Drop table | Delete a table and every row in it. |
| Find row | The first row matching the given values, or found: false when there is none, so a later step can branch on it. |
| Get server info | The server's product and version, the database and the user signed in. |
| Insert row | Add one row to a table and get it back as stored, with its generated id and defaults. |
| Insert rows | Add up to 1,000 rows in one go, all or nothing. |
| List databases | The databases on the server that accept connections (templates left out). |
| List indexes | A table's indexes with their columns in key order, and which is the primary key. |
| List procedures | The stored procedures and functions the user can see, optionally with each one's parameters. |
| List schemas | The schemas in the connected database. |
| List tables | The tables the user can see, in every schema or one, optionally with views. |
| List views | The views the user can see, in every schema or one. |
| Run query | Run a SQL query and return its rows. |
| Run statement | Run an INSERT, UPDATE, DELETE or DDL statement (such as CREATE TABLE) and get the number of rows it changed. |
| Select rows | Read rows from a table without writing SQL: choose columns, match on values, sort and page. |
| Update rows | Change columns on the rows matching the given values. |
| Upsert row | Update the row with the same key, or insert it if there is none, and get the row back as stored. |
Call procedure
Call a stored procedure or function with arguments and get back the rows it returns. A function's result comes back as rows; a procedure's OUT parameters as one row.
Inputs
- Procedure
proceduretextRequired - The procedure's or function's name, e.g. add_customer. Use 'List procedures' to find one.
- Schema
schematext - The procedure's schema. Leave empty for the default (public, or the user's search path).
- Arguments
argumentslist - The values to pass, in order, e.g. [42, "active"]; or by name, e.g. {"customer_id": 42}.
- Maximum rows
max_rowsnumber - Return at most this many rows from each result set.From 1 to 10000. Default:
1000.
Outputs
proceduretext- The procedure or function called
result_set_countnumber- How many sets of rows it returned
result_setslist- Every set of rows it returned, in order
rowslist- The first result set's rows
Example output
{
"procedure": "add_one",
"result_set_count": 1,
"result_sets": [
{
"columns": [
"result"
],
"row_count": 1,
"rows": [
{
"result": 42
}
],
"truncated": false
}
],
"rows": [
{
"result": 42
}
]
}Count rows
How many rows in a table match the given values, or all of its rows when none are given.
Inputs
- Table
tabletextRequired - The table's name, e.g. customers (or sales.customers to name the schema too). Names are case-sensitive; 'List tables' shows them.
- Schema
schematext - The table's schema. Leave empty for the default (public, or the user's search path).
- Match rows where
filtersobject - Column and value pairs every row must match, e.g. {"status": "active"}. A list matches any of its values; null matches a missing value (NULL), not empty text. To compare, give an operator: {"amount": {">=": 10}}, {"name": {"contains": "ada"}} or {"created_at": {"between": ["2026-01-01", "2026-02-01"]}}. Operators: =, !=, >, >=, <, <=, contains, not_contains, starts_with, ends_with, like, not_like, in, not_in, between, is_null. contains, starts_with and ends_with take the text as written and ignore case; like and not_like take a pattern, with % for any text and _ for one character.
Outputs
countnumber- Rows matching the filters
tabletext- The table, with its schema when one was given
Example output
{
"count": 42,
"table": "customers"
}Create table
Create a table from a list of columns. By default an existing table of that name is left alone.
Inputs
- Table
tabletextRequired - The table's name, e.g. customers (or sales.customers to name the schema too). Names are case-sensitive; 'List tables' shows them.
- Schema
schematext - The table's schema. Leave empty for the default (public, or the user's search path).
- Columns
columnslistRequired - One entry per column, written as name then type, e.g. 'id integer primary key auto_increment', 'email varchar(255) not null unique', 'amount numeric(10,2) default 0', 'created_at timestamptz default current_timestamp'. After the type: primary key, not null, unique, auto_increment, and default followed by a number, 'text', true, false, null or current_timestamp.
