This server is fundamentally non-functional as an MCP implementation. No tool definitions are visible in the provided source code. The repo appears to be an SDK wrapper around a remote MVR API service, not an actual MCP server with registered tools. The packages/mvr-mcp-bridge/src/index.mjs file is truncated mid-function and does not show any tool registration, schema definitions, or parameter documentation. The five tools listed (mvr_first_call, mvr_entity_resolve, mvr_evidence_completeness, mvr_context_compile, mvr_decision_check) are referenced only in a Python file path and do not appear to be implemented in the TypeScript MCP bridge code. Without visible tool definitions, parameter schemas, or descriptions, the server cannot be evaluated against the 54 Agentic Tool Patterns.

• Create explicit tool registration in the MCP bridge using @modelcontextprotocol/sdk. Call server.setRequestHandler(ListToolsRequestSchema, ...) and server.setRequestHandler(CallToolRequestSchema, ...) with complete tool definitions.
• For each tool, define a JSON Schema inputSchema with: type 'object', properties object listing all parameters with type and description, required array naming mandatory params, and additionalProperties: false.
• Write 50-200 character descriptions for each tool following the pattern: 'WHAT it does, WHEN to use it, WHAT it returns.' Example: 'Resolve an entity to confirm market viability. Returns entity details and relational readiness score. Call this first before evidence completeness checks.'
• Add description to every parameter. Example: 'entity_name (string, required): The legal or trading name of the entity being resolved. Used for regulatory lookup and verification.'
• For tool names, adopt clear action verbs: 'resolve_entity' instead of 'mvr_entity_resolve', 'check_decision_readiness' instead of 'mvr_decision_check', 'assess_evidence' instead of 'mvr_evidence_completeness'.
• Document output schemas showing expected response fields. Example: 'Returns object: { entity_id: string, entity_name: string, archetype: string, readiness_score: number, market_verification: { country, city, status } }'.
• Add error handling examples in tool descriptions: 'If entity not found, returns status: entity_not_found with suggested_searches array. Try search_entity with partial name.' Per pattern:recovery-guide.