Hercules MCP has 8 tools with basic schemas and descriptions, but significant gaps prevent higher scoring. All tools have descriptions (10-150 chars, mostly adequate), and input schemas are present with type definitions. However, parameter descriptions are sparse or missing entirely (e.g., retrieve_logs has empty input schema with no description). Output schemas are not documented anywhere. Error handling is absent, no guidance on recovery, retryable vs. fatal errors, or actionable error messages. Tool naming is verb-first and clear (list_, create_, run_, execute_, get_, interrupt_, delete_, retrieve_), but descriptions lack context on when to use each tool vs. alternatives. No tool annotations (readOnlyHint, destructiveHint, idempotentHint) despite clear risk levels (READ_ONLY, WRITE, DESTRUCTIVE). Parameter constraints are minimal, no enums, ranges, or format specs. The execute_bash tool is particularly risky: accepts arbitrary commands with no validation, sanitization, or injection guards documented. Overall, definitions are functional but lack the depth and rigor expected of production tools.

• Add output schemas to all tools. Document the structure of returned objects, field types, and which fields are required for chaining (e.g., conversation_id from create_conversation must be usable in run_task).
• Add tool annotations: mark delete_conversation and interrupt_task as destructiveHint=true; mark list_conversations and get_task_status as readOnlyHint=true; mark create_conversation as idempotentHint=true if it deduplicates by workspace_path.
• Document retrieve_logs input schema fully. Specify what parameters filter logs (e.g., conversation_id, log_level, start_time, end_time) and their types/constraints.
• Add validation and sanitization to execute_bash. Document which commands are blocked, what injection attacks are prevented, and return clear errors for invalid commands (e.g., 'Command contains forbidden pattern: rm -rf').
• Add error handling guidance to all tools. Document common failure modes (e.g., 'conversation_id not found', 'workspace_path does not exist') and recovery steps (e.g., 'Call list_conversations to find valid IDs').

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

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