- Skip if it exists
if_not_existstrue/false - Leave an existing table alone instead of failing.Default:
true.
Outputs
createdtrue/false- False when the table already existed and was left alone
tabletext- The table, with its schema when one was given
Example output
{
"created": true,
"table": "customers"
}Delete rows
Remove the rows matching the given values. At least one 'Match rows where' entry is required.
Inputs
- Table
tabletextRequired - The table's name, e.g. customers (or sales.customers to name the schema too). Names are case-sensitive; 'List tables' shows them.
- Schema
schematext - The table's schema. Leave empty for the default (public, or the user's search path).
- Match rows where
filtersobjectRequired - Column and value pairs every row must match, e.g. {"id": 42}. Required, so the whole table can never be changed by accident. To compare, give an operator: {"amount": {">=": 10}}, {"name": {"contains": "ada"}} or {"created_at": {"between": ["2026-01-01", "2026-02-01"]}}. Operators: =, !=, >, >=, <, <=, contains, not_contains, starts_with, ends_with, like, not_like, in, not_in, between, is_null. contains, starts_with and ends_with take the text as written and ignore case; like and not_like take a pattern, with % for any text and _ for one character.
Outputs
rows_affectednumber- Rows the statement changed; null when the database doesn't say
tabletext- The table, with its schema when one was given
Example output
{
"rows_affected": 1,
"table": "customers"
}Describe table
A table's or view's columns in order, with type, nullability and default, and its primary key.
Inputs
- Table
tabletextRequired - The table's name, e.g. customers (or sales.customers to name the schema too). Names are case-sensitive; 'List tables' shows them.
- Schema
schematext - The table's schema. Leave empty for the default (public, or the user's search path).
Outputs
columnslist- The table's columns, in order
countnumber- How many columns the table has
primary_keylist- The primary key's columns, in order
tabletext- The table, with its schema when one was given
Example output
{
"columns": [
{
"character_maximum_length": null,
"column_default": "nextval('customers_id_seq'::regclass)",
"column_name": "id",
"column_type": "int4",
"data_type": "integer",
"is_nullable": false,
"is_primary_key": true,
"numeric_precision": 32,
"numeric_scale": 0,
"ordinal_position": 1
}
],
"count": 1,
"primary_key": [
"id"
],
"table": "customers"
}Drop table
Delete a table and every row in it. This cannot be undone.
Inputs
- Table
tabletextRequired - The table's name, e.g. customers (or sales.customers to name the schema too). Names are case-sensitive; 'List tables' shows them.
- Schema
schematext - The table's schema. Leave empty for the default (public, or the user's search path).
- Ignore if missing
if_existstrue/false - Succeed without doing anything if it is not there.Default:
true.
Outputs
droppedtrue/false- False when there was no such table
tabletext- The table, with its schema when one was given
Example output
{
"dropped": true,
"table": "customers"
}Find row
The first row matching the given values, or found: false when there is none, so a later step can branch on it. Without a sort column, which matching row comes back is up to the database.
Inputs
- Table
tabletextRequired - The table's name, e.g. customers (or sales.customers to name the schema too). Names are case-sensitive; 'List tables' shows them.
- Schema
schematext - The table's schema. Leave empty for the default (public, or the user's search path).
- Match rows where
filtersobject - Column and value pairs every row must match, e.g. {"status": "active"}. A list matches any of its values; null matches a missing value (NULL), not empty text. To compare, give an operator: {"amount": {">=": 10}}, {"name": {"contains": "ada"}} or {"created_at": {"between": ["2026-01-01", "2026-02-01"]}}. Operators: =, !=, >, >=, <, <=, contains, not_contains, starts_with, ends_with, like, not_like, in, not_in, between, is_null. contains, starts_with and ends_with take the text as written and ignore case; like and not_like take a pattern, with % for any text and _ for one character.
