OmniFocus-MCP

Executes targeted queries against the OmniFocus database through AppleScript/JXA automation, supporting field selection, filtering predicates, and sorting parameters to retrieve specific task subsets without full database export. The query_omnifocus tool implements a schema-based filtering interface that translates LLM parameters into AppleScript queries, enabling efficient retrieval of tasks by status, project, due date, and custom metadata without loading the entire database into context.

Exports the entire OmniFocus database as a structured JSON snapshot through AppleScript automation, providing comprehensive access to all tasks, projects, folders, and metadata for holistic analysis or context-aware reasoning. The dump_database tool serializes the complete OmniFocus object hierarchy (projects, tasks, contexts, perspectives) into a single JSON payload, enabling LLMs to perform cross-cutting analysis, dependency detection, or full-system optimization in a single context window.

Validates tool parameters against JSON schemas before execution, ensuring MCP protocol compliance and preventing invalid requests from reaching AppleScript execution. Each tool defines a schema that specifies required/optional parameters, types, and constraints (e.g., ISO 8601 date format, valid project names), and the MCP server validates incoming requests against these schemas before invoking handlers. This approach provides early error detection, clear error messages to LLM clients, and prevents malformed AppleScript commands from being generated.

Creates individual tasks in OmniFocus through AppleScript automation, accepting natural language task names and optional structured metadata (project assignment, due dates, tags, notes, priority flags) via MCP schema parameters. The add_omnifocus_task tool translates LLM-generated task specifications into AppleScript commands that instantiate new task objects with full metadata binding, enabling AI assistants to create tasks with context-aware properties without requiring users to manually enter details.

Creates new projects in OmniFocus with optional parent project assignment, enabling hierarchical task organization through AppleScript automation. The add_project tool accepts project names and parent project references, translating them into AppleScript commands that instantiate project objects with proper hierarchy binding, allowing AI assistants to organize tasks into logical groupings and sub-projects based on conversation context or task analysis.

Updates existing OmniFocus tasks or projects by modifying specific fields (name, due date, project assignment, tags, notes, completion status, priority flags) through AppleScript automation. The editItem tool accepts item IDs and a map of field updates, translating them into AppleScript property assignments that modify task/project objects in-place, enabling AI assistants to refine task details, reschedule work, or update metadata based on conversation context without full item replacement.

Removes tasks or projects from OmniFocus through AppleScript automation, supporting both individual item deletion and cascading removal of nested items (deleting a project removes all contained tasks). The removeItem tool accepts item IDs and executes AppleScript delete commands that remove objects from the OmniFocus database, enabling AI assistants to clean up completed work, archive obsolete projects, or reorganize task structures based on analysis or user intent.

Creates multiple tasks in OmniFocus in a single MCP call through AppleScript automation, accepting an array of task specifications with metadata (names, due dates, projects, tags, notes) and executing sequential AppleScript commands to instantiate all tasks atomically. The batch_add_items tool optimizes bulk task creation by reducing MCP round-trips and AppleScript process overhead compared to individual add_omnifocus_task calls, enabling AI assistants to populate projects with related tasks or implement task breakdowns in a single operation.

Removes multiple tasks or projects from OmniFocus in a single MCP call through sequential AppleScript execution, accepting an array of item IDs and executing delete commands for all items atomically. The batch_remove_items tool optimizes bulk deletion by reducing MCP round-trips and AppleScript process overhead compared to individual removeItem calls, enabling AI assistants to clean up large sets of obsolete items or reorganize task structures efficiently.

Manages the OmniFocus MCP server initialization, tool registration, and stdio transport setup, implementing the Model Context Protocol specification through the @modelcontextprotocol/sdk framework. The server instantiates an McpServer object, registers eight tools with their schemas and handlers, establishes StdioServerTransport for process-based communication with AI assistants (Claude, etc.), and handles connection lifecycle with error handling, enabling seamless integration between LLMs and OmniFocus through standardized MCP protocol.

Executes AppleScript/JXA commands to interact with the OmniFocus application process, enabling task/project CRUD operations and database serialization through direct object model access. The architecture uses AppleScript as the integration layer between the Node.js MCP server and the OmniFocus application, translating tool parameters into AppleScript commands that instantiate, modify, query, or delete OmniFocus objects, and serializing results back to JSON for LLM consumption. This approach leverages OmniFocus's scriptable object model (projects, tasks, contexts, perspectives) to provide programmatic access without requiring database file manipulation.