HopperMCP is a STDIO-only server exposing 11 tools for reverse-engineering code analysis via the Hopper disassembler. The server definition is extremely minimal. Tool names are present but follow non-standard naming conventions (Request suffix, CamelCase instead of verb_noun), descriptions are present but generic and brief (10-50 chars), and no input/output schemas are visible in the provided source. The main source file references are SDK headers and sample tool implementations from Hopper itself, not actual MCP tool registration code. No evidence of JSON Schema, parameter types, or structured output definitions. No error handling, validation, or recovery guidance is visible. This server appears to be in early prototype or proof-of-concept stage.

• Rename all tools to follow verb_noun convention in snake_case: ListDocumentsRequest → list_documents, CurrentAssemblyRequest → get_current_assembly, AssemblyByNameRequest → get_assembly_by_name, etc. This lets LLMs parse intent from names alone.
• Add comprehensive input schemas for all tools. Each schema must include: parameter names, types (string, number, boolean, object, array), descriptions (50-100 chars explaining what the param controls and valid formats), required/optional flags, and enums where values are constrained (e.g., 'output_format' with enum [asm, hex, disasm]). Example: { 'document_id': { type: 'string', description: 'Hopper document UUID. If empty, uses current document.' }, 'address': { type: 'string', description: 'Memory address as hex string (e.g., 0x1000)' } }
• Expand descriptions to 100-200 chars. Include: (1) what the tool returns, (2) when to use it vs. similar tools, (3) any prerequisites. Example for get_current_assembly: 'Returns x86-64 assembly code from the active document at the currently selected address. Use this for quick inspection; use get_assembly_by_address for specific addresses without navigation. Requires an open Hopper document.'
• Document output schemas for all tools. Specify return types and all fields. Example for list_documents: { documents: [ { id: string, filename: string, format: string, function_count: number } ], total: number }
• Add error handling descriptions. Document common failure modes (document not open, invalid address, permission denied) and recovery hints. Example: 'If no document is open, call list_documents() first, then pass a valid document_id.'