Server has 6 tools with basic schemas and descriptions in German. All tools have input schemas with types and descriptions, which is positive. However, descriptions are generic and lack LLM-optimized guidance on WHEN to use each tool vs. alternatives. Parameter descriptions are minimal (e.g., 'Name des Kandidaten' without format hints). No output schemas documented. Error handling is minimal, most tools return raw API responses or generic error dicts without recovery guidance. No tool annotations (readOnlyHint, etc.). Tools lack composition guidance (e.g., which tool to call first for discovery). German descriptions reduce LLM clarity in English-dominant environments.

• Add explicit output schemas to each tool. Document field names, types, and what each field represents. E.g., get_candidate_fundraising should declare it returns {kandidat: string, partei: string, einnahmen_gesamt: number, ...}.
• Rewrite descriptions in English and optimize for LLM selection. Include WHEN to use (e.g., 'Use this to get total fundraising for a single candidate. For industry-wide donor trends, use get_top_donors_by_industry instead.'). Target 50-200 chars with actionable context.
• Replace example values in parameter descriptions with enum constraints or format patterns. E.g., chamber: enum ['senate', 'house'] instead of 'chamber: "senate" oder "house"'.
• Add tool annotations: set readOnlyHint=true on all tools since they only read data. This signals to agents that these calls are safe to retry.
• Implement error recovery guidance. Catch httpx.HTTPError and return structured errors like {error: 'Candidate not found', suggestion: 'Try search_candidates() with a partial name', retryable: false}.