Baserow

Enables read and write operations on Baserow table rows through MCP protocol, exposing individual row creation, retrieval, update, and deletion as discrete tool calls. Implements row-level mutations with field-value validation against the table's 20+ typed field definitions (text, number, select, date, links, etc.), returning structured row objects with metadata. Works by translating MCP tool invocations into Baserow's internal row storage layer, respecting workspace and table-level permissions defined in the hosting tier.

Automatically validates and transforms input data against Baserow's 20+ typed field definitions (single-line text, long text, number, rating, boolean, date/time, URL, email, file, select, link-to-table, lookup, rollup, collaborator, count, duration, autonumber, UUID, password, etc.) before persisting rows. Implements field-specific coercion rules (e.g., converting ISO date strings to date fields, validating email format, enforcing select options) and returns validation errors with field-level details. Enables LLMs to understand table schema constraints and generate valid mutations without trial-and-error.

Exposes single-select and multi-select field options as queryable enumerations through MCP, enabling LLMs to understand available choices and enforce constraints when populating select fields. Implements option enumeration by fetching the list of valid options for a select field and returning them with metadata (option ID, label, color). Validates mutations against the option list, rejecting invalid selections and returning constraint violation errors.

Supports reading and writing date and date/time fields through MCP with timezone awareness, enabling LLMs to work with temporal data correctly. Implements date/time handling by accepting ISO 8601 formatted strings or date objects and converting them to Baserow's internal format, with timezone information preserved. Returns dates in ISO 8601 format with timezone metadata, enabling agents to reason about temporal relationships and schedule-based workflows.

Supports reading and writing rating (1-5 star) and boolean (true/false) fields through MCP, enabling LLMs to populate simple categorical fields and understand binary states. Implements rating fields as integer values (1-5) with validation, and boolean fields as true/false values. Returns typed values in row responses, enabling agents to reason about ratings and boolean states.

Validates and stores email, URL, and phone number fields through MCP with format-specific validation rules, enabling LLMs to populate contact fields correctly. Implements validation by checking email format (RFC 5322), URL format (valid protocol and domain), and phone number format (international or regional), rejecting invalid values with detailed error messages. Returns validated values in row responses, ensuring data quality for contact information.

Supports writing to password fields through MCP with secure hashing and storage, enabling LLMs to set passwords or secrets in Baserow records. Implements password storage by accepting plaintext passwords and hashing them using Baserow's secure hashing algorithm before storage, with read access restricted to prevent plaintext exposure. Returns only a masked indicator on retrieval, preventing password leakage.

Supports reading and writing long text fields through MCP with optional rich-text formatting (markdown, HTML), enabling LLMs to store and retrieve formatted content. Implements long text fields as plain text or rich-text content with optional formatting metadata, returning formatted content in row responses. Enables document-like content storage within Baserow records.

Enforces Baserow's hierarchical permission model (workspace → database → table → field level) at the MCP tool invocation layer, restricting read/write access based on the authenticated user's role and tier. Implements role-based access control (RBAC) for workspace members and team workspaces, with field-level and view-level permissions available in Advanced tier. Prevents unauthorized mutations by validating credentials against Baserow's permission store before executing row operations, returning permission-denied errors for disallowed operations.

Exposes Baserow table schemas (field names, types, constraints, linked tables) as queryable MCP resources, enabling LLMs to discover available tables, understand field definitions, and generate valid mutations without hardcoding schema details. Implements schema caching or on-demand introspection to retrieve field metadata (type, required, unique, linked-table references, select options) and returns structured schema objects. Allows agents to dynamically adapt to schema changes without code updates.

Enables querying and traversing linked-record relationships (link-to-table fields) and computed fields (lookups, rollups) through MCP, allowing LLMs to fetch related rows and aggregated data without separate API calls. Implements relationship traversal by following link-to-table field references to fetch rows from linked tables, and read-only access to lookup and rollup fields that compute values from related records. Returns nested or flattened data structures depending on the query pattern.

Supports batch creation, update, or deletion of multiple rows in a single MCP invocation, reducing round-trip latency and enabling efficient bulk data operations. Implements batch processing by accepting arrays of row objects or mutations and executing them as a group, with optional transaction semantics or partial-failure handling. Returns success/failure status per row and aggregated results, enabling LLMs to perform data migrations or bulk corrections efficiently.

Enables querying rows through Baserow's view definitions (Grid, Gallery, Form, Kanban, Calendar, Timeline), which encapsulate filters, sorts, and grouping rules. Implements view-based queries by applying the view's filter and sort configuration to row retrieval, allowing LLMs to fetch pre-filtered datasets without specifying filter logic manually. Returns rows in the order and grouping defined by the view, enabling context-aware data retrieval.

Supports reading and writing file attachments in Baserow file fields through MCP, enabling LLMs to reference, upload, or retrieve files associated with rows. Implements file handling by accepting file references (URLs, file IDs, or base64-encoded content) and storing them in file fields, with metadata (filename, size, MIME type) returned on retrieval. Enables document processing workflows where LLMs can attach files to records or fetch file metadata for downstream processing.

Enables assigning Baserow workspace users to collaborator fields through MCP, allowing LLMs to assign tasks, responsibilities, or ownership to team members. Implements user assignment by accepting user IDs or email addresses and validating them against the workspace's user roster, returning user metadata (name, email, avatar) on retrieval. Enables workflow automation where LLMs assign work items to appropriate team members based on rules or context.

Supports reading auto-generated autonumber and UUID fields through MCP, enabling LLMs to understand and reference system-generated identifiers. Implements auto-generation by allowing the database to populate these fields on row creation, with LLMs reading the generated values after insertion. Returns generated IDs in row responses, enabling agents to reference newly created records in subsequent operations.