Both tools have minimal descriptions (under 100 chars) and lack parameter descriptions entirely. chart_vision_analyze has an input schema with one enum parameter ('region'), but chart_vision_screenshot has an empty schema with no properties. Neither tool documents output structure, error handling, or recovery paths. No tool annotations (readOnlyHint, etc.). Naming is descriptive but descriptions are insufficient for LLM tool selection. This server falls well below production baseline (194 chars avg description, 100% of A+ tools have param descriptions).

• Add detailed descriptions (150-250 chars) explaining WHAT each tool does, WHEN to call it, and what it returns. E.g. 'Captures a screenshot of the trading chart and analyzes candle colors (green/red bias), support/resistance levels via edge detection, and text via OCR. Returns color percentages, level count, and extracted prices. Use this to assess chart state without manual inspection.'
• Document the 'region' parameter: 'Region to capture: "full" = entire screen (includes UI), "chart" = chart area only (excludes toolbars). Use "chart" for focused analysis.'
• Define output schemas for both tools. chart_vision_analyze should document: {status, candles: {bias, gruen_pct, rot_pct}, colors: {name: pct}, levels: {level_count}, zones: {name: pct}, ocr: {full_text, bottom_text, prices_found}}. chart_vision_screenshot should document: {success, file_path, size_bytes}.
• Add error classification and recovery hints. E.g. 'Screenshot failed: no display detected. Ensure DISPLAY env var is set or PIL/ffmpeg is installed. Retryable: yes.' instead of bare 'Kein Screenshot moeglich'.
• Add tool annotations: both tools are read-only (readOnlyHint: true) and idempotent (idempotentHint: true). This helps agents reason about retry safety.

Inferred effective spec: <=2025-11-25.

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