Duffel exposes 3 tools with basic descriptions and partial schemas. Tool names follow verb_noun convention (search, read, append), which is good. However, descriptions are minimal (10-50 chars), parameter descriptions are absent or trivial, and output schemas are not documented. No error handling guidance, no pagination support for search results, and no indication of which parameters are required vs optional. The tools lack the depth needed for reliable LLM invocation.

• Expand tool descriptions to 100-200 characters. For duffel_search: 'Search notes by keyword or phrase. Returns up to 20 matching file paths and snippets. Use this to discover relevant notes before calling duffel_read to fetch full content.'
• Add parameter descriptions: 'query' → 'Search term or phrase (1-500 chars, case-insensitive)'; 'path' → 'File path relative to notes root (e.g., projects/2024/q1.md)'.
• Document output schema for each tool. E.g., duffel_search returns {results: [{path: string, snippet: string, score: number}], total: number, next_cursor?: string}.
• Add pagination to duffel_search: include offset/limit or cursor-based pagination, return total count, and cap results at 20 by default.
• Add error handling guidance in descriptions. E.g., duffel_read: 'If file not found, try duffel_search() to locate the correct path. Returns 404 if path does not exist.'