MCPlug SDK defines 4 tools with basic schemas and descriptions, but exhibits significant quality gaps. Tool naming is inconsistent (get_weather vs get-weather; search_web lacks verb clarity). Descriptions are present but minimal (10-50 chars, below the 50-200 char baseline for LLM optimization). Input schemas exist but lack parameter descriptions, critical for LLM reasoning. The search_web tool exposes an API key as a required parameter, violating secret-injection patterns. No output schemas are documented. Error handling is minimal; the with_error tool provides no recovery guidance. Tool composition is weak: get_weather and get-weather are duplicates with different naming conventions, forcing LLM disambiguation overhead.

• Remove _GOOGLE_API_KEY from search_web parameters. Inject the API key server-side via environment variables and access it in the handler via the 'env' context parameter.
• Consolidate get_weather and get-weather into a single tool with snake_case naming (get_weather). Remove the duplicate.
• Expand tool descriptions to 50-200 characters. Example: 'Get current weather conditions for a city. Returns temperature, humidity, and forecast. Call this when the user asks about weather.'
• Add descriptions to all input parameters. Example: 'city (string, required): City name or code (e.g., 'San Francisco', 'SFO'). Accepts city names or ISO 3166-1 alpha-2 country codes.'
• Document output schemas in the Tool class. Define what fields (temperature, humidity, forecast) are returned and their types so LLMs can chain tools correctly.

Inferred effective spec: 2025-06-18+.

• Relies on Logging (deprecated) - log to stderr or use OpenTelemetry