- Sort by column
order_bytext - Column to sort on. Leave empty for no order.
- Sort direction
sort_directiontext - Which way to sort, when a sort column is given.One of:
ASC(Ascending),DESC(Descending). Default:ASC.
Outputs
foundtrue/false- True when a row matched
rowobject- The first matching row, or null
tabletext- The table, with its schema when one was given
Example output
{
"found": true,
"row": {
"email": "ada@example.com",
"id": 1,
"name": "Ada Lovelace"
},
"table": "customers"
}Get server info
The server's product and version, the database and the user signed in. The quickest way to check the connection works.
Inputs
None.
Outputs
current_schematext- Where unqualified table names are looked up
databasetext- The database connected to
producttext- PostgreSQL, MySQL or MariaDB
server_timetext- The server's clock, ISO 8601
user_nametext- The user signed in
versiontext- The version number, such as 16.4
version_stringtext- The server's full version text
Example output
{
"current_schema": "public",
"database": "postgres",
"product": "PostgreSQL",
"server_time": "2026-10-16T09:30:00+00:00",
"user_name": "lodol",
"version": "16.4",
"version_string": "PostgreSQL 16.4 on x86_64-pc-linux-gnu"
}Insert row
Add one row to a table and get it back as stored, with its generated id and defaults.
Inputs
- Table
tabletextRequired - The table's name, e.g. customers (or sales.customers to name the schema too). Names are case-sensitive; 'List tables' shows them.
- Schema
schematext - The table's schema. Leave empty for the default (public, or the user's search path).
- Values
valuesobjectRequired - Column and value pairs, e.g. {"name": "Ada", "email": "ada@example.com"}. Leave out generated and default columns. An object or list is stored as JSON.
Outputs
rowobject- The inserted row as stored, with generated ids and defaults
rows_affectednumber- Rows the statement changed; null when the database doesn't say
tabletext- The table, with its schema when one was given
Example output
{
"row": {
"email": "ada@example.com",
"id": 43,
"name": "Ada Lovelace"
},
"rows_affected": 1,
"table": "customers"
}Insert rows
Add up to 1,000 rows in one go, all or nothing. A column missing from a row is stored as null.
Inputs
- Table
tabletextRequired - The table's name, e.g. customers (or sales.customers to name the schema too). Names are case-sensitive; 'List tables' shows them.
- Schema
schematext - The table's schema. Leave empty for the default (public, or the user's search path).
- Rows
rowslistRequired - A list of rows, each a column and value object, e.g. [{"name": "Ada"}, {"name": "Grace"}].
Outputs
rows_affectednumber- Rows the statement changed; null when the database doesn't say
tabletext- The table, with its schema when one was given
Example output
{
"rows_affected": 2,
"table": "customers"
}List databases
The databases on the server that accept connections (templates left out).
Inputs
None.
Outputs
countnumber- How many are listed
databaseslist- The databases the user can see, without the system's own
Example output
{
"count": 1,
"databases": [
{
"name": "postgres"
}
]
}List indexes
A table's indexes with their columns in key order, and which is the primary key.
Inputs
- Table
tabletextRequired - The table's name, e.g. customers (or sales.customers to name the schema too). Names are case-sensitive; 'List tables' shows them.
- Schema
schematext - The table's schema. Leave empty for the default (public, or the user's search path).
Outputs
countnumber- How many are listed
indexeslist- The table's indexes
tabletext- The table, with its schema when one was given
Example output
{
"count": 1,
"indexes": [
{
"columns": [
"id"
],
"is_primary": true,
"is_unique": true,
"method": "btree",
"name": "customers_pkey"
}
],
"table": "customers"
}List procedures
The stored procedures and functions the user can see, optionally with each one's parameters.
Inputs
- Schema
schematext - Only this schema. Leave empty for every schema.
- Include parameters
include_parameterstrue/false - Also list each one's parameters and types.Default:
false.
