n2-ark provides 5 tools with clear safety-focused naming (ark_check, ark_status, ark_load_rules, ark_stats, ark_reset). All tools have descriptions (avg 120 chars, within 10-1024 range). Input schemas are present and typed. However, parameter descriptions are sparse or missing entirely, most params lack guidance on format, constraints, or usage context. Output schemas are not documented. Error handling is minimal; responses lack recovery guidance. The tool set is well-composed (single responsibility per tool) but lacks the LLM-optimization depth expected of A-grade tools.

• Document output schemas for all 5 tools. For ark_check, specify: { allowed: boolean, reason?: string, layer?: string, action_id?: string }. For ark_status, list fields like { rules_loaded: number, state_machines: [...], layers: {...} }. This enables LLMs to extract and chain data.
• Add parameter descriptions with format and constraint details. E.g., 'rules (string): Rules in .n2 format (see https://nton2.com/syntax). Max 10KB. Invalid syntax returns a parse error with line number.' For ark_stats, add 'days (number): Lookback window in days. Range: 1 - 365. Default: 7.'
• Add error handling guidance to tool descriptions. E.g., 'If rules fail to parse, the response includes error_line and error_message. Retry after fixing syntax, or ask the user to review the rule format.'
• Clarify ark_check usage context. Expand description: 'Call before executing any tool or API call to verify it complies with loaded safety rules. Returns allowed=true if safe, false if blocked. Use the reason field to inform the user why an action was denied.'
• Add per-parameter validation examples. E.g., 'ark_load_rules: rules param must be valid .n2 syntax. Example: "rule: block_delete_all if action == delete_database"'. This helps LLMs construct valid inputs on first try.