This server has severe definition quality issues. Tool 1 (execute_skill_action) has no visible input schema in the source code, only a description and risk label. Tool 2 (source_wholesale_produce) has a basic schema with 3 parameters, but descriptions are minimal and lack actionable detail. The server does not register tools via standard MCP mechanisms; instead, it prints a JSON status message to stdout. Tool definitions appear inferred from client.py rather than explicitly registered. No error handling, no output schema documentation, no parameter validation guidance. Descriptions are generic and do not explain WHEN to use each tool or what downstream actions they enable.

• Add explicit tool registration in mcp_server.py using MCP's tool definition format (name, description, inputSchema, outputSchema). Do not rely on client.py class methods.
• For execute_skill_action: define a complete inputSchema with required parameters (e.g., action_type, farm_id, quantity). Specify enum values for action_type (e.g., 'source', 'verify_cold_chain', 'schedule_delivery').
• For source_wholesale_produce: expand parameter descriptions to 50-100 chars each. Example: 'farm_region: Geographic region for sourcing (e.g., Central Valley, Salinas Valley, Monterey County). Determines available farms and harvest schedules.'
• Document output schemas for both tools. For source_wholesale_produce, declare: {batch_id: string, farm_source: string, harvest_timestamp: ISO8601, temperature_log_celsius: number, cold_chain_compliant: boolean, wholesale_cost_discount_pct: number, last_mile_cold_truck_eta_hours: number, restaurant_delivery_ready: boolean}.
• Add error handling: return structured errors with actionable guidance. E.g., 'Invalid farm_region: got "Unknown Valley". Valid regions: Central Valley, Salinas Valley, Monterey County. Call list_available_regions() to discover all options.'