Scoring was not performed

• Document output schema for both tools. Define the structure of SearchArticlesResult and SearchImagesResult, what fields are in articles[] and images[], their types (string, int, date), and which are always present vs optional. Example: {articles: [{url: string, title: string, publish_date: string, tone: number, ...}], query: string, total_count: int}
• Add explicit min/max to numeric parameters in descriptions. Change 'Maximum articles to return (1 - 250, default 50)' to include in both enum/constraint and description: max_records (integer, 1-250, default 50). Same for max_records in search_images.
• Add validation error guidance to descriptions. For start_date_time/end_date_time: 'Must be YYYYMMDDHHMMSS (e.g. 20240115093000). If invalid, tool returns error: Invalid date format.'
• Document mutually exclusive parameters. In search_articles: 'Cannot use both timespan AND start_date_time/end_date_time together. Use timespan for relative searches (e.g. 7d, 1month) OR explicit date range, not both.'
• Categorize errors and include recovery guidance in error messages. Instead of raising ValueError(f'Failed to search articles: {exc}'), map exception types: if timeout → return 'API timeout (retryable): Reduce max_records or try again in 30s'; if invalid_query → return 'Invalid query syntax: Check Boolean operators (AND, OR) are uppercase'; if auth_failure → return 'Authentication failed (fatal): Check GDELT API key configuration'.
• Document pagination if GDELT API supports it. If results can be large, add next_cursor to response or document how to paginate via offset. If not supported, add to tool description: 'Results limited to max_records. GDELT API does not support pagination, for larger result sets, use multiple queries with different timespans.'