Descriptions are present but generic (under 100 chars each). No parameter documentation exists because there are no parameters. Tools are read-only and simple, but lack the structure and guidance needed for production use. No error handling, no output schema documentation, no pagination or result limits despite returning structured data.

• Add formal JSON Schema to both tools. Even with no parameters, declare outputSchema with typed fields (e.g., rules: string, entry_points: array, roster_constraints: object). This enables LLM planning.
• Expand descriptions to 100-200 characters. Explain WHEN to call each tool vs using resources. E.g., 'get_game_rules: Fetch verified rules and roster constraints. Use this when the agent needs to validate a roster or explain game mechanics. Prefer this over site://17-0/rules for programmatic access.'
• Document output structure. Return structured JSON with fields like {rules: string, entry_points: [{name, url}], roster_constraints: {positions: [...], limits: {...}}} instead of plain text.
• Add error handling. Return {error: string, recovery: string} on failure. E.g., 'error: "Failed to fetch rules", recovery: "Try again in 30 seconds or check https://17-0.site directly."'
• Consider removing tools if resources are sufficient. If agents can access site://17-0/rules and site://17-0/links directly, tools are redundant. If tools are kept, they must add value (e.g., caching, validation, transformation) not duplication.