Server has 6 tools with visible schemas and descriptions, but quality is inconsistent. All tools have descriptions (10-60 chars, mostly short), and input schemas are present with type definitions. However, descriptions are generic and lack actionable context for LLM selection. Parameter descriptions are minimal. No output schemas documented. Error handling is absent from tool definitions. Tools follow verb_noun naming (watch_*, drain_events) but lack clarity on when to use each watch variant. No tool annotations (readOnlyHint, etc.) despite all being read-only. Composition is reasonable, tools chain via watcher_id, but descriptions don't explain this dependency.

• Expand tool descriptions to 100-200 chars. Example: 'watch_path: Monitor a filesystem path for file creation, modification, or deletion events using native macOS FSEvents. Returns a watcher_id; call drain_events to retrieve accumulated events. Use this for file-based triggers (config changes, log rotation, etc.).'
• Document output schema for each tool. Add a 'Returns' section: 'Returns {watcher_id: string, kind: string, timestamp_ms: number, data: object}. Use watcher_id with drain_events to retrieve events.'
• Add range constraints to numeric parameters. Example: 'interval_ms: Polling interval in milliseconds (100 - 60000; default 2000). Lower values increase CPU usage; higher values increase latency.'
• Add tool annotations to schema. Mark all tools with 'readOnlyHint: true' to signal safety. Example in rmcp: tool.readOnlyHint = true.
• Clarify watch_power behavior. Describe what events it emits (sleep, wake, battery status change) and note that it takes no parameters. Explain the watcher_id is returned in the first event or via a separate initialization response.