discord-mcp has 4 tools with clear verb-noun naming (get_servers, get_channels, read_messages, send_message). All tools have descriptions (10-80 chars) and input schemas with type definitions. However, descriptions are minimal (under 100 chars) and lack context about when to use each tool or what they return. Parameter descriptions are present but terse. Output schemas are not documented, callers must infer structure from code. Error handling is basic (ValueError for constraints) but lacks recovery guidance. No tool annotations (readOnlyHint/destructiveHint). Missing pagination for list results and no guidance on rate limits or Discord API constraints.

• Add output schema documentation to each tool. E.g., 'Returns: list of {id: string, name: string, type: string}' for get_channels. This enables LLMs to plan chained calls.
• Expand tool descriptions to 50-150 chars with WHAT/WHEN/RETURN structure. E.g., 'List all Discord servers you have access to. Call this first to discover available servers before reading channels or messages. Returns: [{id, name}]'
• Add tool annotations: readOnlyHint=true for get_* and read_messages; destructiveHint=true for send_message. This signals safety to agents without requiring description parsing.
• Implement pagination for get_servers and get_channels. Add optional limit (default 20, max 100) and cursor parameters. Return total_count and next_cursor in response.
• Enhance error messages with recovery guidance. E.g., 'Invalid hours_back: 10000. Must be 1-8760 (1 year). Try hours_back=24 for recent messages or hours_back=168 for the past week.'

Inferred effective spec: <=2025-11-25.

• Relies on Logging (deprecated) - log to stderr or use OpenTelemetry