Enhanced Apple Notes MCP server with search, bulk operations, and workflow integration
This server provides 8 well-named tools with explicit JSON schemas and basic descriptions, but lacks critical details needed for production LLM agents. All tools are directly visible in server/index.js with proper registration. Naming follows verb_noun convention (list_notes, search_notes, get_note_content, add_note, update_note_content, list_folders, create_folder, bulk_export). However, descriptions are minimal (average ~80 chars, well below the 194-char baseline for A+ tools), parameter descriptions lack detail about constraints and formats, output schemas are undocumented, and error handling provides no recovery guidance. The server lacks output schema documentation entirely, LLMs cannot determine what fields to expect from tool responses, forcing them to guess or make follow-up calls. Tools like list_notes and bulk_export accept limit parameters but provide no pagination guidance. Critical: no documentation of what data structures are returned, whether operations are idempotent, or how errors should be handled and recovered from.
Create a new note in Apple Notes
Export multiple notes as plain text with metadata
Create a new folder in Apple Notes
Get the content of a specific note by its name
List all folders in Apple Notes
List all notes from Apple Notes app with optional filtering
Search notes by title or content
No documented output schemas. LLMs cannot determine what fields to expect from any tool response (e.g., what does list_notes actually return? field names, data types, nesting structure). This forces agents to guess or make wasteful follow-up calls.
Parameter descriptions lack constraint details. 'Maximum number of notes to return (default: 20)' does not specify min value, whether the limit is enforced, what happens if limit exceeds available notes, or pagination strategy. LLMs cannot validate inputs or understand valid ranges.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-22 | F | 49 | 2026-07-28+ | v2 |
| 2026-03-09 | F | 19 | - | v1 |
Update the content of an existing note
Tool descriptions are too brief (40-55 chars), well below the 194-char baseline for A+ tools. Descriptions lack context on WHEN to use each tool vs. alternatives (e.g., when should an agent use search_notes vs. get_note_content?), what happens on error, or prerequisites. Example: 'List all notes from Apple Notes app with optional filtering' does not explain pagination, result limits, or what metadata is returned.
No error handling or recovery guidance in tool definitions. What happens if a note does not exist? If folder creation fails due to a duplicate name? If AppleScript execution times out? LLMs receive no guidance on whether to retry, ask the user, or escalate.
No idempotence declaration. Operations like add_note and update_note_content modify state. If an LLM retries due to a transient AppleScript error, does the operation happen twice (creating a duplicate note)? Agents need explicit idempotence guidance or confirmation patterns.
No pagination or result-limiting strategy documented. list_notes and bulk_export accept a limit parameter but do not explain: (1) Does limit cap the returned results, or is it a hint? (2) Are results paginated with a cursor or offset? (3) What is returned if there are more results than the limit? (4) Can the agent fetch the next page?
Parameter 'folder' is optional in list_notes, search_notes, and get_note_content, but behavior when omitted is undocumented. Does list_notes return notes from all folders, or a default folder? Does search_notes search all notes or only in a default location? Ambiguous behavior forces agents to guess.
bulk_export 'format' parameter uses enum correctly, but description does not explain what the difference is between 'json' and 'markdown' output or when each is appropriate. Also does not document the structure of the returned data (JSON schema) or how to consume markdown export.