Outputs
countnumber- How many are listed
procedureslist- The procedures and functions, by schema and name
Example output
{
"count": 1,
"procedures": [
{
"parameters": [
{
"data_type": "text",
"mode": "IN",
"name": "customer_name",
"position": 1
}
],
"return_type": null,
"routine_name": "add_customer",
"routine_schema": "public",
"routine_type": "PROCEDURE"
}
]
}List schemas
The schemas in the connected database.
Inputs
- Include system schemas
include_systemtrue/false - Also list pg_catalog, information_schema and the other pg_ schemas.Default:
false.
Outputs
countnumber- How many are listed
schemaslist- The schemas; the system's own only when asked for
Example output
{
"count": 1,
"schemas": [
{
"schema_name": "public"
}
]
}List tables
The tables the user can see, in every schema or one, optionally with views.
Inputs
- Schema
schematext - Only this schema. Leave empty for every schema.
- Include views
include_viewstrue/false - Also list views.Default:
false.
Outputs
countnumber- How many are listed
tableslist- The tables, by schema and name
truncatedtrue/false- True when there were more than the limit and the rest were left out
Example output
{
"count": 1,
"tables": [
{
"table_name": "customers",
"table_schema": "public",
"table_type": "BASE TABLE"
}
],
"truncated": false
}List views
The views the user can see, in every schema or one.
Inputs
- Schema
schematext - Only this schema. Leave empty for every schema.
Outputs
countnumber- How many are listed
truncatedtrue/false- True when there were more than the limit and the rest were left out
viewslist- The views, by schema and name
Example output
{
"count": 1,
"truncated": false,
"views": [
{
"table_name": "active_customers",
"table_schema": "public",
"table_type": "VIEW"
}
]
}Run query
Run a SQL query and return its rows. Use %(name)s placeholders with 'Parameters' for any value that comes from data. The user's privileges decide what it can touch.
Inputs
- SQL query
querytextRequired - The SQL to run, e.g. SELECT * FROM customers WHERE status = %(status)s. Write a literal % as %% when using parameters.
- Parameters
parametersobject - Values for each %(name)s in the SQL, e.g. {"status": "active"} for WHERE status = %(status)s. A list is sent as an array, for = ANY(%(ids)s). Passing values here, not pasting them into the SQL, keeps a quote in the data from changing the statement.
- Maximum rows
max_rowsnumber - Return at most this many rows.From 1 to 10000. Default:
1000.
Outputs
columnslist- Column names, in order
row_countnumber- Rows returned
rowslist- The rows, each keyed by column
rows_affectednumber- Rows the statement changed; null when the database doesn't say
truncatedtrue/false- True when more rows existed than the limit allowed
Example output
{
"columns": [
"id",
"name",
"email"
],
"row_count": 1,
"rows": [
{
"email": "ada@example.com",
"id": 1,
"name": "Ada Lovelace"
}
],
"rows_affected": null,
"truncated": false
}Run statement
Run an INSERT, UPDATE, DELETE or DDL statement (such as CREATE TABLE) and get the number of rows it changed. Use %(name)s placeholders with 'Parameters' for values.
Inputs
- SQL statement
statementtextRequired - The SQL to run.
- Parameters
parametersobject - Values for each %(name)s in the SQL, e.g. {"status": "active"} for WHERE status = %(status)s. A list is sent as an array, for = ANY(%(ids)s). Passing values here, not pasting them into the SQL, keeps a quote in the data from changing the statement.
Outputs
rows_affectednumber- Rows the statement changed; null when the database doesn't say
Example output
{
"rows_affected": 1
}Select rows
Read rows from a table without writing SQL: choose columns, match on values, sort and page. A limit always applies.
Inputs
- Table
tabletextRequired - The table's name, e.g. customers (or sales.customers to name the schema too). Names are case-sensitive; 'List tables' shows them.
- Schema
schematext - The table's schema. Leave empty for the default (public, or the user's search path).
- Columns
columnslist - Columns to return. Leave empty for all.
