Flameshot MCP has complete input schemas with proper types and descriptions for all 6 tools. Tool names follow verb_noun convention (capture_*, list_*, annotate). Descriptions are present and moderately detailed (50-150 chars), explaining WHAT each tool does. However, descriptions lack WHEN to use guidance and dependency hints. Parameter descriptions are adequate but generic in places. Output schemas are not explicitly documented, responses are inferred from code. Error handling exists but lacks recovery guidance. No tool annotations (readOnlyHint/destructiveHint). Overall: solid foundation, but missing LLM-optimization patterns and composition guidance.

• Document output schemas explicitly. Add a docstring to each tool listing returned fields: 'Returns: {ok: bool, path: str, bytes: int, size: str, wsl_path?: str, image?: bytes}'. This enables agents to extract and chain results.
• Enhance tool descriptions with WHEN-to-use guidance. E.g. capture_screen: 'Captures a single monitor by index. Use this when the user specifies a monitor number; use capture_full for all monitors at once; use capture_gui for interactive selection.'
• Add tool annotations via @mcp.tool(destructiveHint=True) or equivalent to declare write side effects. This signals to agents that these tools modify state and may need confirmation or careful retry logic.
• Improve error messages with recovery hints. E.g. 'Flameshot not found. Install via: apt install flameshot (Linux), brew install --cask flameshot (macOS), or download from https://flameshot.org/ (Windows). Or set FLAMESHOT_BIN env var.'
• Add regex pattern or enum constraint for 'region' parameter. E.g. description: 'Region in format WIDTHxHEIGHT+X+Y (e.g. 800x600+100+50). Width, height, X, Y must be positive integers.' Or validate in code and return 'Invalid region format: got "800x600 100x50", expected "800x600+100+50"'.