Discord-mcp has 6 tools with complete input schemas and descriptions, but quality is inconsistent. Tool names follow verb_noun convention (send-message, read-messages, list-channels, get-user-info, react-to-message), which is good. However, descriptions are generic and lack actionable context for LLM selection. Parameter descriptions exist but are minimal (avg ~40 chars vs baseline 72). No output schemas documented. Error handling is present but not recovery-focused. The 'channel' parameter accepts both names and IDs but lacks clear guidance on format (e.g., '#general' vs 'general' vs channel ID). No pagination support on read-messages despite accepting a limit parameter. Missing idempotency guidance for write operations.

• Document output schemas for all tools. E.g., send-message should return {message_id, channel_id, timestamp, author_id} so agents can chain to react-to-message or read-messages.
• Split 'channel' parameter into explicit channel_name (with # prefix) and channel_id variants, or add enum/pattern constraint to description: 'Channel name (prefix with #, e.g. #general) or numeric ID (e.g. 123456789)'.
• Add pagination to read-messages: accept cursor/offset and limit (1-100, default 20), return {messages: [...], next_cursor, total_count} to prevent context window overflow.
• Expand tool descriptions to 100-200 chars with WHEN/WHY context. E.g., send-message: 'Send a message to a Discord channel. Use this to notify users, post updates, or respond to events. Returns message ID for reactions or edits. Fails if bot lacks channel permissions.'
• Add error recovery guidance in tool descriptions and error responses. E.g., 'If channel not found, call list-channels to discover available channels. If permission denied, verify bot has Send Messages permission in the target channel.'