Apollo.io MCP server has clear tool naming (verb_noun pattern) and complete input schemas with type definitions and descriptions. However, output schemas are entirely undocumented, the code shows API calls but no structured response documentation. Parameter descriptions are present but generic (72 chars avg, within baseline). Error handling is minimal: no recovery guidance, no categorization of retryable vs fatal errors, no actionable error messages. Tool descriptions are adequate (100-150 chars) but lack WHEN to use guidance and dependency hints. No tool annotations (readOnlyHint, destructiveHint, idempotentHint) despite all tools being read-only. Missing pagination documentation for list tools despite per_page parameters.

• Document output schemas for all 7 tools. For each, specify the response object structure, field types, and which fields are required for downstream tool calls. Example: apollo_search_people returns {people: [{id, name, email, title, company, ...}], total_count, page, per_page}.
• Add try-catch error handling with actionable messages. Catch axios errors and return: {error: 'API key invalid, check APOLLO_API_KEY environment variable'} or {error: 'Rate limited, retry in 60 seconds'} instead of raw HTTP status codes.
• Add readOnlyHint: true to all 7 tool definitions in the inputSchema. This signals to agents that these tools are safe to retry and have no side effects.
• Enhance tool descriptions with pagination guidance. For search_people, search_organizations, search_sequences: 'Returns paginated results (default 10-25 per page). Use page parameter to iterate. Total count included in response.'
• Improve apollo_get_email_accounts description from 'Get list of email accounts' to 'Retrieve all email accounts connected to your Apollo account. Returns account IDs, email addresses, and connection status. Call this first to identify which email to use for sending sequences.'