Overboard Studio

Creates new collaborative whiteboards on Overboard Studio through natural language commands processed by an MCP server that handles OAuth 2.0 PKCE authentication flow and persists board state to Overboard's backend. The MCP tool translates conversational intent ('create a board for sprint planning') into authenticated API calls that instantiate a new board resource with configurable metadata, returning a shareable board URL and ID for downstream operations.

Adds sticky notes, shapes, and text elements to existing whiteboards via MCP tool calls that accept natural language descriptions and translate them into structured element creation requests. The system maps conversational positioning hints ('top-left corner', 'center of the board') to canvas coordinates and supports element styling (color, size, font) through optional parameters, persisting elements to the board's collaborative state.

Supports bulk creation, update, or deletion of multiple board elements in a single MCP tool call with transactional guarantees — either all operations succeed or all fail, preventing partial board state corruption. The system accepts an array of element operations, validates them against the current board state, and applies them atomically, returning a detailed result set indicating success/failure for each operation.

Searches board elements by content, type, position, or metadata using MCP tool calls that support both exact matching and fuzzy/semantic search. The system accepts query parameters (text search, element type filter, bounding box region) and returns matching elements with relevance scores, enabling AI systems to locate specific content on large boards without retrieving the entire board state.

Exports whiteboard content to standard formats (JSON, SVG, PNG, PDF) via MCP tool call, enabling integration with external tools and documentation systems. The system serializes board state to the requested format, handling layout preservation, styling conversion, and asset embedding, returning a downloadable file or data URL that can be processed downstream.

Manages reusable whiteboard templates (predefined layouts, element sets, styling) via MCP tools that support creating templates from existing boards and instantiating new boards from templates. The system stores template metadata and element definitions, enabling rapid board creation with consistent structure and styling across teams.

Enables users to manage board access and sharing through natural language commands (e.g., 'share this board with the design team'), with the MCP server translating intent into collaborator invitations and permission updates. The implementation infers collaborator lists from context, resolves email addresses, and applies appropriate permissions based on the described sharing intent.

Creates visual connectors (lines, arrows) between sticky notes and shapes on a whiteboard to represent relationships, dependencies, or workflows. The MCP tool accepts source and target element IDs and optional connector styling (arrow type, line style, label), translating these into graph-like relationship data that persists in the board's collaborative state and renders in real-time for all connected users.

Fetches current board state (all elements, connectors, metadata) from Overboard Studio via MCP tool call, returning a structured representation of the whiteboard's content. The system supports filtering by element type or region and includes real-time synchronization metadata (last modified timestamp, collaborator presence), enabling AI systems to reason about board state and make context-aware decisions for subsequent operations.

Invites users to a whiteboard and manages access permissions via MCP tools that accept email addresses and permission levels (view, edit, admin). The system sends invitations through Overboard's notification system and persists permission metadata to the board's access control list, enabling real-time collaboration with granular permission enforcement across all connected clients.

Fetches the activity log for a whiteboard, including all element modifications, collaborator actions, and timestamps, via MCP tool call. The system returns structured activity records (who changed what, when) that enable AI systems to understand collaboration patterns, detect conflicts, and reconstruct board evolution over time without requiring manual log parsing.

Implements OAuth 2.0 PKCE (Proof Key for Code Exchange) authentication within the MCP protocol, handling the full authorization flow (authorization request, code exchange, token refresh) without exposing credentials to the LLM. The MCP server maintains session state across tool calls, enabling secure, long-lived authenticated access to Overboard resources without requiring users to re-authenticate for each operation.

Retrieves and updates board-level metadata (name, description, visibility, created/modified timestamps) via MCP tools that accept configuration parameters and persist changes to Overboard's backend. The system supports read-only metadata retrieval for context awareness and write operations for board customization, enabling AI systems to manage board properties programmatically as part of larger automation workflows.

Enables simultaneous editing by multiple collaborators through MCP tool calls that are merged into a conflict-free collaborative state using operational transformation or CRDT-like semantics. Each tool call (element creation, modification, deletion) is assigned a unique operation ID and timestamp, allowing the system to deterministically resolve concurrent edits without requiring explicit locking or consensus, ensuring all collaborators converge to the same board state.

Generates shareable board URLs with embedded access tokens and optional expiration, enabling external collaborators to access whiteboards without requiring Overboard accounts. The MCP tool accepts permission levels and optional time-based expiration, returning a unique shareable link that enforces access control at the HTTP layer while maintaining the underlying board's permission model.