Solid foundation with clear naming (verb_noun pattern) and comprehensive parameter descriptions. All 8 tools have input schemas with types and descriptions. However, tool descriptions vary in depth, some lack context on when to use them or what they return. Output schemas are not formally documented. Error handling returns structured dicts but lacks recovery guidance. No tool annotations (readOnlyHint/destructiveHint) despite clear risk profiles. Missing idempotency and confirmation patterns for irreversible operations like flash_firmware.

• Add tool annotations to server.py tool registration: mark flash_firmware with destructiveHint=true and idempotentHint=false; mark read-only tools (list_ports, board_info, fs_ls, fs_get, tail_serial) with readOnlyHint=true.
• Document output schemas for all 8 tools in docstrings or a schema registry. Example: 'Returns {port: str, ok: bool, bytes_written: int, error?: {kind: str, message: str}}'.
• Add a confirm_flash parameter to flash_firmware (separate from confirm_erase) requiring explicit 'true' before any flash operation proceeds, following confirmation-request pattern.
• Enhance error dicts with recovery hints. E.g., BoardError.to_dict() → {kind, message, recovery_hint: 'Check serial connection and try again' or 'Call list_ports() to verify port exists'}.
• Expand tool descriptions to clarify intent and differentiation. E.g., tail_serial: 'Capture raw serial output without REPL protocol, use when board is unresponsive to REPL or you need to see boot logs. For executing code, use repl_exec instead.'