The server exposes a single tool 'query' with an empty input schema ({}). No parameter descriptions, no output schema documentation, and no tool description visible in the source. The tool name 'query' is generic and violates verb_noun naming conventions. Tool registration appears to occur in src/mcp.rs but the actual definition is not shown in the provided code sample, making it impossible to verify schema structure or descriptions. This is a critical failure against baseline patterns.

• Rename 'query' to a specific action verb: 'search_campaigns', 'get_marketing_metrics', 'query_salesforce_data', or similar. Verb_noun naming lets LLMs parse intent from the name alone.
• Add a comprehensive tool description (50-200 chars): state WHAT it does, WHEN to use it, and what it returns. Example: 'Search marketing campaigns and analytics data from Salesforce. Returns campaign name, cost, conversions, and ROI metrics. Use this to answer questions about campaign performance.'
• Define input parameters with types and descriptions. If the tool accepts a search query, add: {"type": "object", "properties": {"query": {"type": "string", "description": "Search term or campaign name (e.g., 'Q4 2024 email campaign')"}}, "required": ["query"]}
• Document the output schema with typed fields. Example: {"type": "object", "properties": {"campaigns": {"type": "array", "items": {"type": "object", "properties": {"id": {"type": "string"}, "name": {"type": "string"}, "cost": {"type": "number"}, "conversions": {"type": "integer"}, "roi": {"type": "number"}}}}, "total_count": {"type": "integer"}}}
• Add error handling guidance. If a query fails, return: 'Invalid search query. Try: search_campaigns(query="email") or search_campaigns(query="Q4 2024")' instead of a bare error code.

Inferred effective spec: <=2025-11-25.

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