Server provides 7 read-only tools with basic descriptions and input schemas. All tools follow verb_noun naming (file-read, file-list, etc.) and have parameter descriptions. However, descriptions are terse (10-40 chars), lack context on WHEN to use each tool, and omit output schema documentation. No error handling guidance, no parameter constraints (enums/ranges), and no indication of pagination support for list/search tools. Tool names use hyphens (Claude.ai compatibility layer) but underlying schemas are visible and properly typed. Security posture is strong (read-only, no secrets exposed), but definition quality lags production baselines.

• Expand each tool description to 100-200 chars. Include: what it does, when to use it vs similar tools, and what it returns. E.g. 'file-list: List markdown files in a directory. Use this to explore vault structure; use file-find for pattern matching across the whole vault. Returns array of {path, size, modified}.'
• Document output schemas formally. For file-list, specify: returns {entries: [{path: string, size: number, modified: ISO8601}], total: number}. For search-content, specify hit structure including line numbers and context snippets.
• Add pagination to list/search tools. Add optional limit (default 20, max 100) and offset/cursor parameters. Document result caps in descriptions: 'Returns up to 20 results; use offset to paginate.'
• Add parameter format hints. E.g. 'file: Path to markdown file (relative to vault root, .md extension required)' and 'pattern: Glob pattern (e.g. *.md, notes/*.md, **/2024-*.md)'.
• Add error recovery guidance. E.g. 'If file not found, try file-list to explore available paths' and 'If search returns no results, try a simpler pattern or broader query'.