Hive MCP server has 5 tools with basic schemas and descriptions, but significant gaps limit production readiness. All tools have descriptions (10-60 chars, mostly short), and input schemas are present with type definitions. However, descriptions are sparse and lack LLM-optimization guidance (pattern:tool-description baseline 194 chars avg). No output schemas documented. Parameter descriptions are minimal. Error handling and recovery guidance absent. No tool annotations (readOnlyHint/destructiveHint). hive_remember is marked WRITE but lacks confirmation/dry-run pattern. Tools lack composition guidance (e.g., hive_route output should feed into downstream decisions). Overall structure is present but shallow.

• Expand tool descriptions to 100-200 chars. Include WHAT the tool does, WHEN to use it (vs similar tools), and WHAT it returns. E.g., 'hive_route: Route a mechanical decision locally without LLM. Use when the decision is deterministic (e.g., pick best tool from a list). Returns the selected tool name or escalation flag.'
• Document output schemas for all tools. Specify return type (object/array/string), required fields, and field types. E.g., hive_route returns {tool_name: string, confidence: number, escalate: boolean}.
• Add parameter descriptions with format/constraint hints. E.g., 'goal: The decision to route (string, 1-500 chars). Examples: "pick best tool", "classify intent". Do not include sensitive data.'
• Add tool annotations: mark hive_route, hive_compress, hive_recall, hive_search with readOnlyHint=true; mark hive_remember with destructiveHint=true (or idempotentHint if keyed updates are safe).
• Implement confirmation pattern for hive_remember: add optional 'dry_run' parameter (default false) or return a confirmation request (MRTR pattern) before persisting writes.