Three well-defined tools with complete input schemas and descriptions in Portuguese. All tools have clear names (cortex_briefing, cortex_remember, cortex_recall) that follow verb_noun convention. Descriptions are detailed (150-250 chars) and explain WHEN to use each tool. Input schemas are properly typed with enums and constraints. However, output schemas are not documented, LLMs cannot predict response structure. Parameter descriptions are good but could be more concise. No error handling guidance visible in tool definitions.

• Document output schemas for all three tools. Example: cortex_briefing returns {entries: [{id, type, text, tags, timestamp, superseded_by?}], total_chars_used, budget_remaining}. This lets LLMs plan downstream calls.
• Add error handling guidance to cortex_remember description: 'Returns error if text exceeds 2000 chars or type is invalid. Retryable if storage temporarily unavailable.'
• Revise cortex_briefing description to state: 'budget_chars: integer, 500 - 20000 (default 6000). Larger budgets return more entries but consume more tokens.'
• Revise cortex_recall description to state: 'limit: integer, 1 - 50 (default 10). Results are sorted by recency; use offset for pagination if needed.'
• Add composition guidance: 'Typical flow: (1) cortex_briefing to load context, (2) cortex_remember to log decisions, (3) cortex_recall to search before re-deriving. Call cortex_briefing first in each session.'