Context

Agents have a PersonalityConfig model (traits, communication_style, risk_tolerance, creativity, description) but the framework lacks depth in how personality impacts agent behavior. This issue researches best practices for LLM persona engineering and implements a comprehensive personality system.

Acceptance Criteria

• Research: survey existing multi-agent persona systems (CrewAI, AutoGen, MetaGPT, etc.)
• Research: evaluate impact of personality traits on LLM output quality
• Extended personality model (interaction styles, decision-making patterns, collaboration preferences)
• System prompt construction that incorporates personality effectively
• Personality compatibility scoring (for team composition)
• Unit tests

Dependencies

• Issue Define default configurations and company template schema #60 (company template schema — minimal personality presets)

Design Spec Reference

Sections 3.1 (Agent Identity Card), 3.4 (Dynamic Roles)