Base & Tables Tools

Name for the new database/base

Optional description of the database

Icon name (e.g., "database", "folder")

Hex color code (e.g., "#3B82F6")

Database ID from listDatabases

Database ID from listDatabases

New name for the database

New description

New icon name

New hex color code

Database ID to delete. WARNING: This permanently deletes all tables and data.

Database ID from listDatabases where the table will be created

Name for the new table

Optional description of the table

Icon name (e.g., "table", "users")

Hex color code (e.g., "#10B981")

Limit to one base — omit for all tables across all bases

Table ID from listTables

Table ID from listTables

Table ID from listTables

Table ID from listTables

New name for the table

New description

New icon name

New hex color code

Table ID to delete. WARNING: This permanently deletes all rows and fields.

Table ID from listTables

Name for the new field/column

Field type: "text", "number", "select", "multiselect", "date", "datetime", "checkbox", "url", "email", "phone", "file", "formula", "rating", "currency", "percent", "duration", "status", "button", "link", "lookup", "rollup". READ-ONLY computed types (writes to them are dropped): "hash" = Hash ID, a read-only stable per-row identifier — it does NOT hash a value you supply (to store a hashed value, compute it yourself and use text/secret); "created_time"/"updated_time" = system timestamps; "formula"/"lookup"/"rollup" = computed. Numbers are exact only up to 2^53 — unsafe integers are rejected with guidance; store long IDs, account numbers, and IMEIs as text.

Optional description of the field

Must be false or omitted. Required fields are not supported - UI creates blank rows first.

Whether values must be unique (default: false)

Default value for new rows

For select/multiselect: [{value: "opt1", label: "Option 1", color: "#3b82f6"}]

For status: [{value: "todo", label: "To Do", color: "#6b7280"}]

For formula: {formula: "{Field A} * {Field B}", resultType: text|number|date|checkbox}. Non-volatile formula values are materialized (stored on the row, auto-recalculated when inputs change — including via links) and are filterable/queryable everywhere; formulas using NOW()/TODAY() are volatile: display-only, NULL in SQL, not filterable

For button: {actionType: "flow", flowId: "xxx"}

For number: {decimalPlaces: 2, useThousandSeparator: true}

For currency: {currencySymbol: "$", symbolPosition: "before"}

For percent: {decimalPlaces: 1}

For lookup: {relationshipFieldId: "<link field id>", sourceFieldId: "<field id on linked table>", aggregation?: CONCATENATE|FIRST|LAST|ARRAY|COUNT|SUM|AVG|MIN|MAX (default CONCATENATE; COUNT counts NON-EMPTY VALUES — use a rollup COUNT to count linked records), filter?: {logic, conditions:[{fieldId,operator,value}]} over LINKED-table fields — only matching linked rows are aggregated}. Computed by Postgres; scalar (non-chained) aggregations are filterable everywhere — view filters, tables_query SQL, filterRows, and published-app entities.filter()

For rollup: {relationshipFieldId: "<link field id>", sourceFieldId: "<field id on linked table>", aggregation: COUNT|COUNT_EMPTY|COUNT_FILLED|COUNT_UNIQUE|SUM|AVG|MIN|MAX|MEDIAN|CONCATENATE|ARRAY|ARRAY_UNIQUE|ARRAY_COMPACT (default COUNT; COUNT counts LINKED RECORDS incl. empty values — COUNT_FILLED matches lookup COUNT), filter?: {logic, conditions:[{fieldId,operator,value}]} over LINKED-table fields — only matching linked rows are aggregated}. Computed by Postgres; scalar (non-chained) aggregations are filterable everywhere — view filters, tables_query SQL, filterRows, and published-app entities.filter()

For link field: target table ID

Link relationship type

Fractional order string (e.g., "a0", "b0")

Field ID from getFieldsByTableId

New name for the field

Field type

New description

Must be false or omitted. Required fields are not supported - UI creates blank rows first.

