Zeuxis demonstrates solid tool definition quality with consistent naming patterns, comprehensive parameter schemas, and clear descriptions. All 12 tools follow verb_noun naming (list_*, capture_*). Schemas are well-structured with type definitions, ranges, and constraints. However, descriptions are minimal (10-50 chars) and lack context about when/why to use each tool. Output schemas are not documented. Error handling guidance is absent. The server is read-only (low risk), but descriptions should explain the capture workflow and preset differences more thoroughly.

• Expand tool descriptions to 50-150 chars. For capture tools, explain the use case: 'Capture a specific window by ID from a prior list_windows snapshot. Use when you need a deterministic, repeatable capture of a known window.' For discovery tools: 'List available monitors to select a target for capture_screen or capture_monitor_region.'
• Document output schemas for all capture tools. Specify that results include fields like 'image_data' (base64 or URI), 'format' (png/jpeg/webp), 'dimensions' (width, height), 'timestamp', and 'applied_settings'. This enables LLMs to chain results to downstream analysis tools.
• Add error handling guidance to capture tool descriptions: 'Returns error if window not found, permission denied, or capture times out. Retry with a longer delay_ms or try capture_cursor_window as fallback.'
• Clarify mutually exclusive parameters in CommonCaptureParams description: 'Provide either delay_ms OR delay_seconds, not both. If both are supplied, delay_ms takes precedence.'
• Document OutputInput union behavior: 'Pass a preset string ("analysis", "exact", "compact") for quick defaults, or a detailed object with custom format/max_dimension/jpeg_quality for fine control. Preset and custom modes are mutually exclusive.'

Inferred effective spec: 2025-06-18+.

• Relies on Logging (deprecated) - log to stderr or use OpenTelemetry