Database Tables

Tables, records (rows), and schema columns (table fields) for org Database Tables. 17 tools.

Cross-cutting patterns

Same conventions as pipe building: introspect_type on inputs such as CreateTableFieldInput / UpdateTableFieldInput, debug=true on mutations, extra_input where the tool exposes it.
IDs: table_id, record IDs, and related parameters are strings in GraphQL (numeric strings for many org tables, or opaque tokens such as fIVcd19N). Prefer quoted strings in MCP/JSON; unquoted numeric JSON may be coerced where the tool uses the shared ID type. See Pipes & cards — Pipefy IDs.
Pagination: get_table_records and find_records support first / after. With the unified MCP envelope, read top-level pagination.has_more and pagination.end_cursor (and pagination.page_size) and pass after=end_cursor for the next page (default page size for listing records is 50; caps apply — see tool docstrings).

Domain	Tools	Notes
Read	`get_table`, `get_tables`, `get_table_records`, `get_table_record`, `find_records`
Table CRUD	`create_table`, `update_table`, `delete_table`	`delete_table` uses preview + `confirm=true` (like `delete_pipe`).
Record CRUD	`create_table_record`, `update_table_record`, `delete_table_record`, `set_table_record_field_value`, `upload_attachment_to_table_record`	`upload_attachment_to_table_record` runs presigned URL + S3 PUT + `setTableRecordFieldValue` for attachment columns. One file per call: to attach multiple files, call the tool once per file. Same inputs as card upload: `file_path` is the local filesystem path the MCP server reads (supports `~` expansion); `file_name` is inferred from the path's basename when omitted. Files larger than 100 MiB are rejected locally before any network call. `field_id` must be the field slug, not the uuid.
Field CRUD	`create_table_field`, `update_table_field`, `delete_table_field`	Schema columns; `delete_table_field` is destructive (confirm with the user).