Whether values must be unique

Default value for new rows

For select/multiselect: [{value: "opt1", label: "Option 1", color: "#3b82f6"}]

For status: [{value: "todo", label: "To Do", color: "#6b7280"}]

For formula: {formula: "{Field A} * {Field B}", resultType: text|number|date|checkbox}. Non-volatile formula values are materialized (stored on the row, auto-recalculated when inputs change — including via links) and are filterable/queryable everywhere; formulas using NOW()/TODAY() are volatile: display-only, NULL in SQL, not filterable

For link field: target table ID

Link relationship type

How to display linked records

Which field to show from linked records

Whether multiple links are allowed

For link fields: what happens to linking rows when a linked record is deleted. RESTRICT blocks the delete; CASCADE deletes the linking rows too (chains, capped at 500 rows per delete); SET_NULL (default) just removes the connection.

For button: {actionType: "flow", flowId: "xxx"}

For number: {decimalPlaces: 2, useThousandSeparator: true}

For currency: {currencySymbol: "$", symbolPosition: "before"}

For percent: {decimalPlaces: 1}

For lookup: {relationshipFieldId: "<link field id>", sourceFieldId: "<field id on linked table>", aggregation?: CONCATENATE|FIRST|LAST|ARRAY|COUNT|SUM|AVG|MIN|MAX (default CONCATENATE; COUNT counts NON-EMPTY VALUES — use a rollup COUNT to count linked records), filter?: {logic, conditions:[{fieldId,operator,value}]} over LINKED-table fields — only matching linked rows are aggregated}. Computed by Postgres; scalar (non-chained) aggregations are filterable everywhere — view filters, tables_query SQL, filterRows, and published-app entities.filter()

For rollup: {relationshipFieldId: "<link field id>", sourceFieldId: "<field id on linked table>", aggregation: COUNT|COUNT_EMPTY|COUNT_FILLED|COUNT_UNIQUE|SUM|AVG|MIN|MAX|MEDIAN|CONCATENATE|ARRAY|ARRAY_UNIQUE|ARRAY_COMPACT (default COUNT; COUNT counts LINKED RECORDS incl. empty values — COUNT_FILLED matches lookup COUNT), filter?: {logic, conditions:[{fieldId,operator,value}]} over LINKED-table fields — only matching linked rows are aggregated}. Computed by Postgres; scalar (non-chained) aggregations are filterable everywhere — view filters, tables_query SQL, filterRows, and published-app entities.filter()

Display order position

Whether field is visible

Field ID to delete. WARNING: This permanently deletes all data in this column.

Table ID from listTables

Object where keys are FIELD IDs from getFieldsByTableId, values are cell values. Example: {"fieldId1": "text", "fieldId2": 123}

Fractional order string for positioning. Leave empty to append at end.

Table ID from listTables

Table ID from listTables

Rows to return, 1-1000 (default: 100). Values outside the range are REJECTED, not clamped — page with offset for more

Number of rows to skip for pagination

Base (database) ID from listDatabases

ONE read-only SELECT over this base's tables BY NAME, e.g. SELECT c."Name", SUM(o."Amount") FROM "Customers" c JOIN "Orders" o ON o.id = ANY(...) ... Columns are field NAMES; every table also has id, "createdAt", "updatedAt". Computed lookup/rollup columns are queryable. FORMULA columns are queryable too (values are materialized on write) — EXCEPT volatile formulas using NOW()/TODAY(), which read as NULL and are listed in the result's unavailableColumns. Use $1..$n with params.

Positional parameters for $1..$n placeholders

Row ID from getRows

Object with field IDs as keys and new values. Only provided fields are updated.

Optional metadata to merge with existing

Row ID to delete

Row ID to reorder

New fractional order string

Table ID from listTables

Array of row updates

Table ID to export

Table ID from listTables

Array of row data. When matchFieldsByName=true, use field NAMES as keys. When false, use field IDs. LINKED-RECORD columns resolve per value: sibling "<Field> (id)" column → row IDs → JSON-array string (comma-safe) → exact label → unique case-insensitive label; unresolvable values are returned in unresolvedLinks (row still imports, link skipped — never guessed, never auto-created). Relations are written both directions.

