Four tools with complete input schemas and descriptions, but significant gaps in output documentation, error handling, and parameter clarity. All tools accept free-form strings where enums or constraints would prevent hallucination. No output schemas documented, LLMs cannot plan downstream calls or validate responses. Descriptions are adequate (100-200 chars) but lack actionable context on when to use each tool vs. alternatives. No error recovery guidance. Tools are read-only (low risk) but lack idempotency hints and confirmation patterns for a marketing context where mistakes are costly.

• Document output schemas for all tools. E.g., find_subreddits should return {subreddits: [{name, subscribers, fit_score, rules_summary, recommended_post_type}], total_count}. This enables LLMs to extract subreddit names for draft_reddit_post.
• Add length constraints and format hints to string parameters. E.g., product_description: 'string, 50-500 chars, describe what the product does and target audience' and your_story: 'string, optional, 0-300 chars, personal founder context'.
• Wrap Groq API calls in try-catch with actionable error messages. E.g., 'Groq API timeout. Retry with a shorter product_description (max 300 chars) or try again in 30 seconds.'
• Add readOnlyHint: true to all tool definitions to signal idempotency and prevent LLM hesitation on repeated calls.
• Clarify tool boundaries in descriptions. E.g., find_subreddits: 'Discover subreddits and their rules. Use this first to explore options.' plan_reddit_campaign: 'Build a multi-phase campaign timeline across multiple subreddits. Use after you have a product and target audience.'