Strong tool naming (verb_noun pattern, clear action semantics) and comprehensive descriptions (avg 180 chars, well above 34-char baseline). All 20 tools have descriptions and most parameters are typed. However, input schemas lack formal JSON Schema constraints (no enums, min/max bounds, or pattern validation). Output schemas are undocumented, LLMs cannot predict response structure. Error handling is generic (returns {error, message} dict) without recovery guidance or categorization. Security is well-designed (confirm/dry_run guards, control-char stripping, secret redaction), but no per-tool permission declarations or audit logging visible.

• Document output schemas for all tools. Use JSON Schema in tool descriptions or a separate schema registry. Example: 'Returns {devices: [{name, mac, model, status, radios: [{band, channel, width, tx_power}]}]}'.
• Add formal enum constraints to input schemas. For band, use enum: ['2.4GHz', '5GHz', '6GHz']. For tx_power_mode, use enum: ['auto', 'low', 'medium', 'high', 'custom']. For channel, use pattern or range validation.
• Implement error categorization. Return {error, message, category: 'retryable'|'user_fixable'|'fatal', recovery_hint: '...'}. Example: 'AP not found. Try unifi_list_devices() to see available APs.'
• Add per-tool scope declarations in descriptions. Example: 'Requires scopes: read:network, write:radio. Requires local-admin API.' This enables least-privilege agent configs and audit trails.
• Add pagination to list tools. unifi_list_devices(limit=50, offset=0) and return {devices: [...], total: N, has_more: bool}. Cap default limit at 50 to prevent context explosion.

Inferred effective spec: 2026-07-28+.

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