Scout MCP provides 21 search tools with consistent naming (verb_noun pattern: *_search) and reasonable parameter schemas. However, descriptions are generic and lack LLM-optimization guidance. Most tools accept query + per_page + optional sort/site/category, but descriptions do not explain WHEN to use each tool vs. alternatives, what makes results actionable, or how to chain tools. Output schemas are not documented, LLMs cannot predict what fields to extract. Error handling is absent from visible code. Tool descriptions average ~120 chars, below the 194-char baseline for A+ tools. No parameter descriptions explain format constraints, valid ranges, or dependencies. Security: x_search and other tools call external APIs but no visible rate-limiting, timeout handling, or secret injection patterns in tool definitions.

• Document output schemas for all 21 tools. For each, specify: field names, types (string|number|array|object), and whether fields are always present or optional. Example: hackernews_search returns [{id: number, title: string, score: number, url: string, comments: number}].
• Expand tool descriptions to 150-250 chars. Add: (1) What makes this source unique vs. similar tools? (2) When should the agent prefer this over alternatives? (3) What downstream actions are typical? Example: 'Search HackerNews for tech news and startup discussions. Use when seeking community sentiment on emerging technologies. Returns discussion links and comment counts, good for gauging interest before deep research.'
• Add parameter descriptions explaining format, constraints, and dependencies. Example: per_page: 'Number of results per page (1-50). Higher values return more results but increase latency and token cost. Default 10 balances coverage and speed.'
• Implement error handling with recovery guidance. Return structured errors: {error: string, code: string, retryable: boolean, suggestion: string}. Example: {error: 'Rate limit exceeded', code: 'RATE_LIMIT', retryable: true, suggestion: 'Wait 60 seconds or try a different search source.'}
• Add tool selection guidance in scout_report description. Clarify which sources are in 'balanced' (14 free) vs 'comprehensive' (18 with X), estimated cost per call, and when to use scout_report vs. individual tools for cost/latency tradeoffs.