Teable MCP (Official)

Read data with a GET request to the Teable REST API (reference: https://help.teable.ai/en/api-doc/overview): views, collaborators, automations and anything else search-api finds as a GET endpoint. Find the url template with search-api first. Never changes data.

Example: call-api-read({ url: "/table/{tableId}/view", params: { tableId: "tblXXX" } })

Create, change or delete data with a POST, PUT, PATCH or DELETE request to the Teable REST API (reference: https://help.teable.ai/en/api-doc/overview): tables, fields, views, automations and anything else search-api finds. Find the method and url template with search-api first; get-doc explains payload formats (formulas, filters, link fields).

Example: call-api-write({ method: "POST", url: "/base/{baseId}/table", params: { baseId: "bseXXX" }, data: { name: "Tasks" } })

Create records using compact array format (saves tokens).

FORMAT: { "tableId": "tblXXX", "header": ["Name", "Status", "Priority"], "records": [ ["Task A", "Done", "High"], ["Task B", "Pending", "Low"] ] }

RULES:

• header: Array of field names (define column order)

• records: 2D array where each row matches header order

• Empty cells: use null or ""

VALUE TYPES:

• Text: "Hello"

• Number: 42 or 3.14

• Boolean: true / false

• Date: "2025-01-27"

• Select: "Option A"

• MultiSelect: ["A", "B"]

• Link: "Record Name"

Delete records from a table.

Parameters:

• tableId (required): The table ID

• recordIds (required): Array of record IDs to delete (max 1000)

Returns:

• success: Whether the operation succeeded

• deletedCount: Number of records deleted

Example: { tableId: "tblXXX", recordIds: ["rec001", "rec002"] }

Get comprehensive documentation for Teable features and configurations. Use this tool to understand configuration formats before creating or updating resources. Workflow:

1. Identify the category needed (field, view, or record)

2. Query the specific subtopic (e.g., "field.formula" or "field.colors")

3. Use examples from documentation to construct correct configurations

Get fields metadata from a table.

Parameters:

• tableId (required): The table ID

• fieldId (optional): Filter by specific field ID

• fieldName (optional): Filter by specific field name

Returns field information:

• id: Unique field identifier

• name: Display name

• type: Field type (text, number, date, etc.)

• description: Field description

• dbFieldName: Database column name

• options: Field configuration

For DATA QUERIES (search, filter, paginate). For current view rows, use get-grid-rows.

Parameters:

• tableId (required)

• viewId - Apply view filters (optional)

• recordId - Fetch a specific record by ID. When provided, search/take/skip are ignored (optional)

• take - Records to return, default 100, max 1000 (optional)

• skip - Records to skip for pagination (optional)

• projection - Field IDs or names to return (optional). If not provided, only first 50 fields will be returned

• search - Search filter object: { value: "keyword", fieldId: "Name" } (fieldId accepts field name OR ID; omit for global search) (optional)

Important:

• If user refers to rows in current view ("first 3 rows", "last row", "row N"), use get-grid-rows instead.

• This tool queries server-persisted data. For the user's live view state (unsaved filters/sorts), use get-grid-rows.

Response (tabular format): {header: ["id", "Name"], rows: [["rec001", "Task A"]]}

Examples: All: { tableId: "tblXXX" } By ID: { tableId: "tblXXX", recordId: "recXXX" } Search: { tableId: "tblXXX", search: { value: "keyword", fieldId: "Name" } } Pagination: { tableId: "tblXXX", take: 50, skip: 100 }

Get tables metadata in a project.

Parameters:

• baseId (required): The project ID

• tableId (optional): Filter by specific table ID

• tableName (optional): Filter by specific table name

Returns table information:

• id: Unique table identifier

• name: Display name

• dbTableName: Database table name (use this for SQL queries)

• description: Table description

• icon: Table icon

List the projects (bases) the caller can access, with their IDs, names, spaces and the caller's role, one page at a time. Call this first to find the baseId other tools need; pass search or spaceId to find a specific base instead of paging through everything.

Search the Teable REST API (reference: https://help.teable.ai/en/api-doc/overview) by keyword or description. Returns matching endpoints within the caller's permission level: method, url template, parameters and request body schema. Run GET endpoints with call-api-read and POST/PUT/PATCH/DELETE endpoints with call-api-write, passing method and url unchanged.

Parameters:

• query (required): search keywords in any language

• tags: filter by API category (e.g., ["record", "field"])

• limit: max results (default 5, max 10)

Run a read-only PostgreSQL query (SELECT only) on the tables of one project. Use it for counts, sums, grouping and reads that join several tables.

Use database names, not display names or IDs:

• Table: the dbTableName from get-tables-meta, already in "schema.table" form; quote each part, e.g. FROM "bseXXX"."tblYYY"

• Column: the dbFieldName from get-fields, in double quotes, e.g. "Status"

• System columns: "__id", "__auto_number", "__created_time", "__last_modified_time", "__created_by", "__last_modified_by"

Add LIMIT 100 to queries that return rows. Timestamps are stored in UTC; convert with timezone('Asia/Shanghai', "col") when needed.

Update records using compact array format.

FORMAT: { "tableId": "tblXXX", "header": ["recordId", "Name", "Status"], "records": [ ["recAAA", "Updated Name A", "Done"], ["recBBB", "Updated Name B", "Pending"] ] }

RULES:

• header: First element MUST be "recordId"

• records: First value in each row is the record ID to update

• Empty string "": Skip field (no change)

• null: Clear the cell

VALUE TYPES: Same as create-records

REORDER (optional): "order": { "viewId": "viwXXX", "anchorId": "recYYY", "position": "before"|"after" } Moves record(s) before/after the anchor in the given view. Swap rows A↔B: call twice with different anchors.