If true, match columns by field name. If false, use field IDs as keys (default: true)

Optional field definitions to create new fields during import

Table ID to filter

Filter object. Keys are field IDs OR field names (case-insensitive). Multiple keys are AND-ed; { "$or": [ {...}, {...} ] } and { "$and": [...] } group full filter objects and nest arbitrarily. Each value is either a plain primitive (equality), or an operator dict. Operators: $eq, $ne, $gt, $gte, $lt, $lte, $between ([min,max] inclusive), $contains, $startsWith, $endsWith, $regex (Postgres syntax, case-sensitive — prefix (?i) for insensitive), $in, $nin (blanks count as not-in-list), $not (complement of an inner operator dict), $isEmpty, $isNotEmpty. Field-to-field: $eqField/$neField/$gtField/$gteField/$ltField/$lteField compare against ANOTHER field of the same row by name/id — { "Spent": { "$gtField": "Budget" } } finds over-budget rows. Example: { "Status": "todo", "Priority": { "$between": [5, 9] }, "$or": [{ "Tags": { "$contains": "urgent" } }, { "Due": { "$isEmpty": true } }] } Validation: unknown field keys, unknown operators, and operator/value type mismatches throw 400 with the valid options — typos surface immediately instead of returning all (or wrong) rows. Semantics: $contains is case-insensitive and matches LITERALLY (%, _ and \ are escaped); $eq is exact and case-sensitive with NO type coercion ({Score: "50"} does not match the number 50); { field: null } means "is empty"; $in: [] matches NOTHING. Field types: multiselect filters by MEMBERSHIP ($contains value, $in = any-of, plain scalar = has that tag). created_time/updated_time compare as dates ($gte/$lt ranges; ISO strings). Link fields filter by linked ROW ID via the relation table ($contains id, $in [ids], $isEmpty/$isNotEmpty). Caveat: $gt/$gte/$lt/$lte on text fields that store numeric-looking strings ("16", "8") use LEXICOGRAPHIC comparison, so "8" > "10" is true. Use these operators only on fields whose values are stored as proper numbers, or filter then sort/compare client-side for text-stored numerics.

Sort spec — rowOrder, createdAt, updatedAt, or ANY FIELD NAME/ID, optionally prefixed with "-" for descending (e.g. "-Total"). Defaults to rowOrder asc. Numbers sort numerically, text case-insensitively, blanks always last; scalar lookup/rollup and self-contained formulas sort server-side too. Link fields, volatile formulas and list aggregations are not sortable (clear 400). Unknown keys throw 400 — never silently ignored.

Max rows per page (default: 100, max: 1000)

Skip results for pagination

Include human-readable field-name keys in output (extra cost — set true only when you need it; default false)

Table ID to search in

Search query string. Matches anywhere in the row's serialized data (field IDs and values).

Max results per page (default: 50, max: 500)

Skip results for pagination

Include human-readable field names in output

Table ID

Field ID to match on (must be unique or will update first match)

Value to match against

Row data with field IDs as keys

Include human-readable output

Table ID

Output format: json (full), array (simple), markdown (documentation)

Source table ID to duplicate

Name for the new table

Target database ID (defaults to same database)

Copy all rows from source table

Copy views configuration

Topic: "overview", "fields", "rows", "views", "linked", "buttons"

Table ID from listTables

Access rules per operation (create/read/update/delete). Each is a Condition: a single rule, a legacy flat array (OR), or an { all | any | not } group. Rule types: public, authenticated, creator, field_match, user_property (with op: eq/ne/gt/gte/lt/lte/in/nin/contains/isEmpty/isNotEmpty), has_role, email, function. AND/OR/NOT groups + operators enable any combination.

Whether to track who creates each row (default: true)

Table ID from listTables