# Update

[back to query](README.md) /// [home](../README.md)

Modify existing rows in a table.

## Request

```json
{
  "action": "update",
  "table": "items",
  "data": {
    "price": 29.99,
    "status": "active"
  },
  "filter": {
    "column": "id",
    "op": "=",
    "value": 42
  }
}
```

## Fields

| Field    | Type   | Required | Notes                        |
|----------|--------|----------|------------------------------|
| `action` | string | yes      | must be `"update"`           |
| `table`  | string | yes      | target table                 |
| `data`   | object | yes      | column value pairs to update |
| `where`  | object | yes*     | simple key value filter      |
| `filter` | object | yes*     | structured filter            |

*at least one of `where` or `filter` is required. you cant do an unfiltered update.

## Response

```json
{
  "success": true,
  "rows_affected": 1
}
```

## Behaviors

- a WHERE clause is **mandatory**. the server refuses to run an update without one. this is a safety thingy yk
- columns you dont have write access to are silently stripped. `warning` is included if json derulo stripped for you
- `last_modified_by` auto set to your user ID
- `last_modified_at` is handled by sqls ON UPDATE trigger
- ownership scoping applies: if you have `rwo` you can only update rows where `pinned_to` equals your user ID uness you have rwa. same deal with `rwg` for group scoping
- limit is applied if set and capped to your max
- auto generation rules apply on update too if configured for the update action