Server has 9 tools with basic verb-noun naming (get_, create_, update_, search_). All tools have descriptions (10-150 chars, mostly adequate). However, input schemas are present but minimal: most parameters lack detailed descriptions, enums are absent where needed (e.g., status in update_campaign_status accepts free-form 'Active'/'Paused' without enum constraint), and output schemas are completely undocumented. Error handling returns generic error strings without recovery guidance. No tool annotations (readOnlyHint, destructiveHint). Descriptions are functional but terse (avg ~80 chars vs. production baseline 194 chars). Parameters like 'status' and 'include_deleted' lack constraint documentation. No pagination support despite potential for large result sets (e.g., get_campaigns, get_ad_groups, get_keywords). Responses are free-text strings rather than structured JSON, forcing LLM parsing.

• Convert all tool responses from free-text strings to structured JSON objects with typed fields. E.g., get_campaigns should return {campaigns: [{id, name, status, daily_budget, type}], total_count, next_cursor}.
• Add enum constraints to status parameters: update_campaign_status(status: enum['Active', 'Paused']). Document in parameter description.
• Implement pagination for list tools: add limit (default 20, max 100) and offset/cursor parameters. Return total_count and next_cursor to enable multi-page iteration.
• Enhance parameter descriptions with format/constraint details. E.g., 'daily_budget: float, range $0.01 - $10,000 per day' and 'campaign_id: integer, the numeric ID returned by get_campaigns or create_campaign'.
• Replace generic error handling with actionable recovery guidance. E.g., 'Campaign not found. Call get_campaigns() to list available campaigns.' Categorize errors as retryable vs. user-fixable.

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

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