- Match rows where
filtersobject - Column and value pairs every row must match, e.g. {"status": "active"}. A list matches any of its values; null matches a missing value (NULL), not empty text. To compare, give an operator: {"amount": {">=": 10}}, {"name": {"contains": "ada"}} or {"created_at": {"between": ["2026-01-01", "2026-02-01"]}}. Operators: =, !=, >, >=, <, <=, contains, not_contains, starts_with, ends_with, like, not_like, in, not_in, between, is_null. contains, starts_with and ends_with take the text as written and ignore case; like and not_like take a pattern, with % for any text and _ for one character.
- Sort by column
order_bytext - Column to sort on. Leave empty for no order.
- Sort direction
sort_directiontext - Which way to sort, when a sort column is given.One of:
ASC(Ascending),DESC(Descending). Default:ASC. - Maximum rows
limitnumber - Return at most this many rows.From 1 to 10000. Default:
100. - Skip rows
offsetnumber - Skip this many rows first, for paging. Set a sort column too, or the pages may overlap.From 0 to 100000000. Default:
0.
Outputs
columnslist- Column names, in order
row_countnumber- Rows returned
rowslist- The rows, each keyed by column
tabletext- The table, with its schema when one was given
truncatedtrue/false- True when more rows existed than the limit allowed
Example output
{
"columns": [
"id",
"name",
"email"
],
"row_count": 1,
"rows": [
{
"email": "ada@example.com",
"id": 1,
"name": "Ada Lovelace"
}
],
"table": "customers",
"truncated": false
}Update rows
Change columns on the rows matching the given values. At least one 'Match rows where' entry is required.
Inputs
- Table
tabletextRequired - The table's name, e.g. customers (or sales.customers to name the schema too). Names are case-sensitive; 'List tables' shows them.
- Schema
schematext - The table's schema. Leave empty for the default (public, or the user's search path).
- New values
valuesobjectRequired - Column and value pairs to set, e.g. {"status": "inactive"}.
- Match rows where
filtersobjectRequired - Column and value pairs every row must match, e.g. {"id": 42}. Required, so the whole table can never be changed by accident. To compare, give an operator: {"amount": {">=": 10}}, {"name": {"contains": "ada"}} or {"created_at": {"between": ["2026-01-01", "2026-02-01"]}}. Operators: =, !=, >, >=, <, <=, contains, not_contains, starts_with, ends_with, like, not_like, in, not_in, between, is_null. contains, starts_with and ends_with take the text as written and ignore case; like and not_like take a pattern, with % for any text and _ for one character.
Outputs
rows_affectednumber- Rows the statement changed; null when the database doesn't say
tabletext- The table, with its schema when one was given
Example output
{
"rows_affected": 1,
"table": "customers"
}Upsert row
Update the row with the same key, or insert it if there is none, and get the row back as stored. The key defaults to the table's primary key; any other key needs a unique constraint on those columns.
Inputs
- Table
tabletextRequired - The table's name, e.g. customers (or sales.customers to name the schema too). Names are case-sensitive; 'List tables' shows them.
- Schema
schematext - The table's schema. Leave empty for the default (public, or the user's search path).
- Values
valuesobjectRequired - Column and value pairs for the row, including its key, e.g. {"email": "ada@example.com", "name": "Ada"}.
- Match on columns
key_columnslist - The columns that identify the row. Leave empty to use the primary key.
Outputs
actiontext- INSERT or UPDATE; on MySQL, NONE when the row was already the same
key_columnslist- The columns the row was matched on
rowobject- The row as stored after the upsert
rows_affectednumber- Rows the statement changed; null when the database doesn't say
tabletext- The table, with its schema when one was given
Example output
{
"action": "UPDATE",
"key_columns": [
"email"
],
"row": {
"email": "ada@example.com",
"id": 43,
"name": "Ada"
},
"rows_affected": 1,
"table": "customers"
}