ZK Agent provides 7 tools with basic descriptions and input schemas, but falls short of production quality. All tools have descriptions (10-100 chars, mostly 50-80), and input parameters are typed with descriptions. However, output schemas are completely undocumented, LLMs cannot predict what fields to expect from responses. Parameter descriptions lack actionable constraints (no enums, ranges, or format specs). Error handling is absent, no guidance on retryability, user-fixable vs fatal errors, or recovery paths. Tool naming is clear (verb_noun pattern), but composition could be tighter: semantic_search_objects and search_whiteboards overlap in intent without clear disambiguation. No tool annotations (readOnlyHint/destructiveHint) despite clear risk levels. Average tool score: 58/100.

• Document output schemas for all tools. E.g., semantic_search_objects should specify: returns {objects: [{id, name, type, preview}], total_count}. get_whiteboard_with_objects should list fields in the returned dict.
• Add error handling guidance to descriptions. E.g., 'Returns error if whiteboard_id not found, try search_whiteboards() first. Retryable on network timeout.'
• Constrain parameters with enums, ranges, or format specs. E.g., resultObjectTypes should be enum ['card', 'journal', 'whiteboard']. Date parameters should specify ISO 8601 format explicitly.
• Clarify tool disambiguation. Rename search_whiteboards to search_whiteboard_by_name and semantic_search_objects to search_objects_by_content to make intent explicit.
• Add tool annotations in MCP registration: mark save_to_note_card and append_to_journal with destructiveHint=true; mark all read tools with readOnlyHint=true.